Microfluidic droplet cultivation preserves microalgae diversity in screening systems Authors: Satoshi Yamamoto, Kazuma Yoshimura, Kayo Konishi, Reiko Nozaki, Ikuo Hirono, Keiichiro Koiwai Published: 2025 ABSTRACT AbstractEfficient screening of microalgae is critical for their research and industrial applications. However, conventional methods of microalgae screening are often considered inefficient and are applicable to a narrow range of species. We conducted a proof-of-concept study on the applicability of a water-in-oil droplet (WODL)-based microfluidic system to high-throughput screening of microalgae and effect of microfluidic droplet cultivation on the species diversity within microalgal samples. First, we confirmed that microalgae with diverse morphologies could be successfully cultured within droplets. Second, we conducted culture tests using a mock community derived from monocultured microalgal strains and a field community derived from the natural environment. Subsequent long-read metabarcoding targeting the 16 S rRNA gene, followed by diversity analysis, revealed that droplet culture is more effective than bulk mixed-species culture in maintaining species diversity. Our results lay the groundwork for the future development of WODL-based high-throughput screening systems, allowing access to a richer microalgal biodiversity.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-31621-x.Keywords: Microalgae, Water-in-oil droplet, High-throughput screening, Microfluidics, Species diversity, Diversity analysisSubject terms: Biological techniques, Biotechnology, Microbiology, Plant sciences Introduction Microalgae have gained increasing attention in recent years as a promising biological resource across various industrial sectors1,2. There are two major advantages of utilizing microalgal resources. One is their versatility, attributed to the vast species diversity of microalgae. They live not only in aquatic environments, including marine and freshwater ecosystems, but also in terrestrial habitats. This ecological diversity contributes to their ability to produce a wide range of valuable compounds beneficial to human society, such as lipids, carotenoids, polysaccharides, and bioactive metabolites. Consequently, microalgae have highly diverse applications, including the extraction of high-value biochemical compounds, as mentioned above; the use of whole biomass as feed in aquaculture or as dietary supplements for humans and pet animals; and bioremediation, where living cells are employed for the purification of soil and wastewater1,3. Another key advantage of microalgae is their ability to combine high productivity with environmental sustainability. Marine microalgae can be cultivated using seawater, the most abundant natural resource on the planet, without competing for arable land or freshwater essential for food production. In terms of biomass production, microalgae can yield 5–10 times more biomass per unit area than terrestrial crop plants4,5. Moreover, some species have been reported to exhibit CO₂ fixation rates that are 10–50 times higher than those of land plants, highlighting their potential as a valuable biological resource from the perspective of carbon neutrality, which has become increasingly important to industry in recent years6,7. Industrial utilization of microalgae, along with the related fundamental and applied research, begins with the screening of strains possessing desirable traits from the diverse microalgal communities found in natural environments8. The first step in this process is the isolation of the target or candidate strains from environmental samples, which are then established as stable unialgal cultures and serve as the basis for all downstream studies8,9. Microalgal isolation techniques originated with the pioneering research by Beijerinck8,10, and have since evolved into a range of methods, including gravity-based separation using centrifugation or sedimentation and phototaxis-based isolation. Currently, the most commonly used approaches are single-cell isolation using a micropipette, serial dilution, and agar plate culture8. Given that each method has its own advantages and limitations, the choice of method depends on both the morphological and ecological characteristics of the target species, such as cell size, shape, robustness, motility, and growth behavior, and properties of the environmental samples, including their origin (e.g., marine, freshwater, soil, or extreme environments such as acidic, hypersaline, or cryospheric habitats), microalgal abundance, and degree of contamination. When working with samples from poorly characterized environments, it is often necessary to combine multiple isolation methods and apply preprocessing steps, such as preculture, concentration by filtration or sedimentation to increase cell density, and chemical treatments to exclude non-target organisms. Rapid and efficient screening of the target or of diverse candidate strains from natural environments is important for research aimed at industrial applications. However, conventional microalgal isolation methods have limited throughput, efficiency, and adaptability to the morphological and ecological diversity of microalgae. For instance, agar plate isolation is time-consuming, often requiring one to two months for the microalgae to form colonies large enough for picking, and is restricted to colony-forming species. Serial dilution is ineffective in isolating rare species owing to their low abundance in samples. Micropipette-based methods, although precise, are technically demanding and lack scalability. Flow cytometry–based methods such as fluorescence-activated cell sorting (FACS) address some of these limitations, but the sorting process exposes cells to various physical stresses—including high fluidic pressure, laser irradiation, electrostatic charging, high-voltage deflection, and collisions with container surfaces—which can damage cells. Consequently, the probability of successfully establishing clonal cultures from sorted microalgal cells is generally low11. In recent years, to overcome these drawbacks, droplet microfluidics-based technology is emerging as a powerful tool for high-throughput screening of microorganisms12–15. By encapsulating single cells into individually separated water-in-oil droplets (WODLs), this technology enables “million screening” and has the capacity to generate hundreds of thousands of droplets per minute. Originally developed for studies involving bacteria14,15, fungi16, yeast17, bacteriophages18, cultured cells19, and enzymes20, it has recently been applied to microalgae. For example, Kim et al.21–23 used WODLs to screen high-growth, high-lipid-producing strains from mutant libraries of Chlamydomonas reinhardtii and Botryococcus braunii. Similarly, Yu et al.24 developed a WODL-based method to rapidly identify genetically modified Phaeodactylum tricornutum and Nannochloropsis gaditana cells expressing green fluorescent protein, thereby streamlining algal transformation workflows. However, there have been no reports of applying this technology to the screening of natural environmental samples containing diverse microalgal species, and information on such applications remains limited. In this study, we present a proof-of-concept for constructing a high-throughput screening system for microalgae using WODL. We demonstrated that single cell isolation and cultivation using WODL can serve as a first step in such screening while preserving species diversity in complex microalgal samples. We encapsulated mixed-species microalgal communities into droplets at the single-cell level and examined the effect of the approach on the species diversity within the community. This study consisted of two parts: one using a mock community containing several monocultured microalgal strains, and the other using a field community derived from environmental samples containing wild microalgae. In both cases, the effects of droplet encapsulation and internal cultivation on the community diversity was evaluated through long-read metabarcoding targeting the 16 S rRNA region, followed by diversity analysis. The findings of this study lay the groundwork for applying WODL-based microfluidic systems to high-throughput screening of microalgae from natural environments. Results and discussion Encapsulated microalgae grow successfully in droplets To evaluate the compatibility of WODL with microalgal cultivation, we first examined the growth dynamics of seven microalgal strains belonging to different taxonomic groups and exhibiting diverse morphologies, each individually encapsulated in droplets (Fig. 1). Fluorescence microscopy targeting chlorophyll a autofluorescence confirmed that single cells encapsulated in the droplets successfully increased in number (Fig. 2a, b). After 144 h of cultivation, the average number of cells per droplet was 61.7 ± 1.4 for N. oculata, 42.0 ± 2.1 for P. tricornutum, 29.1 ± 0.44 for Pavlova lutheri, 10.8 ± 0.71 for Chaetoceros neogracile, 10.2 ± 0.42 for Tetrasekmis tetrathele, and 7.6 ± 0.31 for Rhodomonas sp. Because each droplet had an internal volume of 1 nL, these values correspond to the cell densities per nanoliter. These cell densities were comparable to or even higher than those achieved in standard bulk cultures, indicating that the tested microalgae could grow efficiently even within the confined space of the droplets. In contrast, Isochrysis galbana showed no detectable growth in the droplets, despite growing normally during the preconditioning. Comparison among the species revealed a trend in which smaller-celled species exhibited greater growth. This suggests that the general hypothesis that smaller cells with a higher surface area-to-volume ratio can absorb nutrients more efficiently and exhibit a growth advantage under nutrient-limited conditions25,26 also applies to culture within droplets. An exception to this explanation was C. neogracile, which, despite its relatively small size (5–8 μm), reached a similar cell number per droplet as that of T. tetrathele (12–16 μm), the largest-celled species in this study. This result is likely because of the existence of long setae extending from the four corners of C. neogracile cells (Fig. 2c). As proposed by Liu et al.27, nutrient availability and physical space are key resources determining microbial growth, and the competition for these resources strongly affects growth and proliferation. In this context, the setae likely reduced the usable physical space within the droplets, limiting cell division. In addition, species such as T. tetrathele and Rhodomonas sp., which possess multiple long flagella and exhibit active motility, were observed to form cell aggregates within the droplets because of entanglement of their flagella (Fig. 2c). Given the physical space constraints described above, this cell aggregation may also act as a limiting factor for cell growth. These two phenomena seem specific to WODL-based cultivation and are rarely observed in conventional bulk cultures using flasks. Although this study demonstrated the feasibility of microalgal cultivation within WODL, future research must consider issues unique to microscale environments caused by species-specific morphology. Growth dynamics of seven microalgae encapsulated in droplets. Values represent mean ± standard error (n = 3). Microscopic images of microalgae. Overlay fluorescence images of Tetraselmis tetrathele as a representative among the seven microalgae, encapsulated in droplets at (a) 0 h and (b) 96 h. Scale bar: 200 μm. Bright-field images of the mock community at 96 h, taken from the (c,d) Droplet sample and (e) Bulk sample. Scale bar: 100 μm. Mock community test confirms that cultivation in droplets maintains microalgal diversity To verify the utility of the WODL in microalgal screening, we conducted a preliminary investigation of the effect of droplet encapsulation and internal cultivation on the species diversity in microalgal communities by using a mock community containing seven monocultured microalgal strains. Figure 2c-e show the Droplet and Bulk samples after 96 h of cultivation in the mock community. As expected, the Bulk sample showed a bias in the species present after 96 h of cultivation, whereas in the Droplet samples, no substantial bias was observed among the species, all of which remained independently distributed within the droplets. In this study, to validate these observations, we performed a metabarcoding analysis targeting the 16 S rRNA gene. A total of 21 samples were sequenced to analyze the mock community cultured in the bulk and droplet conditions. Each sample yielded between 4,489 and 11,130 effective reads, which were used to generate the operational taxonomic unit (OTU) abundance tables (see Supplementary Table S1). Figure 3a shows the changes in the species composition under bulk and droplet conditions in the mock community. Among the seven species, Chaetoceros became dominant in the Bulk samples throughout the experiment. Its abundance increased from 42.2 to 43.4% at 0 h to 79.4–86.7% at 48 h, 90.7–93.4% at 96 h, and 93.8–96.1% at 144 h, accounting for the majority of the total reads. Interestingly, the general trend described above that smaller-celled species grow better did not hold under the mixed-species conditions of the Bulk sample. For example, at 144 h, the abundances of Nannochloropsis and Phaeodactylum were below 0.1% and 0.5%, respectively, much lower than that of Chaetoceros. A possible explanation for this is the competitive growth strategy of diatoms against the growth of other species. Previous studies have shown that diatoms have competitive advantages, such as lower carbon and energy requirements for cell division, because they use silica-based cell walls, which are more energy-efficient than organic matter28,29. In our experiment, the culture medium contained silica, which could be exclusively utilized by diatoms but not by other microalgae, likely giving them an additional advantage. In addition, some diatom species are known to exert allelopathic effects that inhibit the growth of other microalgae30–32. Although it remains unclear whether such effects influenced the species in this study, it is likely that these multiple factors contributed to the dominance of a single fast-growing species in the Bulk samples. By contrast, the Droplet samples tended to maintain their initial species diversity throughout the experimental period. At 144 h, the abundance of Chaetoceros remained at a lower level of 42.2–45.5%, whereas the other species showed higher values than those in the Bulk samples: Tetraselmis accounted for 14.4–20.2%, Pavlova for 15.2–17.2%, Rhodomonas for 12.6–15.3%, Phaeodactylum for 5.4–6.2%, and Nannochloropsis for 2.1–2.6%. The preserved species diversity in the Droplet samples is because of the physical separation of cells by the droplets (Fig. 2c, d), which minimizes interspecific competition and allows slower-growing species to persist. This observation aligns with the findings of Geersens et al.33, who demonstrated that microfluidic droplet cultivation can reduce representation bias caused by fast-growing species, thereby facilitating the enrichment of slower-growing or rare taxa. In addition, they also reported that cell compartmentalization in the droplets helps prevent the diffusion of allelochemicals released by competing species, which can inhibit growth. These results highlight the capacity of droplet microfluidics to mitigate the advantages provided by the culture conditions, such as silica availability for diatoms, and interspecific interactions, thereby helping to preserve the community composition during cultivation. As an exception, Isochrysis, which did not show growth in the single-species WODL cultivation test conducted in this study, also failed to grow here, with its abundance remaining extremely low at below 0.5%. This result suggests that cultivation using the WODL may not fully reflect the species compositions of the original community. Genus-level taxonomic compositions in Initial, Bulk, and Droplet samples based on 16 S rRNA gene sequencing. (a) Mock community at 0, 48, 96, and 144 h. (b) Field community at 0 and 96 h. Abundances were calculated based on OTU assignments at the genus level. Numbers shown above each bar in panel (b) indicate the number of detected genera. Only the genera with relative abundances ≥ 1% are shown individually; the remaining taxa are grouped as “Others.” Detailed taxonomic data are provided in the Supplementary Table S1 and S2. Using WODL supports the preservation of species diversity in the field community The findings from the mock community experiment provided preliminary validation of the approach of applying WODL to microalgal screening. Based on the results, we validated the approach on a field community derived from natural surface seawater for further practical verification. A total of 9 samples were sequenced to analyze the field community cultured under bulk and droplet conditions. Each sample yielded between 37,510 and 54,970 effective reads, which were used to generate the OTU abundance tables (see Supplementary Table S2). Rarefaction curves based on the Chao1 index confirmed that the sequencing depth was sufficient to capture the major species components in each sample (Fig. 4a). As shown in Fig. 3b, the dominant genera in the Initial samples were Pyrenomonadales_XXX, accounting for 16.6–39.9%, followed by Chaetoceros with 6.2–12.8%, Arcocellulus with 5.8–12.9%, and Prymnesiophycidae_XXXXX with 5.8–8.4%. After 96 h of cultivation in the Bulk sample, the diatom Arcocellulus became the dominant taxon, accounting for 80.1–91.9% of the total community. This strong dominance was accompanied by a substantial decline in the species diversity, mirroring the pattern seen in the mock community. In contrast, the Droplet samples exhibited a more balanced composition. Although Arcocellulus remained the most abundant genus, its abundance was relatively lower than that in the Bulk sample, ranging from 56.6 to 61.1%. Furthermore, other genera detected in the Initial samples were also retained to a greater extent in the Droplet samples when compared with the Bulk samples, reflecting a similar trend of diversity preservation as observed in the mock community. Quantitatively, the number of detected genera ranged from 36 to 42 in the Initial samples, 11 to 13 in the Bulk samples, and 22 to 26 in the Droplet samples, indicating that approximately 1.9-fold additional genera were recovered under droplet conditions than in bulk culture. This pattern was consistent with the alpha diversity metrics, which showed significantly higher values in the Droplet samples across all three indices (Shannon, Simpson, and Chao1), suggesting that taxonomic richness and evenness were better preserved in droplets (Fig. 4c). Beta diversity analysis further supported these observations. The principal coordinates analysis (PCoA) revealed clear clustering of the three sample types, with PERMANOVA confirming statistically significant differences among them (p = 0.006) (Fig. 4b). The first principal coordinate (PC1), which explained a remarkably high proportion of the total variation (84.32%), separated the Initial samples from both cultivated groups. However, the Droplet and Bulk samples were located close to each other along this axis. This result indicates that the cultivation process itself, whether in bulk or in droplets, was the primary factor driving the overall community shift in this study. One likely reason for this strong effect is that natural microalgal communities often contain many species that cannot grow under laboratory conditions2,34–36. Some species may require specific factors, such as temperature, light, or nutrients, which were not fully reproduced in our experimental setup. As a result, cultivation may have filtered out many sensitive or slow-growing species early in the process, leading to similar changes in both the droplet and bulk cultures. The second coordinate (PC2), although explaining only 11.4% of the variation, contributed to the separation between the Droplet and Bulk samples. This indicates that, beyond the initial filtering of sensitive or non-culturable species during cultivation, the use of WODL influenced the resulting community structure, likely by reducing interspecific competition and allowing a greater number of taxa to persist. Diversity analysis of the field community based on 16 S rRNA gene sequencing. (a) Rarefaction curves based on the Chao1 index for Initial, Bulk, and Droplet samples. (b) Principal coordinates analysis (PCoA) based on OTU composition. (c) Alpha diversity indices (Shannon, Simpson, and Chao1) for each sample type. Different letters indicate significant differences (p < 0.05). Limitations of the study Although cultivation conditions were designed to reflect both the storage environment of culture stocks and the conditions at the time of field sampling, they may not have fully replicated the natural habitat of the field community. In particular, laboratory light regimes and nutrient concentrations likely differed from in situ conditions. These discrepancies may have influenced which taxa persisted during cultivation, potentially altering the proportion of species retained between bulk and droplet cultures. A more detailed evaluation of environmental parameters will be necessary in future work to refine experimental designs and enhance the ecological relevance of cultivation-based diversity assessments. Although cultivation conditions were designed to reflect both the storage environment of culture stocks and the conditions at the time of field sampling, they may not have fully replicated the natural habitat of the field community. In particular, laboratory light regimes and nutrient concentrations likely differed from in situ conditions. These discrepancies may have influenced which taxa persisted during cultivation, potentially altering the proportion of species retained between bulk and droplet cultures. A more detailed evaluation of environmental parameters will be necessary in future work to refine experimental designs and enhance the ecological relevance of cultivation-based diversity assessments. In this study, the chloroplast 16 S rRNA gene was selected as the target for metabarcoding analysis for two principal reasons. First, Alonso et al.37 demonstrated that both chloroplast 16 S and nuclear 18 S rRNA markers can reliably discriminate microalgal taxa at the genus level, indicating that chloroplast 16 S can provide taxonomic resolution comparable to 18 S in certain contexts. Second, chloroplast 16 S rRNA benefits from curated databases such as PhytoREF38, which are tailored to photosynthetic eukaryotes. Nonetheless, plastid 16 S coverage remains incomplete, and some groups may be underrepresented. Incorporating complementary markers, such as nuclear 18 S rRNA or the chloroplast-encoded rbcL gene, is recommended for future studies to achieve a more comprehensive assessment of natural microalgal communities. In this study, the chloroplast 16 S rRNA gene was selected as the target for metabarcoding analysis for two principal reasons. First, Alonso et al.37 demonstrated that both chloroplast 16 S and nuclear 18 S rRNA markers can reliably discriminate microalgal taxa at the genus level, indicating that chloroplast 16 S can provide taxonomic resolution comparable to 18 S in certain contexts. Second, chloroplast 16 S rRNA benefits from curated databases such as PhytoREF38, which are tailored to photosynthetic eukaryotes. Nonetheless, plastid 16 S coverage remains incomplete, and some groups may be underrepresented. Incorporating complementary markers, such as nuclear 18 S rRNA or the chloroplast-encoded rbcL gene, is recommended for future studies to achieve a more comprehensive assessment of natural microalgal communities. The mechanism by which Isochrysis inhibited growth within the droplets remains unclear. Several factors may impair microbial growth in droplet microfluidic systems, including shear stress generated during microfluidic flow, the effects of fluorinated surfactants on cell membranes, altered gas permeability of fluorinated oils, nutrient limitations in confined volumes, and pH shifts caused by accumulated metabolic byproducts39. Further investigations are needed to determine which of these factors may have influenced microalgal species in this technology. The mechanism by which Isochrysis inhibited growth within the droplets remains unclear. Several factors may impair microbial growth in droplet microfluidic systems, including shear stress generated during microfluidic flow, the effects of fluorinated surfactants on cell membranes, altered gas permeability of fluorinated oils, nutrient limitations in confined volumes, and pH shifts caused by accumulated metabolic byproducts39. Further investigations are needed to determine which of these factors may have influenced microalgal species in this technology. Finally, this WODL-based screening system enables only primary screening. Conditions inside droplets differ fundamentally from those in large-scale outdoor cultivation, which is the primary context for algal industries. Therefore, once candidate strains are identified using this approach, their growth must be evaluated under different cultivation scales and environmental settings using conventional methods. Finally, this WODL-based screening system enables only primary screening. Conditions inside droplets differ fundamentally from those in large-scale outdoor cultivation, which is the primary context for algal industries. Therefore, once candidate strains are identified using this approach, their growth must be evaluated under different cultivation scales and environmental settings using conventional methods. Conclusion This study presented a proof-of-concept study for the application of the WODL to the high-throughput screening of natural microalgae. We investigated the growth dynamics of various microalgal species in droplets and examined how single-cell encapsulation and cultivation in droplets affect the species diversity in the samples by using both a mock community and natural field community. Our results demonstrated that a variety of microalgal species can be successfully cultivated in droplets and that encapsulating the microalgal communities at the single-cell level helps maintain species diversity in samples when compared with conventional bulk cultivation. These findings highlight two key advantages of using WODL over traditional methods such as agar plating, serial dilution, micropipette-based single-cell isolation, and flow cytometry-based methods such as FACS and FlowCAM. First, the WODL-based microfluidic system enables the isolation of a considerably broader range of microalgal species, including those that do not form colonies on solid media, occur at low abundance, or grow slowly—groups often inaccessible by conventional approaches. This expands the range of species available for research, overcoming the limitations of methods like FACS, which require species-specific optimization of sorting conditions, and FlowCAM, which is not designed for isolation or cultivation. Second, the WODL-based approach achieves much higher throughput, with the capacity to encapsulate up to 10,000 individual cells per minute for direct use in primary screening. This integration of cell isolation and culture testing streamlines workflows and accelerates the identification of microalgae with desirable traits, marking a significant shift in screening strategies. Beyond improving efficiency, this approach expands access to biologically valuable but previously overlooked species, offering new microbial resources with ecological and industrial relevance. In the future, the WODL-based platforms may serve as a foundation for screening workflows in microbiology research. While this study focused on isolating and maintaining individual cells within microalgal communities, the next step will involve selective recovery of target cells. Sorting methods based on cell size or fluorescence emission within droplets have already been developed, as demonstrated by Hoshino et al.18 and Mori et al.14 By applying such approaches, it should be possible to establish large-scale screening systems for microalgae. In particular, because microalgae contain a wide variety of photosynthetic pigments, selecting specific taxonomic groups based on their characteristic fluorescence wavelengths should be relatively straightforward compared to bacteria and other microorganisms. We are currently pursuing this line of research. As industrial interest in microalgae continues to grow, the screening system using WODL represents a promising tool for unlocking the full potential of microalgal biodiversity. The findings presented here provide foundational evidence of its applicability to diverse and ecologically relevant communities and lay the groundwork for future large-scale screening and exploitation of microalgal resources. Methods Preparation of microalgal communities Seven microalgal strains maintained at the Fisheries Technology Institute, Japan Fisheries Research and Education Agency, were used to prepare the mock community: C. neogracile (NRIA-0023), P. tricornutum (NRIA‐0065), I. galbana (NRIA‐0074), P. lutheri (NRIA‐0080), T. tetrathele (NRIA‐0090), N. oculata (NRIA‐0098), and Rhodomonas sp. (NRIA‐0113). Each strain was precultured at least three times after recovery from the stock culture to stabilize its physiological state before use in the experiments. The cultures were maintained in 500 mL Erlenmeyer flasks containing Guillard 2 F medium prepared with sterilized seawater at 20 °C, with filtered airflow and continuous illumination by white light emitting diodes (LEDs) at 80 µmol m⁻² s⁻¹. Log-phase cells were diluted with sterilized seawater to 1.0 × 10⁵ cells mL⁻¹ and pooled together. The resulting mixed-cell suspension was designated as the mock community. Environmental seawater samples were collected from Gokasho Bay (Minamiise-cho, Mie, Japan) in November 2024. The surface seawater was concentrated 1,000-fold using a 5 μm mesh phytoplankton net sequentially filtered through strainers (500, 200, 100, and 20 μm), and centrifuged to adjust the cell density to 1.0 × 10⁵ cells mL⁻¹. The resulting concentrate was used as the field community. The environmental conditions at the time of sampling were as follows: water temperature 20 ± 1 °C, surface salinity 33 psu, and continuous sunny weather for at least seven days including the sampling date. For both communities, nutrients were added to formulate the Guillard 2 F medium, and ampicillin was supplemented at 100 µg mL⁻¹. The resulting cell suspensions were used in the subsequent experiments. Experimental design The experimental design is briefly summarized in Fig. 5. The cell suspensions of each community prior to cultivation were defined as the “Initial samples.” These were then divided into two groups: one group, referred to as the “Bulk sample,” was batch-cultured in 100 mL test tubes, whereas the other, referred to as the “Droplet sample,” was encapsulated as single cells in droplets and cultured internally using the method described below. Both samples were maintained at 20 °C under continuous illumination by white LEDs at 80 µmol m⁻² s⁻¹, with aeration provided only to the Bulk samples. The cultivation period was set to six days for the mock study and four days for the field study. In addition to the above experiments, to assess the growth characteristics of each microalgal species within the droplets, each single-species cell suspension used to formulate the mock community was individually encapsulated and cultured under the same conditions. Summary of the experimental design. Droplet generation Droplets were generated using the 2D chip-1100DG equipped with an On-chip Droplet Generator (On-chip Biotechnologies, Tokyo, Japan). The oil phase consisted of 2% (w/w) 008-FluoroSurfactant in HFE-7500 (On-chip Biotechnologies). The sample phase comprised the Initial sample. Droplet formation was carried out under a sample pressure of 25 kPa and oil pressure of 30 kPa, yielding approximately 100,000 droplets per minute with an average diameter of 125 μm, corresponding to an internal volume of 1 nL. Because the number of microalgal cells per droplet theoretically follows a Poisson distribution40, setting the cell concentration to 1.0 × 10⁵ cells mL⁻¹ (corresponding to λ = 0.1 for a droplet volume of 1 nL) resulted in approximately 10% of the droplets containing a single cell. Under these conditions, the probability of multiple-cell encapsulation was calculated as P (≥ 2) = 1 – e^(–λ) (1 + λ) = 0.00468, corresponding to 0.47% of all droplets. In this study, approximately 700,000 droplets were generated per sample and transferred to 2 mL tubes for cultivation. Sampling and microscopic visualization Samples for DNA extraction were collected every 48 h for the mock study and at 96 h for the field study. For the Bulk sample, 20 mL of the suspension was collected by centrifugation at 8,000 × g for 10 min. The resulting pellets were washed three times with sterilized seawater and then recollected by centrifugation at 12,000 × g for 5 min. The final pellets were stored at − 80 °C until DNA extraction. For the Droplet sample, an equal volume of 10% (w/w) 1 H,1 H,2 H,2 H-perfluoro-1-octanol (FUJIFILM Wako Chemicals, Osaka, Japan) diluted with HFE-7500 was added to the droplet suspension. The mixture was gently shaken under exposure to a plasma globe—an electric discharge device that generates an electric field—to disrupt the droplets. The resulting aqueous phase containing the microalgal cells was collected using a micropipette, filtered through a GF/F glass fiber filter, and the retained material was stored at − 80 °C until DNA extraction. For microscopic visualization, the Droplet samples were imaged using a fluorescence microscope (BZ-X700, Keyence, Osaka, Japan) equipped with appropriate filter sets. In addition, cell counting was performed for single-species Droplet samples. A 10 µL aliquot of each droplet suspension was loaded onto Countess™ Cell Counting Chamber Slides (Invitrogen, California, USA), and the number of microalgal cells contained within each droplet was determined under an optical microscope (BX51, Olympus, Tokyo, Japan). Counting was conducted every 48 h after the start of cultivation. DNA extraction, library preparation, and 16 S rDNA sequencing DNA was extracted using the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany) following the instructions of the manufacturer. For microalgal community analysis, we targeted the plastid 16 S rRNA gene, which is widely used as a phylogenetic marker for photosynthetic eukaryotes. Specifically, a region of approximately 750 bp was amplified using the MALGAS_F (5’-CGA AAG CCT GAC RGA GCA ATA-3’) and MALGAS_R (5’-TAA GGG GCA TGM TGA CTT GAC GT-3’) primers, originally designed to cover diverse plastid 16 S sequences of microalgae37. Polymerase chain reaction (PCR) amplification was performed using TAKARA Ex taq (Takara Bio Inc., Shiga, Japan). To obtain a final reaction volume of 20 µL, a PCR reaction mixture comprising 2.0 µL of 10×Ex taq buffer, 1.6 µL of dNTPs at 2.5 mM, 0.4 µL of each primer (final concentration of 1.0 µM), 0.1 µL of Ex taq polymerase, 1.0 µL of the DNA template, and 14.5 µL of distilled water was used. For the Droplet group samples, the DNA template volume was increased to 3.0 µL and the distilled water volume was adjusted to 12.5 µL so that the final reaction volume remained 20 µL. The thermocycling conditions included initial denaturation at 95 °C for 3 min, followed by 30 cycles of denaturation at 95 °C for 30 s, an annealing step at 56 °C for 30 s, and extension step at 72 °C for 1 min, with a final extension at 72 °C for 3 min. For the field community, the number of cycles was reduced to 25 to minimize the PCR bias and facilitate the detection of a broader diversity of species. Amplicons were sequenced using Native Barcoding kit 24 V14 (Oxford Nanopore Technologies, Oxford, UK) and flow cell R10.4.1 (Oxford Nanopore Technologies) on a GridION platform (Oxford Nanopore Technologies). All amplicons were purified using 1× volume of AMPure XP beads (Beckman Coulter, Inc., IN, USA) to remove primer dimers, and their concentrations were quantified using Qubit (Thermo Fisher Scientific Inc., MA, USA). For the library preparation, 200 fmol of each amplicon was used according to the instructions of the manufacturer. Sequencing was terminated when each sample reached 50,000 raw reads. All procedures were performed separately for the mock and field studies. Fastq files for each sample were generated from pod5 files using Guppy 7.1.4 with super-accurate basecalling mode, a minimum read length of 200 bp, and minimum Q score of 10. Data analysis OTUs were defined using a multi-step pipeline. The raw reads were first trimmed with Seqkit (version v2.10.0)41 to remove the primer sequences and retain reads with lengths between 700 and 900 bp. The filtered reads were then aligned to the PhytoREF database38, a curated reference for plastidial 16 S rRNA genes from photosynthetic eukaryotes, using BLAST. The top hits were extracted, and sequences with < 97% identity were excluded. Taxonomic assignments were based on the best hits, and OTU abundance tables were generated by counting the reads assigned to each genus. Downstream diversity analyses were conducted in QIIME 2 (q2cli version 2024.10.1)42. The alpha diversity was estimated using three indices: the Shannon index, which accounts for both richness and evenness; Simpson index, which emphasizes the dominant taxa and provides a measure of the community evenness; and Chao1 index, which estimates the species richness based on the number of rare OTUs. The beta diversity was evaluated using PCoA to visualize the differences in community composition among the samples. Rarefaction curves were also generated to assess the sequencing depth. All analyses and visualizations were performed in a WSL Ubuntu environment using custom scripts developed by the authors. Quantification and statistical analysis Statistical tests were conducted in QIIME 2. The Kruskal–Wallis test followed by Dunn’s test was used to compare the alpha diversity indices, and PERMANOVA was applied to assess the beta diversity. All analyses were performed in a WSL Ubuntu environment. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Acknowledgements This work was funded by the Public Interest Incorporated Foundation, Konno & Lester Foundation. We are grateful to Dr. Yuri Ota of On-chip Biotechnologies Co., Ltd. for providing valuable information during the preparation of the manuscript. We would also like to thank Editage (www.editage.com) for English language editing. The illustration of the sequencer in Figure 5 was obtained from The Togo Picture Gallery (copyright 2016 DBCLS TogoTV / CCBY-4.0). Author contributions S.Y.: Conceptualization, Methodology, Investigation, Data curation and processing, Writing-original draft, Funding acquisition, Supervision. K.Y.: Investigation. K.K.: Investigation. R.N.: Investigation. I.H.: Writing-review and editing. K.K.: Conceptualization, Writing-original draft, Supervision. Data availability 1. This study does not report any original code.2. All data supporting the findings are available within the article and Supplementary Table 3. The long-read metabarcoding sequencing datasets generated during this study have been deposited in the DDBJ Sequence Read Archive and are publicly available under the BioProject accession number PRJDB35649, with associated BioSamples SAMD01597485–SAMD01597514, Experiments DRX684269–DRX684298, and Runs DRR704184–DRR704213.Any additional information required to reanalyze the data reported in this manuscript can be obtained from the corresponding author upon request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Data Availability Statement 1. This study does not report any original code.2. All data supporting the findings are available within the article and Supplementary Table 3. The long-read metabarcoding sequencing datasets generated during this study have been deposited in the DDBJ Sequence Read Archive and are publicly available under the BioProject accession number PRJDB35649, with associated BioSamples SAMD01597485–SAMD01597514, Experiments DRX684269–DRX684298, and Runs DRR704184–DRR704213.Any additional information required to reanalyze the data reported in this manuscript can be obtained from the corresponding author upon request.