Sustained algal-bacterial interactions promote accumulation of recalcitrant particulate carbon with accelerated radiocarbon aging signatures

Abstract

Algal-bacterial interactions regulate the production and fate of marine particulate organic carbon (POC), yet their capacity to generate long-lived carbon sinks remains unclear. Using a two-year, nutrient-self-sustaining co-culture of Synechococcus and its mutualistic microbiota, we demonstrate that sustained algal-bacterial interactions promote the progressive accumulation of recalcitrant POC. Continuous release of algal-derived organic substrates, coupled with repeated microbial enzymatic reworking, produced inert organic particles, with ~26% resisting remineralization and forming a stable recalcitrant particulate carbon pool. Radiocarbon (Δ¹⁴C) analyses revealed an apparent age offset exceeding 400 years for POC formed within only two-year experiment. This “aged” signature likely results from algal assimilation of fossil-fuel-¹⁴C-depleted atmospheric CO₂ followed by accumulation in generated recalcitrant POC. By demonstrating that apparent radiocarbon age can become decoupled from actual residence time, these results provide insights into the interpretation of “old” marine POC and highlight the important roles of algal-bacterial interactions in ocean particulate carbon sequestration.

Sustained algae-bacteria interactions generate recalcitrant particulate organic carbon with aged radiocarbon signatures within years, revealing an overlooked ocean carbon storage pathway and complicating interpretations of old marine carbon pools.

Introduction

Marine algae, particularly phytoplankton, are major engines of global primary productivity and exert a profound influence on oceanic carbon sequestration and climate regulation^1,2. The well-known major pathway by which phytoplankton contribute to oceanic carbon sequestration is the biological carbon pump (BCP), whereby photosynthetically fixed CO₂ is converted into particulate organic carbon (POC), a fraction of which escapes remineralization in surface waters and sinks to the deep ocean, where it can be stored for centuries to millennia³. However, less than ~1% of the phytoplankton-derived POC can ultimately reach the deep sea, as the vast majority is rapidly remineralized by heterotrophic microbes in the upper ocean^4,5. As a result, the effectiveness of oceanic carbon sequestration depends not only on algal carbon fixation, but critically on microbial processing that governs the transformation, stabilization, and persistence of algal-derived organic matter.

Algae and bacteria coexist in tightly coupled, metabolically active consortia that pervade marine ecosystems^6,7. Within these associations, algal exudates and detrital biomass are continuously transformed by heterotrophic bacteria through extracellular enzymatic degradation, metabolic re-synthesis, and aggregation processes^8,9. Sustained microbial reworking can profoundly alter the chemical composition and reactivity of algal-derived organic matter. For instance, a fraction of algal-released dissolved organic carbon (DOC) can be transformed by microbial communities into recalcitrant DOC (RDOC), a process conceptualized as the microbial carbon pump (MCP), which contributes to ocean carbon sequestration in the dissolved phase^10,11. These algal-bacterial interactions therefore play a central role in determining whether algal production is rapidly remineralized or retained in persistent oceanic carbon reservoirs.

While the MCP has provided a powerful framework for understanding the longevity of dissolved organic carbon, far less attention has been given to the formation of recalcitrant particulate organic carbon (RPOC). Not all POC is equally labile. Certain particulate fractions exhibit strong resistance to microbial degradation and may represent a persistent yet underappreciated pathway for carbon sequestration distinct from classical sinking fluxes^12,13. Emerging evidence suggests that bacterial enzymatic reworking, aggregation mediated by transparent exopolymer particles (TEP), and associations with mineral phases can stabilize POC against remineralization^14–19. However, the extent to which such processes generate RPOC during sustained, metabolically active algal-bacterial interactions remains poorly understood. In natural systems, this question is difficult to address due to environmental variability, continuous particle turnover, and the predominance of short-term observations.

A further fundamental uncertainty concerns the apparent “age” of particulate carbon formed through these biological processes. Radiocarbon analyses have shown that deep-ocean organic matter often exhibits depleted Δ¹⁴C signatures and apparent ages spanning from hundreds to thousands of years^20–23. These signatures are generally attributed to a combination of processes, including selective preservation, incorporation of pre-aged carbon, microbial transformation, and prolonged residence within the ocean interior. By contrast, carbon fixed by contemporary photosynthesis is generally considered “modern”, having been recently assimilated from atmospheric CO₂. Since the Industrial Revolution, however, fossil fuel combustion has introduced large quantities of radiocarbon-depleted “old” CO₂ into the atmosphere, lowering the Δ¹⁴C of the global carbon pool²⁴. Whether assimilation of this ¹⁴C-depleted carbon, together with sustained algal-bacterial carbon cycling, can generate particulate organic carbon exhibiting apparently old radiocarbon signatures despite its recent biological origin remains unknown. Resolving this question is important for understanding how biological processes influence the relationship between apparent radiocarbon age and actual carbon residence time in marine environments.

Although algae and bacteria interact continuously in the ocean, our understanding of POC formation and stabilization under sustained algal-bacterial interactions remains limited. To address this gap, we established a long-term, nutritionally self-sustaining co-culture system comprising a marine Synechococcus strain and its mutualistic bacterial community. Following initial inoculation, the system persisted for two years without any artificial external nutrient supplementation, relying entirely on internal carbon and nutrient recycling^6,25. This uniquely stable and prolonged algal-bacterial co-culture provides an exceptional experimental platform to resolve the formation, transformation, stabilization, and radiocarbon aging of POC under controlled yet ecologically relevant conditions.

Here, we demonstrate that prolonged algal-bacterial interactions facilitate the progressive accumulation of inert, carbon-rich RPOC, accounting for a substantial proportion of total POC and exhibiting a pronounced radiocarbon age offset exceeding 400 years, despite forming entirely within only a two-year experimental period. These observed apparently aged radiocarbon signatures likely result from the algal assimilation of fossil-fuel-¹⁴C-depleted atmospheric CO₂, followed by the progressive accumulation and stabilization of this carbon within a long-lived particulate carbon pool during sustained algal-bacterial interactions. By decoupling apparent carbon age from residence time, our results highlight sustained algal-bacterial interactions as a hidden yet significant mechanism contributing to the persistence and apparent aging of particulate carbon in the modern ocean.

Results

Gradual accumulation of POC during long-term algae-bacteria interactions

Synechococcus maintained long-term viability when cocultured with its mutualistic bacterial communities, persisting for more than 720 days under nutrient-recycling conditions, whereas axenic cultures collapsed after approximately 100 days, likely due to the absence of microbial partners capable of supplying essential nutrients⁶ (Supplementary Fig. S1). Throughout the coculture period, both Synechococcus and bacterial populations were consistently maintained at abundances on the order of 10⁵ ~ 10⁷ cells mL⁻¹ (Supplementary Fig. S1), indicating sustained coexistence and continuous biological activity within the system. Synechococcus exhibited similar growth and decline trends in both axenic and coculture systems during the first 100 days, after which axenic cultures collapsed permanently and never recovered, whereas Synechococcus in the coculture system remained viable throughout the 720-day incubation period and underwent successive physiological phases characterized by growth, decline, chlorosis, and recovery. A pronounced transition occurred around day 280, when Synechococcus recovered from chlorosis and exhibited substantial regrowth (Supplementary Fig. S1), reflecting the establishment of a sustained nutrient recycling regime that regenerated sufficient nutrients to support algal regrowth. This internal nutrient regeneration loop enabled algal-bacterial interactions to persist over extended timescales, promoted a pronounced and progressive accumulation of POC, with concentrations increasing more than sevenfold from 1105.6 ± 205.9 µmol L⁻¹ at initiation to 8075.0 ± 600.0 µmol L⁻¹ by day 720 (Fig. 1A). In contrast, axenic Synechococcus cultures reached a maximum POC concentration of 4320.9 ± 726.9 µmol L^-1, coinciding with the culture collapse. POC concentrations in algal-bacterial cocultures exceeded those in axenic algal cultures prior to day 70, but were transiently lower during days 80-100, likely reflecting the abrupt generation of detrital material following Synechococcus mortality in axenic cultures. After day 100, active biological production of POC in the axenic cultures ceased because the cultures were no longer viable, whereas the cocultures remained active and continued to accumulate POC throughout the 720-day incubation.

Fig. 1. Long-term dynamics of particulate organic carbon and particle morphology in Synechococcus-bacteria cocultures.

A Temporal changes in particulate organic carbon (POC) concentrations and (B) molar C:N ratios in long-term Synechococcus-bacteria cocultures compared with axenic cultures. The line plots represent mean ± standard deviation (SD, n = 3 independent biological replicates); individual data points (dots) represent the mean. C Scanning electron microscope (SEM) images of POC showing the temporal evolution of particulate organic matter morphology in the long-term co-culture systems. Representative images are shown from three independent biological replicates at each time point.

Concurrently, the elemental stoichiometry of particulate matter shifted progressively toward higher carbon-to-nitrogen (C:N) ratios over time, with a more pronounced increase in cocultures (from 3.89 ± 0.46 to 5.63 ± 0.40) than in axenic cultures (from 3.70 ± 0.18 to 4.24 ± 0.06; Fig. 1B). This enrichment indicates preferential carbon retention relative to nitrogen during prolonged algal-bacterial interactions.

Scanning electron microscopy (SEM) revealed marked temporal changes in particle morphology. Early-stage particles (day 30) consisted primarily of recognizable cellular structures and loosely associated heterogeneous material. In contrast, particles collected during late-stage cocultivation (days 500–720) formed compact, amorphous aggregates characterized by reduced porosity and indistinct biological features (Fig. 1C), consistent with progressive structural consolidation of particulate matter. Accompanying this morphological transition, SEM–energy-dispersive X-ray spectroscopy (SEM-EDS) consistently detected enrichment of magnesium (Mg: 13.01–14.53%) and silicon (Si: 11.90–13.20%) in late-stage aggregates relative to early-stage particles (Mg: 10.72–11.71%, Si: 2.91–4.42%; Supplementary Table S1). This co-enrichment suggests an increasingly close association between organic matrices and inorganic components, which may contribute to enhanced particle stability.

Fourier-transform infrared (FT-IR) spectroscopy further revealed compositional changes in accumulated POC (Supplementary Fig. S2 and Table S2), notably an increase in oxygen-containing carbonyls and carboxylates. These functional groups likely facilitate interactions between organic matter and mineral phases through cation complexation with Mg²⁺ and binding with Si-bearing associations, as suggested by the SEM-EDS data (Supplementary Table S1). Concurrent increases in total alkalinity during late-stage cocultivation indicate geochemical conditions favorable for carbonate precipitation (Supplementary Table S3). Together, these observations are consistent with the formation of mineral-associated organic carbon (MAOC), which would further enhance the persistence of particulate carbon in the coculture system.

Prolonged algal-bacterial interactions enhance POC recalcitrance

To assess temporal changes in POC bioavailability, we quantified the recalcitrant fraction of POC derived from long-term (720-day; POC_{720-coculture}) and short-term (30-day; POC_30-coculture) Synechococcus-bacteria cocultures, as well as axenic Synechococcus POC (30-day; POC_axenic), using 500-day dark microbial degradation experiments inoculated with natural seawater microbial communities. POC degradation exhibited biphasic kinetics in all treatments (Fig. 2), with an initial rapid phase (Phase 1, days 0–20) followed by a markedly slower phase (Phase 2, days 50–500). Phase 1 decay rates differed significantly among treatments (analysis of covariance, ANCOVA, p = 2.42 × 10⁻⁵), with POC_30-coculture degrading fastest (k1 = 0.0686 d⁻¹), followed by POC_axenic (k1 = 0.0494 d⁻¹) and POC_{720-coculture} (k1 = 0.0342 d⁻¹; Supplementary Table S4). These initial decay kinetics are consistent with the values recently reported for the primary degradation phase of diverse phytoplankton-derived particles²⁶. In contrast, Phase 2 decay rates were statistically indistinguishable among all three treatments (ANCOVA, p = 0.051), with similarly low decay constants across groups (k2 = 0.0005−0.0014 d⁻¹), indicating that the residual recalcitrant fractions share similar inherent resistance to microbial utilization. Critically, the recalcitrant fraction of POC differed significantly among treatments after 500 days of degradation. The 720-day coculture retained 26.41 ± 0.04% of its initial POC as recalcitrant material, significantly higher than fractions remaining in the 30-day coculture (15.94 ± 0.02%) and the axenic culture (12.92 ± 0.03%) (one-way ANOVA with Bonferroni post-hoc test, p = 0.0035; Supplementary Table S4). No significant difference was observed between the 30-day coculture and axenic culture (p = 0.79). These results indicate that prolonged algal-bacterial interactions promoted the formation and accumulation of recalcitrant particulate carbon. POC loss during Phase 1 coincided with rapid microbial growth during the first few days, followed by a decline in microbial abundance as labile substrates became depleted (Supplementary Fig. S3A).

Fig. 2. Long-term microbial degradation of particulate organic carbon (POC) derived from axenic cultures, short- (30-day) and long-term (720-day) cocultures in dark incubations.

Natural log-transformed percentages of remaining POC (ln(%POC)) are plotted as a function of incubation time. A biphasic decay model was fitted to the data, with circles and triangles representing the initial rapid decay phase (Phase 1) and the subsequent slower decay phase (Phase 2), respectively. The values of k1 and k2 denote the decay rate constants for the two phases. The corresponding coefficients of determination (R²) for Phase 1 and Phase 2 were 0.9130 and 0.8225 for the axenic culture, 0.8758 and 0.5808 for the 30-day coculture, and 0.8733 and 0.7320 for the 720-day coculture.

Although POC also declined measurably in uninoculated control incubations, final POC concentrations in all controls remained significantly higher than those in the corresponding inoculated treatments (Welch’s t test, all p < 0.05; Supplementary Fig. S3B and Table S4). The observed POC loss in controls is likely attributable to residual particle-associated microbes retained on the 0.7 µm filters during POC collection from cocultures, passive dissolution of particles into the dissolved organic carbon pool, and spontaneous lysis of Synechococcus cells during prolonged dark incubation^26,27. Together, these results demonstrate that long-term algal-bacterial interactions progressively increase the proportion of recalcitrant particulate carbon, generating a stable particulate carbon pool resistant to sustained microbial degradation.

Carbon isotope analyses reveal rapid apparent aging of POC over two years

To establish the isotopic baseline of the incubation system, we first measured the carbon isotopic composition of the inorganic carbon sources available for Synechococcus photosynthesis. The atmospheric CO₂ entering the incubation system through vent membranes exhibited a Δ¹⁴C value of −39 ± 2‰. These measurements indicate that the carbon available for photosynthetic fixation already carried a depleted radiocarbon signature. Although all POC analyzed in this study was newly produced within the experimental system after inoculation and in the absence of external carbon additions, carbon isotope analyses revealed pronounced isotopic shifts during the two-year algal-bacterial cocultivation period. Both δ¹³C and Δ¹⁴C values of POC decreased significantly between day 30 and day 720 (Fig. 3). Specifically, δ¹³C values declined by 5.5‰, from −12.2‰ on day 30 to −17.7‰ on day 720, indicating a progressive enrichment of particulate carbon in ¹²C during long-term cultivation. More strikingly, Δ¹⁴C values decreased from −66‰ in the POC_30-coculture to −115‰ in the POC_{720-coculture}, corresponding to conventional radiocarbon ages of approximately 476 and 911 years (years before present, BP), respectively. This shift represents an apparent age offset exceeding 400 years occurring within only two years. Because radioactive decay over this timescale is negligible, the observed decrease in Δ¹⁴C does not reflect true chronological aging. Instead, it indicates that newly produced particulate carbon progressively acquired more depleted radiocarbon signatures during prolonged algal-bacterial cocultivation. The coexistence of increasing POC recalcitrance and declining Δ¹⁴C suggests that sustained algal-bacterial carbon cycling can generate and preserve particulate carbon with apparent radiocarbon ages substantially older than its actual biological age.

Fig. 3. Variation of δ¹³C-POC and Δ¹⁴C-POC during long-term Synechococcus-bacteria cocultivation.

Bar plots show δ¹³C-POC (‰) and Δ¹⁴C-POC (‰) at day 30 and day 720 of long-term cocultivation. Bars represent the mean of independent biological replicates, with error bars indicating the standard deviation (SD) for day 720 (n = 3) and for day 30 (n = 2, due to sampling limitations during the long-term cocultivation experiment); individual data points (dots) represent biologically independent samples.

Functional succession of extracellular enzymes parallels increasing POC complexity

Microbial activity underpins the observed transformations in POC quantity and quality. Given the central role of extracellular carbohydrate-active enzymes (CAZymes) and peptidases in particulate carbon processing²⁸, we analyzed genes encoding these enzymes across the cocultivation period. Metagenomic analyses detected 232 CAZyme families and 137 peptidase families across all samples, with total gene abundances remaining relatively stable over time (9809−11,431 CAZyme genes and 9506−10,857 peptidase genes; Supplementary Data 1). Glycoside hydrolases (GHs), carbohydrate-binding modules (CBMs), and glycosyltransferases (GTs) were consistently enriched in the secretory CAZyme fraction relative to the total fractions, whereas serine peptidases (SPs), metallo peptidases (MPs), and cysteine peptidases (CPs) dominated both total and secretory pools (Fig. 4A). These enzyme classes have the potential to selectively degrade labile substrates and generate chemically complex residues that likely contribute to increasing particulate carbon recalcitrance^29,30.

Fig. 4. Functional classification and phylogenetic affiliation of genes encoding bacterial CAZymes and peptidases in long-term Synechococcus-bacteria cocultures.

A Functional composition of genes encoding CAZymes and peptidases; B Taxonomic variability at the phylum level (class level for Proteobacteria) of genes encoding CAZymes and peptidases.

As POC became increasingly recalcitrant over time, both the α-diversity (Shannon index and richness) and compositional complexity of predicted secretory enzymes increased significantly (p < 0.05; Fig. 5A, B). Taxonomic diversity, including enzyme-associated taxa and the overall microbial community, also increased substantially over the cocultivation period (p < 0.05; Fig. 5A, B; Supplementary Fig. S4), with more pronounced divergence between early (days 0-70) and late (days 450-720) stages. Principal coordinate analyses revealed clear stage-specific clustering of enzyme assemblages (p < 0.05; Fig. 5C). By day 720, several secretory CAZymes involved in murein degradation (e.g., GH24, GH25, CE4) and in the modification or biosynthesis of complex polysaccharides (e.g., GT41, CBM5, CBM85) were significantly enriched relative to day 0^31–34 (Supplementary Data 2; DESeq2, p < 0.05), with log2FoldChange values ranging from approximately 2 to 15. Multiple peptidase families targeting structural and protein-rich substrates were also enriched (Supplementary Data 2). Although enzyme expression alone cannot establish causality, this coordinated functional succession provides a mechanistic framework consistent with progressive POC transformation and stabilization.

Fig. 5. Progressive diversification of genes encoding CAZymes and peptidases and their associated microbial communities during long-term Synechococcus-bacteria cocultivation.

A Temporal changes in the richness and Shannon diversity of genes encoding secretory CAZymes and peptidases and the corresponding microbial communities. B Temporal changes in the richness and Shannon diversity of genes encoding total CAZymes and peptidases and the corresponding microbial communities. Box plots show the median (center line), interquartile range (IQR, 25th–75th percentiles; box), and whiskers indicating the minimum and maximum values; individual data points (dots) represent biologically independent samples (n = 3). Statistical significance was assessed using one-way ANOVA followed by post-hoc Tukey’s HSD test (two-sided); different letters indicate statistically significant differences among time points (p < 0.05), whereas groups sharing at least one letter are not significantly different, and exact adjusted p values for pairwise comparisons are provided in the Source Data for Fig. 5. C Bray-Curtis β-diversity of CAZyme and peptidase gene composition during long-term cocultivation. Data points represent three independent biological samples at each time point (n = 3).

Taxonomic mapping revealed distinct metabolic specialization among bacterial groups. Alphaproteobacteria were the dominant contributors to secretory enzyme genes throughout the cocultivation period, particularly for peptidases (Fig. 4B), while Bacteroidetes and Planctomycetota increased in relative abundance during later stages. From the metagenomic dataset, we recovered 49 high-quality metagenome-assembled genomes (MAGs) encoding a total of 217 secretory CAZyme families and 129 secretory peptidase families (Supplementary Data 3; Fig. S5, S6). MAGs affiliated with Bacteroidota (Cyclobacteriaceae and Flavobacteriaceae), Planctomycetota (Pirellulaceae), and Myxococcota (Sandaracinaceae) exhibited particularly high capacities for secretory CAZyme production, with up to 316 CAZyme genes per genome, while Bacteroidota and Proteobacteria encoded up to 149 peptidase genes per MAG. Collectively, these results demonstrate a coordinated microbial function succession that parallels increasing POC chemical complexity and recalcitrance during sustained algal-bacterial interactions.

Together, these results demonstrate that sustained algal-bacterial interactions facilitate a coordinated progression from POC accumulation to biochemical stabilization and apparent radiocarbon aging, providing experimental evidence that biologically “modern” carbon can rapidly acquire properties characteristic of long-lived, aged particulate carbon.

Discussion

Using a two-year, nutritionally self-sustaining coculture of Synechococcus and its mutualistic microbiota, we demonstrate that prolonged algal-bacterial interactions promote the progressive accumulation and stabilization of recalcitrant particulate organic carbon. Unexpectedly, although all particulate carbon was produced within the two-year experimental period, its radiocarbon composition corresponded to conventional radiocarbon ages of several hundred years. These apparent ages do not represent actual carbon residence times, but rather reflect the development of aged radiocarbon signatures during sustained biological processing. Our study suggests that this pattern may result from the algal assimilation of radiocarbon-depleted atmospheric CO₂ into newly formed particulate carbon, followed by its accumulation during sustained algal-bacterial interactions. Together, these findings indicate that contemporary photosynthetically fixed carbon may exhibit apparently old radiocarbon signatures, independent of long-term physical storage.

The magnitude of POC accumulation observed in our system is substantially higher than typical concentrations in modern open oceans^35–37 (Fig. 1). In contrast to open ocean, where POC levels are regulated by dynamic equilibrium between production, remineralization, and physical export, the elevated POC levels in our coculture system primarily reflect sustained internal recycling, retention, and transformation of organic matter, rather than transient bloom dynamics. Although such concentrations are not directly observed in today’s oceans, geological and sedimentary evidence suggests that comparably high particulate or sedimentary organic carbon reservoirs may have existed in Proterozoic oceans, which were dominated by cyanobacterial primary producers and tightly coupled microbial consortia^38,39. More broadly, biologically mediated carbon retention and stabilization pathways, similar to those demonstrated here, may contribute to the long-term preservation of particulate carbon in the ocean alongside physical burial processes.

A defining feature of long-term algal-bacterial cocultivation was the progressive increase in POC recalcitrance. The higher residual POC fraction and lower decay constant in the long-term cocultures collectively indicate that prolonged algal-bacterial interactions facilitate the accumulation of RPOC, contributing to a stable particulate carbon pool that persists even under sustained microbial activity. This pattern parallels observations from long-term algae-bacteria cocultures where RDOC accumulates over time, supporting a scenario that persistent algal-bacterial interactions contribute to the shared conversion of labile carbon into refractory pools across both particulate and dissolved phases²⁵. Together, these results suggest that RPOC and RDOC formation are not isolated phenomena, but rather represent interconnected outcomes of long-term microbial reworking within tightly coupled autotrophic-heterotrophic systems.

This sustained accumulation of recalcitrant POC also influences carbonate system dynamics within the cocultures. Continuous photosynthetic fixation of CO₂ by Synechococcus, together with bacterial stabilization of a fraction of this carbon into recalcitrant POC, transfers carbon from the dissolved inorganic carbon (DIC) pool into a stable particulate reservoir that is progressively decoupled from rapid remineralization. As recalcitrant POC accumulates, the return flux of CO₂ is reduced, establishing an initial CO₂ deficit relative to air-water equilibrium and promoting continued atmospheric CO₂ influx⁴⁰. Concurrent increases in total alkalinity during late-stage cocultivation further reinforce this imbalance (Supplementary Table S3), enhancing more CO₂ dissolution and its conversion into bicarbonate (HCO₃^-), further deepening the CO₂ deficit and driving atmospheric CO₂ influx^41,42. This sustained atmospheric CO₂ influx replenishes the DIC pool, while Synechococcus maintains primary productivity via carbon-concentrating mechanisms⁴³, supporting continued formation of recalcitrant POC. These coupled pathways constitute a self-reinforcing feedback loop that collectively sustains CO₂ uptake and promotes the accumulation of recalcitrant POC.

The increasing recalcitrance of POC arises from a hierarchy of biologically mediated processes, with selective microbial reworking via extracellular enzymes representing a plausible underlying mechanism. Throughout the cocultivation, metagenomic analyses revealed abundant and diverse genes encoding predicted secretory carbohydrate-active enzymes (CAZymes) and peptidases, indicating substantial microbial potential for the extracellular transformation of particulate organic matter. Although enzyme activities were not directly measured, the diversity and composition of these predicted secretory enzyme repertoires changed markedly and coincided with the shifts in POC composition, stoichiometry, particle morphology, and recalcitrance. Notably, later-stage communities exhibited enrichment of genes encoding predicted enzymes associated with the degradation or modification of structurally complex substrates, including cell wall polymers, murein, and complex polysaccharides^20,28. These patterns suggest functional reorganization of the microbial community in response to adaptation to increasingly complex particulate organic matter during prolonged cocultivation. This enzymatic succession reflects a progressive transition from the utilization of labile substrates to the exploitation of structurally complex particulate matter under continuous carbon supply, thereby facilitating the accumulation of degradation-resistant residues⁴⁴. Interestingly, both microbial community diversity and the diversity of secreted CAZymes and peptidases increased during cocultivation, but their temporal trajectories were not tightly coupled. Although taxonomic diversification occurred throughout the experiment, changes in enzyme diversity and composition were particularly pronounced during the later stages, when POC exhibited greater recalcitrance. This divergence between taxonomic and functional patterns suggests that shifts in extracellular enzymatic potential cannot be explained solely by changes in community composition⁴⁵. Instead, the enrichment of enzyme families associated with the degradation and modification of structurally complex substrates is consistent with the increasing complexity of the accumulated POC pool during prolonged algal-bacterial interactions.

Several secondary processes likely reinforce this enzymatic stabilization. First, long-term algal-bacterial cocultivation produces a continuous supply of cellular detritus, including cell wall fragments, membrane components, and intracellular debris, which are intrinsically more resistant to enzymatic attack than fresh biomass^16,46. Second, the accumulation of RDOC may partially coat particulate surfaces, reducing enzymatic accessibility and further slowing degradation^20,25. Third, the release of transparent exopolymer particles (TEP) promotes aggregation and the formation of dense, compact particulate structures^47–49, as supported by SEM-EDS, FT-IR, and alkalinity data (Supplementary Table S1, S2, S3). Such aggregation can physically shield organic matter from microbial enzymes and enhance resistance to remineralization^47,49,50. Collectively, these processes indicate that POC recalcitrance is not simply inherited from source material, but is actively constructed through sustained microbial processing.

Independent evidence for progressive recalcitrance and aging of POC is also provided by its elemental and stable isotopic characteristics. During long-term cocultivation, POC exhibited a marked increase in C:N ratios (Fig. 1B), consistent with a shift from nitrogen-rich, labile algal-derived substrates toward carbon-enriched, degradation-resistant particulate matter. Elevated C:N ratios are widely associated with reduced bioavailability and slower remineralization, reinforcing the conclusion that prolonged algal-bacterial interactions selectively retain carbon while removing more reactive nitrogenous components^16,27. In parallel, δ¹³C values of POC declined progressively over time (Fig. 3). This trend likely reflects the selective microbial degradation of specific components within the coculture-derived POC. Although all organic matter in our system derives from the same algal source, intracellular fractionation during biosynthesis can generate distinct stable carbon isotopic signatures among different compound classes, with lipids generally more depleted in δ¹³C than carbohydrates and amino acids^51–53. As microbes preferentially remineralize the more labile, δ¹³C-enriched carbohydrates and amino acids, the residual POC becomes relatively enriched in recalcitrant, δ¹³C-depleted lipid-like and other structurally complex compounds^51,54,55. The simultaneous increase in C:N ratios together with a gradual decrease in δ¹³C-POC further support ongoing selective degradation of organic matter during long-term algal-bacterial interactions. While similar δ¹³C shifts in marine POC are often attributed to physical transport and mixing processes⁵⁴, our controlled experiments demonstrate that biological interactions alone can generate such isotopic shifts, highlighting the importance of microbial processing in shaping POC signatures.

A particularly intriguing result of this study is the emergence of increasingly depleted Δ¹⁴C signatures during long-term cocultivation, despite all POC being produced within a two-year experimental period. The decline in Δ¹⁴C corresponds to an apparent radiocarbon age increase of more than 400 years, yet radioactive decay over such a short timescale is negligible. Consequently, the observed shift cannot be interpreted as true chronological aging, but rather as the development of an increasingly aged radiocarbon signature in newly formed organic matter. Notably, this Δ¹⁴C depletion cannot be explained solely by photosynthetic isotopic fractionation. Although photosynthesis preferentially assimilates ¹²C and can therefore decrease δ¹³C values⁵⁶, accelerator mass spectrometry (AMS)-derived Δ¹⁴C values are corrected for mass-dependent isotopic fractionation through δ¹³C normalization⁵⁷. However, this correction does not necessarily exclude a contribution from differential photosynthetic fractionation among ¹²C, ¹³C, and ¹⁴C to the observed Δ¹⁴C depletion. Given that ¹⁴C fractionation is approximately twice that of ¹³C, the observed 5.5‰ decrease in δ¹³C would correspond to an estimated ¹⁴C effect of approximately 11‰, substantially smaller than the observed 49‰ decrease in Δ¹⁴C. Thus, photosynthetic fractionation may account for part of the observed Δ¹⁴C depletion, whereas the remaining shift likely reflects changes in the radiocarbon content of the carbon source and repeated carbon recycling within the incubation. Our baseline measurements indicate that atmospheric CO₂ entering the system carried depleted radiocarbon signatures. Since the Industrial Revolution, fossil-fuel combustion has introduced large quantities of ¹⁴C-depleted CO₂ into the atmosphere, lowering atmospheric Δ¹⁴C through the Suess effect^58,59. Continuous gas exchange between the cultures and the atmosphere, therefore, provided a persistent source of isotopically depleted inorganic carbon available for photosynthetic fixation. Although the Δ¹⁴C of the initial DIC pool in the culture medium was not directly measured, this DIC pool was prepared using an artificial seawater system and exposed to contemporary ambient laboratory air. Consequently, the initial DIC pool would be expected to have carried a depleted radiocarbon signature, although its precise isotopic composition cannot be quantitatively constrained. Assimilation of this carbon by Synechococcus could progressively imprint low-Δ¹⁴C signatures onto newly produced organic matter. While this mechanism can explain the isotopic depletion, sustained algal-bacterial interactions likely played a critical role in preserving and accumulating this carbon as recalcitrant particulate matter. Repeated microbial transformation converted labile algal products into increasingly stable particulate forms, allowing carbon carrying depleted radiocarbon signatures to persist and accumulate over time. In this sense, microbial processing did not necessarily generate the low-Δ¹⁴C signal itself, but facilitated its stabilization within a long-lived particulate carbon pool.

To further evaluate the contribution of sustained algal-bacterial interaction, additional radiocarbon measurements were conducted on particulate material from four-year incubations (Supplementary Table S5). The Δ¹⁴C of coculture-derived POC continued to decline over time, reaching −175‰ and corresponding to an apparent radiocarbon age of approximately 1483 years BP. Compared with Δ¹⁴C values measured at day 720 ( − 115‰; 911 years BP), this shift represents an additional apparent age increase of more than 500 years during the subsequent two years of cocultivation, suggesting that prolonged algal-bacterial carbon cycling can progressively amplify radiocarbon age offsets over time. In contrast, residual particulate material from the long-term axenic incubations, which collapsed by day 100 and were not maintained under strictly sterile conditions thereafter, yielded a less depleted Δ¹⁴C value of −119‰ (953 years BP). Despite these limitations, the observed age offset supports the scenario that sustained algal-bacterial interactions enhance the preservation and accumulation of particulate carbon carrying depleted radiocarbon signatures. Because Δ¹⁴C was measured only for POC throughout the incubation, the relative contributions of atmospheric CO₂ uptake, DIC dynamics, and microbial transformation cannot be fully resolved. Future experiments incorporating time-resolved measurements of DIC, DOC, and POC radiocarbon signatures, together with compound-specific radiocarbon analyses, will be required to quantify the pathways responsible for the observed isotopic evolution. Regardless of the precise mechanism, our results demonstrate that sustained algal-bacterial interactions can generate POC exhibiting centuries-old radiocarbon signatures within years. This finding reveals a biological pathway that decouples apparent radiocarbon age from actual carbon residence time and suggests that at least part of the radiocarbon aging observed in marine particulate carbon may arise from biological processing rather than prolonged isolation alone.

While our controlled algal-bacterial coculture system reveals the biological processes underlying POC stabilization and apparent aging, extrapolating these findings to the natural ocean involves uncertainties. Physical transport, resuspension, grazing, and hydrodynamic mixing will all modulate the accumulation, transformation, and export of particulate carbon in situ. Our incubation was not designed to reproduce these physical processes, and the resulting POC dynamics may therefore differ from those in the natural ocean environment. In the natural ocean, POC produced in the euphotic zone may be remineralized and transformed into DOC or exported downward as sinking particles, where it experiences darkness, cold, high-pressure, and continued microbial processing⁶⁰. The recalcitrant POC signatures reported here reflect processes occurring under illuminated, room-temperature, atmospheric-pressure incubation conditions and therefore do not directly represent the fate of POC in the natural ocean. By disentangling biological effects from physical processes, our study provides a mechanistic foundation for interpreting field observations and generates testable hypotheses regarding the origins of recalcitrant and apparent radiocarbon-aged POC in marine environments.

In all, sustained interactions between algae and heterotrophic bacteria represent a powerful, previously underappreciated pathway that promotes the formation and stabilization of recalcitrant particulate carbon and contributes to the development of apparently aged radiocarbon signatures in contemporary photosynthetically fixed carbon. This process highlights the central role of algal-bacterial interactions in shaping particulate carbon persistence, and underscores the need to reconsider how radiocarbon signatures are interpreted in assessments of oceanic carbon sequestration. Sustained algal-bacterial interactions emerge as a critical yet underappreciated mechanism governing the persistence and apparent aging of particulate carbon in the modern ocean.

Methods

Establishment of a long-term algal-bacterial coculture system

A long-term algal-bacterial coculture system was established using the marine Synechococcus sp. PCC7002 (Pasteur Culture Collection) and a heterotrophic bacterial consortium derived from a nutritionally self-sustaining Synechococcus-bacteria coexistence system reported in an earlier study⁶¹. In brief, the bacterial consortium was collected by sequential filtration of the coculture system (20 mL) through sterilized 1.0 µm and 0.22 µm pore-size polycarbonate membranes (Millipore, Ireland), followed by three washes with sterile artificial seawater (ASW) to remove residual dissolved organic carbon and nutrients. The washed bacterial cells were then inoculated into exponentially growing axenic Synechococcus cultures. Initial abundances of both Synechococcus and bacterial populations were approximately 10⁶ cells mL⁻¹, as determined by flow cytometry. Axenic Synechococcus cultures were verified by epifluorescence microscopy and 16S rRNA gene sequencing prior to inoculation and were maintained as bacterial-free controls throughout the experiment.

All cocultures and axenic controls were established in triplicate in acid-washed borosilicate glass bottles containing modified SN medium⁶² (Supplementary Table S6) prepared with 0.2 µm-filtered and autoclaved ASW. EDTA and vitamin B₁₂ were omitted to minimize the introduction of exogenous organic carbon and to constrain organic matter production primarily to in situ photosynthesis and subsequent microbial transformation⁶². Cultures were incubated at 22 °C under a 12:12 h light-dark cycle with white LED illumination (50 μmol photons m⁻² s⁻¹) in a dedicated, controlled indoor environment without local combustion sources. Culture vessels were sealed with 0.22 µm vent membranes to permit gas exchange with ambient laboratory air while maintaining sterility. Following initial inoculation, no additional inorganic nutrients or organic substrates were supplied for the duration of the incubation, establishing a closed system in which carbon and nutrients were internally recycled through algal production and microbial reworking^6,25. Given that continuous agitation often leads to dense particle clumps⁶³, cultures were instead gently mixed manually once per week to resuspend settled material.

Experimental design and sample collection

Synechococcus-bacteria cocultures were incubated continuously for 720 days as described above. Axenic Synechococcus cultures were maintained under identical conditions and sampled until day 100, after which cultures collapsed due to senescence. Samples were collected from cocultures at 12 time points (days 0, 5, 9, 15, 30, 45, 70, 100, 180, 450, 600, and 720) for the determination of particulate organic carbon (POC), particulate nitrogen (PN), molar C:N ratios, algal and bacterial abundance. Samples collected at days 30 and 720 were selected for detailed carbon isotope analyses and chemical characterization, representing early and late stages of cocultivation. Equal aliquots from triplicate incubations were pooled for FT-IR-based chemical characterization to obtain sufficient sample material. Particle morphology and elemental composition were further examined at days 30, 500, and 720 to characterize the progressive structural evolution of particulate organic matter during cocultivation. To further evaluate the contribution of sustained algal-bacterial interactions to carbon isotopic shift, samples from long-term cocultures and corresponding axenic incubations maintained for four years were collected for isotopic characterization. Carbonate system parameters, including DIC, pH, and total alkalinity (TAlk), were monitored concurrently. Metagenomic samples were collected from cocultures at days 0, 9, 70, 450, and 720 to resolve temporal changes in microbial functional potential. In addition, independent short-term (30-day) Synechococcus-bacteria cocultures were established in triplicate (n = 3) using identical incubation volumes and conditions as the long-term cultures, to generate particulate material for comparative assessments of POC recalcitrance.

Assessment of POC recalcitrance through long-term microbial degradation experiments

The bioavailability of coculture-derived POC was assessed using a 500-day dark microbial degradation experiment. POC harvested from the 720-day (POC_{720-coculture}), 30-day (POC_30-coculture) cocultures, and 30-day axenic (POC_axenic) cultures was collected by gentle filtration onto precombusted (450 °C, 5 h) 0.7 μm pore-size GF/F filters (Whatman, UK). Filters were resuspended in sterile ASW by gentle shaking, recovering approximately 60% of the collected POC¹³. Biodegradation assays were initiated by inoculating POC suspensions with freshly collected natural microbial communities. Initial POC concentrations were adjusted to comparable levels between the POC_{720-coculture}, POC_30-coculture, and POC_axenic treatments. Microbial inocula were obtained from coastal surface seawater (Qingdao, China; 36°05′N, 120°28′E) and transported to the laboratory within 4 h. To minimize background particulate matter and labile DOC carryover, seawater was pre-filtered through precombusted 0.7 μm GF/F filters, incubated in darkness for 48 h to suppress autotrophic activity, and subsequently filtered through a 0.22 μm membrane. Concentrated microbial cells were resuspended in sterile ASW and inoculated into POC suspensions at a 1:5 volume ratio in triplicate. Parallel controls consisted of (i) POC incubated without microbial inoculation and (ii) microbial inocula incubated without added POC. All systems (2 L) were supplemented with NaNO₃ (50 μM) and KH₂PO₄ (3 μM) to prevent nutrient limitation and incubated in the dark at 23 °C for 500 days. Bottles were sealed with 0.22 μm vent membranes to permit gas exchange while maintaining sterility. Flasks were gently shaken prior to subsampling to ensure homogeneous particle distribution. Samples were collected at days 0, 1, 3, 5, 10, 20, 50, and 500 for POC concentration and bacterial abundance analyses²⁶.

Analysis of POC, PN, and chemical parameters

POC and PN concentrations were measured using an ESC 4024 CHNS-O elemental analyzer (NC Technologies, Italy). Samples collected on GF/F filters were acid-fumed with concentrated HCl for 12 h to remove inorganic carbonates, oven-dried at 50 °C, wrapped in aluminum foil, and analyzed using acetanilide as the calibration standard¹³. Analytical precision was ±2%. C:N ratios were calculated on a molar basis. Bulk chemical composition of POC was characterized using Fourier-Transform Infrared (FT-IR) spectroscopy (Nicolet 6700, Thermo Fisher Scientific, USA)⁶⁴. Synechococcus and bacterial abundances were quantified by flow cytometry using a FACSAria II (BD Bioscience, USA)^65,66. Carbonate system parameters were measured following standard protocols^41,67. Additional methodological details are provided in the Supplementary Information (Supplementary Methods).

Carbon isotope analyses and particle morphology

Radiocarbon isotope (Δ¹⁴C) and stable carbon isotope (δ¹³C) analyses were performed on particulate material collected on pre-combusted 0.7 μm GF/F filters. Filters were acidified, dried, sealed in evacuated quartz tubes with CuO and Ag, and combusted at 850 °C for 2 h. The resultant CO₂ was cryogenically purified and manometrically measured on the vacuum pipeline system for isotope analyses. Radiocarbon measurements were performed by accelerator mass spectrometry at the Center for Isotope Geochemistry and Geochronology (CIGG) of the Qingdao National Laboratory for Marine Science and Technology (QNLM) following graphitization by sealed zinc reduction^68,69. The precision of Δ¹⁴C measurements was ± 8.0‰. δ¹³C was determined using a Thermo 253 Plus dual-inlet IRMS with an analytical precision of ± 0.2‰. The Δ¹⁴C values of POC were calibrated using the measured δ¹³C, and conventional ¹⁴C ages (BP) were calculated following the conventions of Stuiver and Polach (1977)⁵⁷. To constrain the isotopic composition of carbon available for photosynthetic fixation, we measured the Δ¹⁴C signatures of atmospheric CO₂. Atmospheric CO₂ was collected using an alkaline trapping method⁷⁰. Briefly, a carbon-free 0.5 M NaOH solution was prepared by pretreatment with saturated BaCl₂ to remove residual inorganic carbon and subsequently exposed to ambient laboratory air for 10 days. Absorbed atmospheric CO₂ was recovered as BaCO₃ following the addition of saturated BaCl₂. The resulting BaCO₃ precipitates were rinsed with ultrapure water, dried at 55 °C, and analyzed for Δ¹⁴C values. These measurements were used to establish the isotopic baseline of inorganic carbon sources available to Synechococcus during cocultivation.

For particle morphological and elemental composition analyses, samples collected on GF/F filters were gently rinsed with ultrapure water to remove salts, freeze-dried, mounted on double-sided conductive tape, and sputter-coated with gold (Au). Samples from three biological replicates at each time point were examined by field emission scanning electron microscopy coupled with energy-dispersive X-ray spectrometry (SEM-EDS, Hitachi S-4800, Japan). Six randomly selected and spatially distinct locations were analyzed per sample to ensure representative compositional characterization^19,71.

Metagenomic functional annotation

DNA extraction, sequencing, assembly, and gene prediction followed previously described protocols (details are provided in Supplementary Methods)²⁵. Genes encoding carbohydrate-active enzymes (CAZymes) were annotated using DIAMOND v2.1.17.171 BLASTp searches against the dbCAN3 CAZy database with an e-value cutoff of 1 × 10^{−20 72,73}. Peptidases were annotated using the MEROPS v12.4 database with the same cutoff⁷⁴. Signal peptides (SPs) were predicted using SignalP⁷⁵ v6.0 to identify secretory enzymes involved in extracellular POC processing. Taxonomic affiliation of enzyme-encoding genes was assigned using DIAMOND blastp searches against the NCBI non-redundant (NR) database (version 2024). The taxonomic composition of the whole metagenomic community was profiled using Kraken2⁷⁶ v2.1.2 against the RefSeq database, and bracken⁷⁷ v2.6.2 was subsequently used to estimate taxon abundances based on Kraken2-assigned reads. Gene abundances were normalized for gene length and sequencing depth using the transcripts (DNA, in this case) per million (TPM) method⁶, representing length- and depth-normalized read counts per million mapped reads in metagenomic datasets. Taxonomic and functional diversity analyses were calculated using the vegan⁷⁸ package v2.7-2 in R for: (i) the overall microbial community profiles at the genus level, (ii) the enzyme-associated taxonomic profiles at the genus level, and (iii) the functional profiles of CAZyme and peptidase families. To quantify temporal changes in functional gene abundance over the cocultivation period, pairwise differential abundance analysis was performed between the late (day 720) and initial (day 0) stages using DESeq2⁷⁹ v1.38.3 in R. The magnitude of enrichment was quantified using the log2FoldChange values generated by the model. Statistical significance was assessed using Benjamini-Hochberg-adjusted p-values.

Statistical analysis

All experiments were conducted in duplicate or triplicate, and data are reported as mean ± standard deviation (SD). All statistical analyses and visualizations were conducted using Prism 8 (GraphPad Software, USA), ImageGP (https://www.bic.ac.cn/BIC/#/), and R (RStudio 4.0), with significance assessed at p < 0.05. The scripts used in this study are available via GitHub at https://github.com/Hanshuang-Zhao/RPOC-associated-script⁸⁰. The recalcitrant fraction (RPOC) was defined as the percentage of POC remaining after 500 days relative to the initial concentration. Differences in RPOC among the 720-day coculture, 30-day coculture, and axenic culture treatments were evaluated using one-way ANOVA, followed by a Bonferroni post-hoc multiple comparisons test. Differences in final POC concentrations at day 500 between uninoculated controls and their corresponding inoculated treatments were assessed using Welch’s t tests. POC decay kinetics were analyzed using a biphasic decay model²⁶. Natural log-transformed percentages of remaining POC (ln(POC%)) were plotted against incubation time and exhibited two distinct degradation phases, consisting of an initial rapid decline followed by a slower long-term decay. Based on the observed temporal patterns, the data were segmented into Phase 1 (days 0-20) and Phase 2 (days 50-500). Phase-specific decay constants (k1 and k2) were calculated as the slopes of linear regressions fitted to natural log-transformed data within each phase. Analysis of covariance (ANCOVA) homogeneity of slopes tests were used to assess differences in decay rates between phases within each treatment and among treatments within each phase.

Author contributions

Y.Z. conceived the study. Y.Z., H.Z. and Z.Z. designed the experiments. H.Z., X.F., S.N., H.Y. and J.S. conducted the experiments and collected the data. H.Z. analyzed the data. H.Z. and Z.Z. wrote the draft manuscript. Y.Z., X.L. and N.J. contributed to manuscript revision. All authors reviewed and edited the manuscript.

Peer review

Peer review information

Nature Communications thanks Rhona Stuart and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the National Key Research and Development Project of China (2023YFE0113100), the global Ocean Negative Carbon Emissions (ONCE) Project, the National Natural Science Foundation of China (42476115, 42476116, 42576100, U1906216), the Taishan Scholar Foundation of Shandong Province (tsqn202408283), Research Project of Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai) (SML2024SP028), the Youth Innovation Promotion Association of the CAS (2023220), the China Postdoctoral Science Foundation (2025M780847).

Data availability

The raw sequence data generated in this study have been deposited in the NCBI’s Sequence Read Archive (SRA) database. The metagenomic raw reads are available in PRJNA978314. The data generated in this study are provided in the Source Data file. Source data are provided with this paper.

Code availability

All scripts used to analyze data and generate figures in this study are available on GitHub at https://github.com/Hanshuang-Zhao/RPOC-associated-script and are archived at https://zenodo.org with the (https://doi.org/10.5281/zenodo.21967954).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77300-x.