Macroalgal and seagrass species generate variable amounts of recalcitrant dissolved organic carbon in coastal Japan

The urgency of climate change has heightened international momentum for harnessing ocean-based solutions within local, regional, and global frameworks^1,2. Oceans and coastal ecosystems take up atmospheric CO₂ and store substantial amounts of organic carbon (OC), which is referred to as blue carbon^3,4. Carbon storage processes that confine organic and inorganic carbon in the ocean over timescales of at least 100 years contribute to removing atmospheric CO₂⁵. Traditionally, blue carbon research has focused on ecosystems, such as mangroves, salt marshes, and seagrass beds, which accumulate OC in their sediments⁶. In addition, recent studies have highlighted that macroalgae contribute substantially to the coastal and ocean carbon budget^7,8,9. Although macroalgae have often been overlooked in blue carbon frameworks, their role has been increasingly recognized alongside those of other ecosystems^10,11.

Recent advances in blue carbon research have highlighted that carbon storage by coastal ecosystems is not limited to the burial of OC in local sediments; it also involves the transport and storage of carbon in external pools^4,7. In this context, submerged coastal vegetation, including macroalgae and seagrasses, are particularly influenced by hydrodynamic processes, making the export of OC to external pools a key pathway for carbon storage^12,13,14. A part of the OC that remains and is altered via biotic and abiotic decomposition processes reaches continental shelves and the deep sea, where the OC derived from these ecosystems can be stored in sediments and the water column over the long term (>100 years)^15,16,17,18.

Dissolved organic carbon (DOC) is a huge carbon reservoir in the ocean, and it plays a crucial role in climate regulation^19,20. The export of DOC and its resistance to decomposition have been proposed to be the principal mechanisms of carbon storage by macroalgae^12,21,22 and seagrasses^23,24,25. Several percent of net primary production (NPP) from macroalgae and seagrasses is estimated to be exported below the mixed layer as DOC, contributing to long-term carbon storage, and this flux is comparable to or may even exceed that of particulate OC storage in local and external carbon pools^7,24,26. However, the magnitude and controls of DOC production, export, and persistence remain poorly constrained^27,28,29.

Recalcitrant DOC (RDOC) is the fraction of DOC that is apparently non-accessible and/or resistant to biotic and abiotic processing, and it plays a major role in global carbon cycles and organic carbon storage³⁰. The export of RDOC derived from macroalgae and seagrasses is thought to be a key carbon flow influencing long-term carbon storage in the ocean^12,25. Evidence from degradation experiments indicates that a substantial fraction of DOC released from macroalgae and seagrasses persists as, or is transformed into, RDOC^21,22,31,32. However, previous experiments have been conducted under a wide range of incubation conditions, including differences in temperature, light regime, and experimental duration²⁷. Moreover, macroalgae encompass a taxonomically and functionally diverse group with substantial variation in physiological traits and ecological roles. Rather than attempting to eliminate this variability through strictly standardized experimental designs, integrating DOC release and degradation measurements across diverse species, sites, and environmental conditions can provide a more comprehensive understanding of macrophyte-derived DOC dynamics. At the same time, explicitly evaluating the uncertainty associated with environmental and methodological variability remains essential for interpreting RDOC estimates.

To quantify the contribution of marine macrophyte-derived RDOC in the context of climate change mitigation, predictions of RDOC stability over the long term (>100 years) are required. The reactivity continuum (RC) model approach is gaining popularity as a powerful statistical tool for realistic description of degradation in a variety of aquatic environments, including for DOC^33,34,35, particulate OC³⁶, and marine sediments³⁷. The RC model assumes a continuous spectrum of organic matter reactivities, where the degradation rate progressively decreases over time. This assumption is consistent with recent findings from high-resolution analytical techniques, which have shown that natural organic matter is composed of an enormous number of molecular compounds with differing chemical properties and levels of degradability³⁸. As degradation proceeds, the remaining DOC pool becomes increasingly enriched in compounds with lower reactivity, justifying the time-dependent decline in the apparent decay rate assumed by the RC model. Thus, the RC model provides a suitable framework for capturing the complex and dynamic nature of bulk DOC degradation because it reflects the emergent behavior of a heterogeneous mixture of compounds with varying reactivity³⁹. The RC model contains various uncertainties, particularly due to its assumption of consistent degradation behavior over time and its limited representation of dynamic microbial and environmental interactions, but it is expected to be applicable to the prediction of long-term DOC stability³³.

To improve understanding of the role of macroalgae and seagrasses in the removal of atmospheric CO₂, it is necessary to empirically quantify carbon sequestration via DOC derived from these habitats. Several approaches can contribute to this goal, including biogeochemical tracers, modeling simulations, and incubation-based experiments. Here, we adopted an empirical approach based on degradation experiments to quantify DOC release rates and the recalcitrant fraction of DOC derived from macroalgae and seagrasses. This approach provides a technically feasible, field-based means of assessing DOC dynamics across diverse coastal settings.

We compiled a large empirical dataset by conducting field-bag experiments on more than 20 macroalgal and six seagrass species distributed across cold-temperate to subtropical regions of Japan (Fig. 1 and Table S1), capturing variability among species, sites, seasons, and in situ environmental conditions. DOC release rates derived from these experiments were combined with degradation experiments (conducted over ~300 days) and RC model predictions to estimate the persistence of macroalgae- and seagrass-derived DOC over decadal to centennial timescales (~100 years). In addition, the organic matter components contributing to RDOC persistence were examined by analyzing changes in fluorescent dissolved organic matter (FDOM) and nutrient concentrations during decomposition. Through this empirically grounded approach, we provide quantitative evidence that DOC export from submerged aquatic vegetation represents a substantial pathway for long-term carbon sequestration in temperate to subtropical coastal ecosystems.

DOC release rate and NPP

Compilation of the field-bag experiment results showed that daily DOC release rate per unit biomass varied by two orders of magnitude (5–462 µmol g-DW⁻¹ d⁻¹); however, when averaged within macrophyte types, mean DOC release rates were broadly similar among types, indicating that most variability occurred within types (and among experiments) rather than between type-level means (Fig. 2). The DOC release rates of macroalgae and seagrasses were similar, averaging 100 µmol g-DW⁻¹ d⁻¹ (95% credible interval [CrI]: 83–116 µmol g-DW⁻¹ d⁻¹) and 137 µmol g-DW⁻¹ d⁻¹ (95% CrI: 81–195 µmol g-DW⁻¹ d⁻¹), respectively. Among the macrophyte types, red algae exhibited the highest DOC release rate, which was significantly higher than that of cold-water kelp and Sargassaceae algae. Within the red algae, cultivated Pyropia yezoensis showed particularly high DOC release rates, ranging from 112 to 350 µmol g-DW⁻¹ d⁻¹ (Table S2). Even within the same macrophyte type, there was substantial variation in DOC release rates across experiments conducted at different times, locations, or with different species.

Daily NPP per unit biomass ranged from −107 to 1850 µmol g-DW⁻¹ d⁻¹, and 96% of the experiments showed positive values (i.e., autotrophic) (Fig. 2). The average NPP values for macroalgae and seagrasses were 312 µmol g-DW⁻¹ d⁻¹ (95% CrI: 254–371 µmol g-DW⁻¹ d⁻¹) and 540 µmol g-DW⁻¹ d⁻¹ (95% CrI: 310–790 µmol g-DW⁻¹ d⁻¹), respectively. The rates were higher in seagrasses and relatively tiny macroalgae, such as other brown algae, red algae, and green algae; the average rates of 371–730 µmol g-DW⁻¹ d⁻¹ were significantly higher than those of warm-water kelp and Sargassaceae algae.

DOC/NPP values exhibit considerable variability across experiments, with average values of 0.24 (95% CrI: 0.21–0.28) for macroalgae and 0.31 (95% CrI: 0.03–0.71) for seagrasses (Fig. 2). Excluding warm-water kelp, which had limited data and high uncertainty, the other macrophyte types had an average DOC release equivalent to 23–31% of their NPP. In experiments conducted during the growing period of macroalgae and seagrasses, DOC release rates were positively correlated with NPP, whereas no correlation was observed during the senescent period (i.e., the period of biomass decline) (Fig. 3).

Recalcitrant fraction of DOC derived from macrophytes

In almost all degradation experiments, the concentration of macrophyte-derived DOC decreased substantially by day 90, after which the degradation rate became very slow (Supplementary Fig. 1). At day 300 of degradation, significant differences were observed among macrophyte types, with seagrasses and Sargassaceae algae showing higher RDOC fractions compared to cold-water kelp and red algae (Fig. 4). The average RDOC fraction at day 300 was 31% (95% CrI: 28–33%) for macroalgae and 42% (95% CrI: 29–53%) for seagrasses. The RDOC fractions showed variability across experiments within each macrophyte type, with values ranging from 7 to 68% across all data.

The RC model estimates for RDOC fractions at years 25 and 100 showed similar trends, with clear patterns observed among the different macrophyte types (Fig. 4). At year 100, the RDOC fraction was 25% (95% CrI: 17–34%) for seagrasses, which was significantly higher than that of macroalgae at 14% (95% CrI: 11–16%). Among the six macroalgae types, Sargassaceae algae showed a slightly higher RDOC fraction compared to the other types. The RDOC fractions at year 100 among the six macroalgae types averaged between 10 and 19%.

Fluorescent DOM release and transformation during degradation

Four FDOM components were identified via parallel factor (PARAFAC) analysis of the samples collected from field-bag experiments. The spectral characteristics of the components (Supplementary Fig. 2) were very similar to those of FDOM in other aquatic environments identified previously by PARAFAC analysis. Based on the fluorescence characteristics, the components could be categorized into humic-like (C1), humic- or polyphenolic-like (C2), marine humic- or microbial humic-like (C3), and protein-like (C4) components^31,40,41.

A 300-day degradation experiment of macrophyte-derived DOM showed that changes in fluorescence intensities (FIs) varied depending on the FDOM components (Fig. 5). After the 300-day incubation, the FIs of humic-like components (C1, C2, and C3) increased by a median factor of approximately 1.2 or more (Mann–Whitney U test, p < 0.01). In contrast, the FIs of the protein-like component (C4) decreased by a median factor of approximately 0.4 during the incubation.

Relationships between RDOC fraction, FDOM, and nutrients

The remaining fraction of RDOC after the 300-day degradation was found to be related to several DOM characteristics. The FI of humic-like components (C1, C2, and C3) normalized to DOC concentration showed significant positive correlations with the RDOC fraction (p < 0.05; Fig. 6a–c). In contrast, no significant correlation was observed between the protein-like fluorescent component (C4/DOC) or the DOC/DON ratio and the RDOC fraction (p > 0.05; Fig. 6d, e). The C/N ratio of the macrophyte body was not significantly related to the RDOC fraction (Fig. 6f). Additionally, no significant relationship was observed between the RDOC remaining fraction and the DOC concentration, DOC/TDN, or DOC/TDP ratios (p > 0.05). Thus, the relative availability of DOC, nitrogen and phosphorus during the degradation experiment could not be considered the dominant factor controlling the degradation rate.

Effect of degradation-promoting treatment

We evaluated the effects of potential limiting processes on RDOC degradation, including photodegradation, nutrient availability, and microbial community. In the cases of Saccharina angustata (site ITK) and Sargassum horneri (site HGN), macrophyte-derived DOC (DOC_M) concentration decreased after the first degradation-promoting treatment (Fig. 7). From the second degradation-promoting treatment until day 500, however, there was no significant change in DOC_M concentration. The RDOC fraction remained approximately constant, exceeding the estimated RDOC fraction of year 25 and year 100.

Annual budget of DOC and RDOC

The DOC and RDOC budgets relative to total NPP (NPP_total) were estimated on an annual timescale, based on the average DOC release rate and the recalcitrant fraction of DOC. NPP_total was defined as the sum of DOC release and net biomass production, the latter of which partially includes particulate OC export and herbivory, although not all loss pathways are fully resolved. The proportions of the annual DOC production (NPP_DOC) to NPP_total were estimated to be 36% (95% CrI: 25–45%) for macroalgae and 33% for seagrasses (20–42%), respectively (Table 1). Among the six macroalgal types, the proportion of NPP released as DOC ranged from 20 to 63%, with red algae exhibiting the highest contribution.

At the 25-year timescale, the RDOC budget accounted for 3–9% of total NPP across the seven macrophyte types (Table 1). Over the 100-year timescale, these values slightly declined, ranging from 2 to 8%. The type-specific patterns of the RDOC budget were consistent across both timescales, with relatively higher RDOC/NPP_total observed in warm-water kelp, Sargassaceae algae, red algae, and seagrasses.

This study empirically demonstrates the presence of recalcitrant DOC persisting over a 100-year timescale across a broad taxonomic and geographic range. Our study contributes to the growing body of knowledge on DOC derived from macroalgae and seagrasses. Macroalgae and seagrasses have been known to be important sources of DOC in coastal waters^23,42, and a large portion of released DOC is exported offshore^26,43. However, the fate of this exported DOC, particularly its potential contribution to carbon sequestration through recalcitrant fractions, remains insufficiently quantified.

Our degradation experiments revealed substantial variation in the RDOC fraction among macrophyte types (Fig. 4). At 300 days, the average recalcitrant fraction was over 40% for seagrasses and approximately 30% for macroalgae, with values ranging from 7 to 68% across all samples. Similar variability has been reported in previous syntheses of degradation experiments conducted under a wide range of conditions, including different taxonomic groups, incubation durations, temperatures, and light conditions, where recalcitrant fractions over timescales of several days to several hundred days ranged from 0 to 85%^27,28,29. In our experimental protocol, the observed variation is likely attributable to differences in DOC chemical composition, microbial community composition, and/or substrate availability, reflecting macrophyte seasonality, physiological condition, and variability in the surrounding environment.

Our RC model projection at the 100-year scale showed that recalcitrant fractions remained considerable, on average 25% (95% CrI; 17–34%) for seagrasses and 14% (95% CrI; 11–16%) for macroalgae, indicating a substantial contribution to long-term carbon sequestration (Fig. 4). The RDOC fractions estimated at the 25- and 100-year timescales in this study correspond to what is commonly referred to as semi-refractory DOC (lifetime: ~decade) in oceanographic research^30,44. Although this does not represent carbon sequestration on geological timescales, it constitutes a meaningful indicator within the temporal framework of human climate change mitigation efforts⁴⁵. While the efficiency of DOC transport from surface to deep ocean layers is known to vary regionally⁴⁶, the decadal-scale persistence of macrophyte-derived RDOC likely enhances its potential for long-term sequestration via the biological carbon pump¹².

The RDOC evaluated in this study is hypothesized to be derived from two key sources that have been previously suggested in the literature^31,32: (1) direct production of chemically recalcitrant DOC by macroalgae and seagrasses, and (2) transformation of initially bioavailable DOC into more recalcitrant forms through microbial degradation and re-synthesis, a process associated with the microbial carbon pump^47,48.

Humic-like components did not show a net decrease during the 300-day degradation experiment; instead, their fluorescence intensities increased (Fig. 5), suggesting that they exhibit recalcitrant characteristics and are poorly utilized by microbial communities. The positive correlation between RDOC fractions and initial humic-like FIs indicated that differences in the quantity of humic-like substances underlie the observed variation in DOC recalcitrance both within and across macrophyte types (Fig. 6). The recalcitrant nature of humic-like substances derived from macroalgae and seagrasses has also been reported in previous studies^31,49,50, which have suggested that phenolic compounds, such as phlorotannins and lignin produced by brown algae and seagrasses, contribute to chemical defense mechanisms against herbivory and microbial degradation^21,51. Polycyclic aromatic compounds and highly aromatic compounds have been reported to be present in kelp-derived RDOC⁵².

Our FDOM results confirm that recalcitrant DOC can be produced through microbial transformation of bioavailable DOC (Fig. 5). Such production of RDOC via the microbial carbon pump is considered a major pathway for long-term carbon storage in the ocean⁵³, and our results indicate that DOC derived from coastal macrophytes also serves as a substrate for this process^31,32. However, it should be noted that our analysis was limited to the fluorescent fraction of DOM, and further research is needed to evaluate the overall carbon budget associated with the microbial carbon pump. Quantifying the relative contributions of different DOM sources represents an important next step in this field. These findings highlight the importance of DOM molecular composition and transformation processes in determining the fate of carbon exported from vegetated coastal habitats.

Our field-bag experiments revealed wide variability in DOC release rates per unit biomass among and within macrophyte types, with values spanning two orders of magnitude (Fig. 2). Similar levels of variability have also been reported in global syntheses, where area-based DOC release rates vary by up to three orders of magnitude²⁷. The large variation observed in our study is likely attributable to species-specific traits^54,55, life stages and seasonality⁵⁶, metabolism and stoichiometry⁵⁷, and differences in environmental conditions across habitats⁵⁵.

The DOC/NPP ratio suggests that a substantial fraction of primary production is allocated to dissolved carbon pathways over daily timescales (on average, 24% for macroalgae and 31% for seagrasses; Fig. 2). These values are consistent with previous estimates, including a global synthesis for macroalgae that reported a mean DOC/NPP ratio of 30% ± 13%²⁷. The positive, albeit weak, correlation between DOC release and NPP observed during the growth phase implies that DOC exudation is actively linked to photosynthetic activity⁵⁸ (Fig. 3), but it may also be modulated by internal stoichiometric balances, such as C:N ratios within the tissues⁵⁷. In contrast, during the senescent phase, DOC release remained relatively high despite very low daily NPP, suggesting that passive leakage of DOC may dominate during this period, as cellular integrity deteriorates and internal organic compounds are released into the surrounding water^31,56.

Annual-scale assessments of RDOC budgets are essential for quantifying the contribution of coastal macrophytes to blue carbon storage and climate change mitigation¹². Expressing RDOC retention as a proportion of annual NPP also facilitates the scaling of local observations to spatially broader carbon budgets. This study provides globally relevant insights by quantifying DOC release and RDOC retention across a broad taxonomic and spatial gradient. Previous estimates relied on literature values and first-order approximations, but these studies established critical baselines^7,24. Our work advances the field by providing empirical measurements through field and degradation experiments. It should be noted that the previous estimates encompass total DOC export to the deep ocean, not specifically the recalcitrant fraction, and are therefore not directly comparable to our RDOC-focused assessment.

Our approach estimated that 36% (95% CrI: 25–45%) of macroalgal NPP_total is released as DOC on an annual basis, which closely aligns with a previous estimate (23%)⁷. The RDOC budget at the 100-year timescale accounted for an average of 4% (3–6%) of macroalgal NPP, with type-specific estimates ranging from 2 to 7%. These values are slightly lower than a previous estimate of deep export (~7%)⁷, yet broadly comparable in magnitude. This convergence supports previous first-order estimates that a substantial portion of the exported DOC would contribute to carbon sequestration.

Our data suggest that 33% (20–42%) of NPP_total is released as DOC from seagrasses. This estimate falls within the range reported in previous studies, including values of ~7%²⁴, ~46%²³ on average from a global synthesis, and 16–64%⁵⁹ from recent empirical studies. These comparisons highlight the substantial variability and uncertainty in previous estimates, while indicating that our results are broadly consistent with the existing range. The 100-year RDOC budget for seagrasses reached up to 8% (4–12%) of NPP_total, which exceeds the previously estimated 2% deep export value²⁴. These findings imply that earlier global assessments may have underestimated the long-term sequestration potential of DOC originating from seagrass meadows.

Our estimates of the RDOC budget generally exceed the ranges estimated for particulate organic carbon export for macroalgae⁷ and seagrasses²⁴, suggesting that DOC export is potentially a more substantial pathway of carbon sequestration^14,26.

This study has several uncertainties and limitations. One limitation is the limited spatiotemporal representativeness of DOC release measurements. Most of the field experiments were conducted within constrained spatial and temporal windows, particularly during the active growth phase of macrophytes, and therefore, they do not fully capture the variability in DOC release across different environmental conditions, seasons, and phenological stages. For instance, DOC release can increase substantially during the senescent phase in some kelp species⁵⁶, and such seasonal variation is not fully accounted for in our dataset. Moreover, DOC production rates are known to vary with plant functional types and environmental drivers, such as temperature, light availability, and nutrient status⁵⁵. In our dataset, however, no significant relationship between in situ water temperature and DOC release rate was detected when examined within each macrophyte type, suggesting that temperature did not systematically explain the observed variability at the macrophyte type level (Supplementary Fig. 3). Brown algae and seagrasses were sampled across a wide range of temperatures, but the number of experiments was limited for some groups—particularly red algae and green algae—potentially introducing stronger seasonal sampling bias for these types. Additionally, a substantial portion of macrophyte biomass subjected to grazing or microbial degradation may be fragmented and transformed into dissolved phases^31,60, representing a potentially important pathway in coastal carbon cycling.

DOC in the surface ocean can be transported below the mixed layer on timescales shorter than 100 years, potentially leading to its isolation from the atmosphere for extended periods⁶¹. In regions where this vertical export process predominates, our 100-year degradation-based estimates of RDOC persistence may represent a conservative (i.e., underestimated) scenario.

Water temperature during DOC degradation plays a critical role in microbial activity and, consequently, DOC decomposition rates⁶². Our degradation experiments were conducted at 22 °C, which is comparable to the global average sea surface temperature (SST) and closely aligned with the annual mean SST in temperate regions. In cold-temperate and subtropical sites, degradation rates may be overestimated and underestimated, respectively. Nevertheless, considering the objective of this study to estimate broadly representative average values, the use of a standardized, intermediate temperature can be regarded as a robust experimental setting.

Our decomposition experiments were conducted under controlled dark and closed conditions, which may overestimate the recalcitrant fraction of DOC compared to natural environments characterized by variable light exposure⁵⁰, nutrient supply²⁸, microbial community²⁸, and substrate concentrations⁶³. In this study, RDOC is operationally defined as the DOC fraction remaining after long-term (>300-day) dark microbial incubation, and we acknowledge that this definition does not imply absolute chemical inertness but rather resistance to microbial degradation under the applied conditions. We did not quantify microbial abundance during the incubations and therefore cannot fully exclude the possibility that microbial biomass limitation contributed to DOC persistence, in addition to intrinsic substrate recalcitrance.

To address this uncertainty, we conducted degradation-promoting experiments that included light exposure, nutrient enrichment, and microbial inoculation, aiming to test whether the persistence of DOC could be explained solely by extrinsic limiting factors. These experiments indicated that, although an initial fraction of DOC was susceptible to enhanced degradation, a substantial DOC pool remained resistant even after repeated treatments, suggesting that RDOC persistence cannot be attributed solely to photodegradation, nutrient availability, or microbial community limitations. However, these degradation-promoting experiments were conducted on only two macroalgal species and did not include seagrass-derived DOC, which represents a limitation in extrapolating these findings across all macrophyte types. The absence of significant correlations between the recalcitrant DOC fraction and indicators of substrate availability further suggests that nutrient or organic matter limitation alone is unlikely to explain the substantial DOC persistence observed in our experiments.

The RC model used to estimate long-term RDOC fractions assumes consistent degradation behavior over time, which may not fully capture dynamic microbial and environmental interactions. Nevertheless, compared to models that assume the existence of a permanently nondegradable refractory fraction, the RC model offers a more realistic framework by representing the continuous degradation of molecules with varying reactivity. This conceptual approach is increasingly supported by recent advances in molecular fingerprinting studies, which have revealed a spectrum of molecular lability rather than a clear-cut division between labile and refractory DOM^33,39.

Future studies are needed to address several key knowledge gaps. First, molecular fingerprinting approaches are necessary to quantitatively confirm the presence of macroalgae- and seagrass-derived DOC in RDOC reservoirs across the ocean. In particular, ultrahigh-resolution mass spectrometry techniques, such as Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS), would provide critical insights into the molecular composition and recalcitrant properties of DOC fractions that resist long-term degradation⁶⁴. Recent studies have shown that kelp can produce forms of DOM regarded as recalcitrant, including carboxyl-rich alicyclic molecules (CRAM)³². Second, microbial community analyses associated with DOC transformation would enhance our mechanistic understanding of RDOC persistence. Third, integrating DOC-based carbon fluxes with physical oceanographic models could improve assessments of the likelihood of deep-ocean transport and long-term sequestration¹⁴. Advancing these research areas will refine estimates of DOC-mediated carbon sequestration originating from vegetated coastal habitats, including macroalgae and seagrasses, and clarify their roles in the global carbon budget as well as the scalability of blue carbon-based climate interventions.

Study sites

We conducted field-bag experiments at 18 sites spanning Japan’s subtropical, temperate, and cold-temperate regions (24–44 °N, 124–145°E; Fig. 1 and Table S1). The experiments were carried out across various seasons from February 2019 to August 2022. Seven of the sites are located in the cold-temperate zone. Among them, four sites (AKS, KSR, ITK, and MRR) along the Pacific coast are influenced by the Oyashio Current, and three sites (SKM, KYD, and BTK) along the Sea of Japan are affected by the Tsushima Warm Current. At these cold-temperate sites, experiments were conducted using species inhabiting the cold-temperate zone, including cold-water kelps (Saccharina angustata, Saccharina japonica, and Saccharina longissima), seagrasses (Phyllospadix iwatensis and Zostera marina), small brown algae (Fucus sp.), green algae (Ulva sp.), and red algae (Mazzaella japonica) (Table S2). Sites TKO and ARS, located along the Pacific coast in the temperate zone and facing the open ocean, are influenced by the Kuroshio current. Experiments at these sites were conducted using warm-water kelps (Ecklonia cava and Eisenia bicyclis) (Table S2). The six sites (FTS, ONK, KNZ, HNN, NRT, and HGN) located in Tokyo Bay and the Seto Inland Sea are in semi-enclosed coastal environments. Experiments at these sites primarily focused on cultivated macroalgae (Pyropia yezoensis, Undaria pinnatifida, and Saccharina japonica), Sargassaceae species (Myagropsis myagroides, Sargassum fusiforme, Sargassum horneri, Sargassum macrocarpum, and Sargassum patens), warm-water kelps (Ecklonia cava ssp. kurome), seagrasses (Zostera caulescens and Zostera marina), small brown algae (Hydroclathrus clathratus, Padina arborescens, and Undaria pinnatifida), green algae (Ulva sp.), and red algae (Gelidium sp. and Gloiopeltis sp.) (Table S2). At WKM, a site facing the Sea of Japan in the temperate zone, experiments were conducted using Sargassaceae species (Sargassum horneri) (Table S2). At the two sites (FKD and MST) on Ishigaki Island, located in the subtropical zone, experiments focused on subtropical seagrasses (Cymodocea rotundata, Enhalus acoroides, and Halodule pinifolia) and Sargassaceae species (Sargassum ilicifolium) (Table S2).

Field-bag experiments

Field-bag experiments^22,49,65 were performed on-site to quantify the metabolic parameters of macroalgae and seagrass species and to collect dissolved organic matter (DOM) released from their thalli and aboveground shoots. Whole detached thalli were used for the experiments of macroalgae; they were acclimated in ambient seawater for over an hour before measurements began. Previous research has confirmed that the detachment process does not influence DOC release from the thalli in short-term experiments⁵⁷. Seagrasses with above- and belowground parts were sampled intact as much as possible to minimize damage. Sediments attached to the roots were rinsed off, and the samples were acclimated in ambient water for over an hour before analysis.

The thallus of each macroalga and the shoots of each seagrass were placed in a plastic bag (polyethylene; volume adjusted according to macrophyte biomass, approximately 1–45 L) containing ambient seawater. Before placement into plastic bags, epiphytes on the macroalgal thalli and seagrass shoots were gently removed by hand. The volume of seawater and the wet weight of the thallus and shoots were recorded for each bag. To prevent air from entering, the open end of the bag was securely tied. For certain macroalgae and all seagrasses, the holdfasts and belowground parts were positioned outside the bag by tightly tying the bag’s opening around the upper portion of the holdfasts and belowground structures, to minimize potential contributions from loosely attached sediments and associated materials. The experimental bags were placed at depths matching the macrophyte’s growth range as closely as possible.

Triplicate sets of transparent and dark bags were prepared to evaluate changes in dissolved inorganic carbon (DIC), total alkalinity (TAlk), and DOC caused by macrophyte metabolism²². To offset the contribution of phytoplankton, a control set of transparent and dark bags containing only ambient seawater was established. Water samples were collected from the bags at the start of the experiment and approximately 4 h later. Bags were retrieved by divers and opened for sampling either at the boat side or in a nearby shallow area.

Since the detachment of macroalgae thalli from substrata and alterations to their surrounding conditions can induce stress, the experimental design minimized incubation time (4 h) and avoided lifting thalli from the sea surface to the air, thereby reducing potential stress as much as possible. Although we cannot rule out that the exposure of the belowground parts of seagrass to the water may have some impact on carbon metabolism, a previous study has shown that there is no difference in photosynthetic activity between cultivation with the belowground parts buried in sediment and hydroponic cultivation with the belowground parts exposed to water⁶⁶. While such stress effects cannot be entirely excluded, their influence on DOC release is likely limited over the short experimental duration; thus, the measurements are considered to provide a reasonable approximation of DOC release under natural conditions. Moreover, the experimental setup is also relevant to understanding DOC dynamics associated with detached and transported biomass in natural environments.

To further minimize handling-induced stress and ensure comparability across a wide range of species and sites, sampling was conducted only at the beginning and end of incubation. Although this two-point measurement introduces some uncertainty, the design reflects a trade-off to balance experimental feasibility and stress reduction in field conditions.

Water samples were processed immediately on board or near the experimental site as described below. Samples for DIC and TAlk were collected in 250-mL Schott Duran bottles and preserved with 1550 μL of 10% benzalkonium chloride solution per bottle (see the Supplementary Methods 1 for a description of the procedure). For DOC analysis, water samples were filtered through 0.2-μm pore size polytetrafluoroethylene (PTFE) filters (DISMIC–25HP; Advantec, Durham, NC, USA) into precombusted glass vials (450 °C for 2 h, 50 mL capacity). The DOC samples were then stored at −20 °C until analysis. Although a previous study has reported that freezing can alter DOC concentrations in relatively high-DOC samples (DOC > 5 ppm)⁶⁷, the DOC concentrations of our samples were below this range. The thalli and shoots in each bag were immediately transported to the laboratory, where they were weighed, dried at 60 °C for more than 24 h until reaching a constant weight, and re-weighed to determine their water content. Water temperatures during the experiments were measured by a conductivity-temperature sensor (Infinity-CT, JFE Advantech, Hyogo, Japan).

Gross primary production (GPP), community respiration (R), net primary production (NPP), and DOC release were determined by analyzing the changes in DIC, TAlk, and DOC during the field-bag experiments. These metabolic parameters per unit biomass were calculated based on the equations proposed by Watanabe et al.²² (see the Supplementary Methods 2 for a description of the calculation). Daily rates of the metabolic parameters were estimated by incorporating the macroalgal and seagrass biomass in each bag, the photoperiod length, and the hourly metabolic rates derived from the transparent and dark bag experiments. The photoperiod was defined as the time between sunrise and sunset, and its duration was obtained using Automated Meteorological Data Acquisition System (AMeDAS) data provided by the Japan Meteorological Agency (available at https://www.jma.go.jp, last accessed: 22 January 2025).

Degradation experiments

To quantify the recalcitrant fraction of that DOC derived from macroalgae and seagrasses due to microbial activity, as well as to assess the change of DOM components, water samples for degradation experiments were collected after the field-bag experiments. DOM samples were obtained from each transparent bag containing macroalgae and seagrasses, as well as from the control bags. The collected water samples were filtered through precombusted (450 °C for 2 h) glass-fiber filters (0.7-μm pore size; GF/F, Whatman, Maidstone, Kent, UK) under reduced pressure. GF/F filters were assumed to allow a substantial fraction of free-living bacteria to pass through into the experimental samples, as suggested in previous studies^21,31.

A total of 40 mL of the filtrates was transferred into precombusted (450 °C for 2 h) 100-mL glass vials, which were then sealed with Teflon-coated rubber stoppers and aluminum caps. Each vial had a 60-mL headspace containing approximately 540 μmol of oxygen, sufficient to support the aerobic microbial degradation of DOC (~220 μmol per bottle), assuming a 1:1 molar ratio of oxygen consumption to the mineralization of DOC into CO₂. The degradation experiments consisted of six incubation periods (0, 10, 30, 90, 150, and 300 days) per field survey, with triplicate vials used for each incubation period. The samples were incubated at room temperature (22 °C) in total darkness until further analysis. The degradation of the organic matter was evaluated under uniform conditions (room temperature) for all samples.

After incubation, the samples were filtered through 0.2-μm PTFE filters (DISMIC–25HP; Advantec) into precombusted (450 °C for 2 h) 100-mL glass vials. The filtrates were then frozen at −20 °C until DOC and fluorescent DOM (FDOM) analyses.

Long-term incubation in closed systems may alter the composition and activity of microbial communities⁶⁸ (i.e., the bottle effect), potentially influencing degradation dynamics. To minimize such effects, experimental conditions were standardized across all samples, and the degradation-promoting experiments described in the next section were conducted to support the interpretation of DOC reactivity.

Degradation-promoting experiments

The experiments described above were conducted in a dark, closed system using natural seawater, where microbial degradation was assumed to be the primary process responsible for DOC decomposition. However, under such experimental conditions, processes such as photodegradation by sunlight, nutrient limitation, and microbial inactivation might lead to an underestimation of the actual degradation potential. To address this, additional experiments were conducted to evaluate whether the fraction of RDOC, estimated in the dark and closed system, deviated significantly when degradation-promoting treatments were applied to macrophyte-derived DOC.

The target species for the degradation-promoting experiments were selected from globally representative macroalgae types: cold-water kelp (Saccharina angustata from the ITK site in May 2022) and Sargassaceae algae (Sargassum horneri from the HGN site in April 2022). As controls, environmental water samples were also used. Samples collected during field-bag experiments were filtered through GF/F filters (nominal pore size 0.7 μm), and 500 mL of filtrate was dispensed into acid-washed 1-L polycarbonate bottles. All treatments and controls were incubated in identical polycarbonate bottles under the same experimental conditions, allowing potential container-related effects (e.g., leaching or sorption) to be accounted for through comparison with environmental water controls.

The degradation-promoting treatment was intended to mimic oceanic DOC processing (mixing, photochemical alteration, and subsequent microbial reworking). Four replicates were prepared for each condition, and the samples were incubated in the dark at room temperature with continuous shaking to maintain homogeneous conditions and simulate physical mixing. The first degradation-promoting treatment was applied on day 150, when the decrease in DOC concentrations had reached a relatively stable level. A second treatment was conducted on day 180 to evaluate whether residual DOC could become degradable after renewed photochemical and microbial processing (i.e., to reassess its recalcitrance).

For the first treatment, sample water from each bottle was filtered through 0.2-μm PTFE filters to remove as many microorganisms as possible and define the DOC pool. The filtrate was transferred to new 1-L polycarbonate bottles and exposed to sunlight for four days. After light exposure, a solution of KNO₃ and KH₂PO₄ was added to each bottle (final concentrations: NO₃⁻, 50 μM; PO₄³⁻, 3 μM), ensuring that nitrogen and phosphorus were not limiting relative to DOC concentrations based on the Redfield ratio. Additionally, 8 mL of environmental water filtered through GF/D filters (nominal pore size, 2.7 μm; collected from the macroalgae bed in Tokyo Bay) was added to each bottle to introduce new microbial communities, and dark incubation was resumed. The addition of the inoculum resulted in an increase in total organic carbon (TOC) concentration of at most 3% relative to the incubation water, which was smaller than the standard deviation of DOC concentrations among the four replicates. Therefore, the contribution of DOC and fine particulate organic carbon potentially introduced with the inoculum was considered negligible. GF/D filtration was used to minimize the introduction of nonmicrobial particulate components while retaining microbial cells for inoculation. Importantly, this inoculation strategy allowed us to assess whether potential size-selective biases associated with GF/F filtration could influence DOC degradation.

For the second treatment, microorganisms were again removed by filtration through 0.2-μm PTFE filters, followed by a 4-day sunlight exposure. Fresh microbial communities filtered through GF/D were then introduced, and dark incubation was continued. Samples for DOC analysis were collected at days 150, 180, 300, and 500.

Chemical analyses

The concentrations of DIC and TAlk were measured using a batch-sample analyzer (ATT-05 and ATT-15; Kimoto Electric, Osaka, Japan) following the method described by Watanabe et al.⁶⁵. The analytical precision of the system, calculated as the standard deviation of multiple reference replicates, was typically within ±2 μmol L⁻¹ for both DIC and TAlk. DOC concentrations were analyzed at least in triplicate using a total organic carbon analyzer (TOC-L; Shimadzu, Kyoto, Japan). The measurements were based on the non-purgeable organic carbon method. Potassium hydrogen phthalate (Wako Pure Industries, Osaka, Japan) was used as the calibration standard. The coefficient of variation for the analyses was less than 2%.

The excitation-emission matrix (EEM) of FDOM was analyzed using a spectrofluorometer (RF6000, Shimadzu, Japan). EEMs were recorded across excitation wavelengths ranging from 250 to 550 nm at 5-nm intervals and emission wavelengths from 290 to 600 nm at 2-nm intervals. To correct for the inner-filter effect, absorbance spectra were obtained using a spectrophotometer (UV-1900, Shimadzu, Japan). The EEM spectra were normalized to Raman units by referencing the area under the Raman scattering peak at 350-nm excitation. To further remove interference from water Raman scattering, the EEM spectrum of MilliQ water (Raman-normalized) was subtracted from the sample EEM spectra. The EEM data were processed using parallel factor analysis (PARAFAC), a multivariate statistical approach that decomposes FDOM fluorescence into distinct components. The PARAFAC analysis was performed with the DOMFluor toolbox in the MATLAB software environment.

For the experimental samples, TOC and TN contents in the plant body were analyzed using an elemental analyzer (Flash EA 1112; Thermo Electron, Bremen, Germany). Concentrations of nutrients (total dissolved nitrogen [TDN], total dissolved phosphorus [TDP], and dissolved organic nitrogen [DON]) in seawater samples collected from the field-bag experiments were quantified using spectrophotometry combined with wet oxidation²⁵.

Data analyses

The reactivity continuum (RC) model was used to simulate long-term decay of macrophyte-derived DOC. Macrophyte-derived DOC (DOC_M) was calculated at each of the six sampling time points during the degradation experiment as the difference between the DOC concentration in the macrophyte treatment and that in the control. In the RC model, DOC degradation is described by the following equation:

where DOC_M0 and DOC_Mt indicate the DOC_M concentrations at the start time of incubation and on day t of decomposition, respectively (Supplementary Fig. 1). In this equation, α represents the average lifetime of more reactive compounds, while v (dimensionless, 0 < v ≦ 1) serves as the shape parameter³⁴. Consequently, the apparent decay coefficient k (in days⁻¹) can be expressed by Eq. 2, emphasizing that the decay rate is not constant; rather, it gradually decreases over time.

The RC model was applied to estimate the RDOC fraction remaining after 25 and 100 years of degradation.

We used R software for statistical analyses. The minpack.lm package was used to fit the RC model to the observed decline in DOC concentration during degradation through a least-squares approach. Additionally, the nlraa package was employed to evaluate the uncertainty of the estimates of the RDOC fraction using a bootstrap method.

Macroalgae and seagrasses were classified into seven types based on their habitats and ecological characteristics (Table S2). The cold-water kelp group included species from the genus Saccharina, while the warm-water kelp group comprised species from the genera Ecklonia and Eisenia. The Sargassaceae algae group included members of the genera Sargassum and Myagropsis. The other brown algae group encompassed species from Undaria, Fucus, Hydroclathrus, and Padina. The red algae group consisted of species from Pyropia, Gelidium, Gloiopeltis, and Mazzaella, and the green algae group included species of the genus Ulva. Finally, the seagrass group comprised species from Zostera, Cymodocea, Enhalus, Halodule, and Phyllospadix. Each type was statistically evaluated based on four metrics: DOC release rate, NPP, DOC/NPP, and RDOC fraction. The data for these metrics were derived from the field-bag and degradation experiments conducted in this study, as well as additional DOC release rate data for warm-water kelp Ecklonia cava obtained from previous field-bag experiments (SMD site, Fig. 1 and Table S2)⁴⁹.

These four metrics were used as effect sizes in the analysis. The average value and uncertainty of each effect size for each macrophyte type were estimated using a random-effects meta-analysis, which accounts for between-experiment heterogeneity arising from differences in species, sites, and experimental conditions. A Bayesian framework was adopted using the metaBMA package, allowing uncertainty to be explicitly propagated through the analysis. We estimated weighted mean effect sizes across experiments using reported standard errors as measures of within-study uncertainty.

Weakly informative priors were used for the overall mean effect size and for the between-experiment heterogeneity, reflecting limited prior knowledge while constraining parameter estimates to realistic ranges. The primary objective of the analysis was to estimate the posterior distribution of mean effect sizes for each macrophyte type. Differences among macrophyte types were inferred from pairwise comparisons of posterior estimates; contrasts were considered supported when the 95% credible intervals for the type-specific mean effect sizes did not overlap. Results were visualized using orchard plots generated in R with the orchaRd package.

Annual DOC budget calculation

Using the DOC release rates and the recalcitrant fraction of DOC obtained in this study, we estimated the annual-scale DOC budget relative to NPP. Here, the annual net primary production including macrophyte biomass (NPP_total) was calculated using Eqs. 3 and 4:

where NPP_DOC and NPP_biomass (g-C m⁻² yr⁻¹) represent the annual DOC production and plant body production, respectively. NPP_DOC was calculated by multiplying the average DOC release rate estimated for each macrophyte type (RR_DOC; g-C g-DW⁻¹ yr⁻¹) by the annual mean biomass (B_mean; g-DW m⁻²) of the macrophyte bed. The estimation error for the annual DOC budget was derived through error propagation using the bootstrap method using R software.

The variables B_mean and NPP_biomass, which are required to calculate the ratio of NPP_DOC to NPP_total, were obtained from compiled literature values (Table S3). The proportion of RDOC release relative to NPP_total was calculated by multiplying the NPP_DOC/NPP_total ratio by the recalcitrant fraction of DOC estimated in this study. Here, NPP_biomass represents net biomass production and implicitly includes biomass losses associated with tissue senescence, detachment, and export occurring during the growth period. The compiled NPP_biomass values are primarily derived from methods, such as leaf marking and stratified clipping techniques, which inherently account for some of these losses. Nevertheless, NPP_biomass may still underestimate total production, because not all loss processes are fully captured.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.