Introduction

Coastal blue carbon ecosystems (CBCEs)1, including mangrove forests, tidal marshes and seagrass meadows, greatly contribute to the total organic carbon (OC) stored in coastal marine sediments, exceeding soil OC sequestration rates per unit area of terrestrial forests by over 20-fold2. CBCEs can be managed to enhance their CO2 burial capacity and/or avoid emissions from extant carbon sinks, contributing up to 841 Tg CO2e withdrawal and emission reduction per year globally3. The restoration of CBCEs as carbon offsetting projects is at its onset4,5,6, in part because a thorough understanding of OC permanence in their soils and its vulnerability to disturbances is lacking7,8. Resolving soil OC dynamics is therefore crucial to assess the long-term CO2 sink capacity of CBCEs and the effect of ecosystem degradation on OC stability and CO2 emissions7,9,10.

The largest OC reservoir in CBCEs is the soil, constituting more than 95% of the total stock in seagrass meadows and tidal marshes, and more than 60% in mangrove forests11,12. Upon burial, a portion of the OC in the soil is mineralized, whereas another portion can be preserved for centuries and millennia13,14. The quality of organic matter, humification, physical protection, anoxia, temperature or bioturbation, have been discussed as OC preservation mechanisms in marine sediments15,16,17. However, the study of OC accumulation/mineralization in BCEs present specific challenges due to the high impact of the plant belowground organs over sediment biogeochemistry10,18,19. Some studies have inferred soil OC decay rates in CBCEs by combining OC content with age-depth models10,19,20,21,22. They rely on the chronologic sequence of these soils where deeper equals older23. As organic matter decays, OC content within the soil is expected to decrease with depth following an exponential decay curve, assuming constant supply. However, previous studies reported either decreasing, increasing or stable OC trends with soil depth and age in a handful of locations18,24. This showcases the complexity of soil OC dynamics in CBCEs linked to the balance among burial, erosion, decomposition processes, and environmental changes over centennial to millennial times scales. Field observations have been contrasted with laboratory and/or field incubations of plant tissues aimed at determining OC decay rates, but there are very few experiments exploring OC decay in soils9,25,26,27,28,29,30. Indeed, the short time span of incubation experiments, encompassing days to up to two years in duration25,26, is likely insufficient to capture soil OC preservation/decay processes at meaningful time scales to aid climate change mitigation strategies31. Furthermore, several natural processes that are difficult to replicate in experiments can influence the preservation/decay of soil OC, including changes in porosity, compaction, bioturbation or the exhaustion of electron acceptors32. Moreover, abrupt changes in microbial community, redox conditions and pH values between the rhizosphere and the subsoil, can have a profound influence on soil OC decay10,33. These processes can play synergic and/or antagonistic roles influencing soil OC preservation and add complexity to understanding soil OC dynamics in CBCEs.

Here, we examined 2174 soil OC profiles from mangrove forests, 1014 tidal marshes, and 573 seagrass meadows across the world to assess OC decay patterns with soil depth (i.e., increasing, stable or decreasing). Furthermore, centennial and/or millennial OC decay rates were modeled in 83 cores where age-depth models were available, had a coherent soil chronostratigraphic sequence, δ13C profile and decreasing OC content with age, following a simple modeling approach developed for peatbogs34.

Results and discussion

Organic carbon trends with depth in coastal blue carbon soil profiles

Tidal marshes and mangroves had higher average OC content in the top 25 cm of soil compared to seagrass meadows (median ± sd; 5.1 ± 7.4, 4.2 ± 8.2, and 0.8 ± 1.7% OC, respectively). From the 3733 cores studied, 63% showed decreasing OC content with increasing soil depth (significant negative correlation, spearman test, p < 0.05, 95% confidence), whereas 15% showed increasing OC content (positive correlation, p < 0.05) and 22% showed relatively stable trends with soil depth (p > 0.05). The percentage of cores with decreasing OC trends was higher in tidal marshes (82%) compared to mangroves and seagrasses (57 and 53%, respectively). The percentage of cores with increasing OC trends ranged from 21% in seagrass to 7% in tidal marshes, whereas the percentage of cores with relatively stable OC downcore trends ranged from 27% to 24% in seagrass and mangroves to 11% in tidal marshes.

Soil OC exponential decay trends with increasing soil depth have been previously reported in terrestrial and freshwater depositional ecosystems such as peats and lakes34,35. These trends are expected in low-energy depositional environments34 where soil depth is a proxy for time36 and organic matter decays exponentially over time26. However, the inputs of OC and inorganic particles to the soil may change along the formation of the deposit and/or the sedimentary record could be disturbed18,24,34, disrupting the expected OC decay trend with increasing soil depth and leading to the development of increasing or stable OC trends. The increasing or stable OC trends found in 18–47% of the cores studied could be explained by multiple synergic or antagonistic biogeochemical interactions, together with anthropogenic and/or natural disturbances acting at different spatial and temporal scales. Mixing of the soil profile due to bioturbation or hydrodynamic, OC inputs from the rhizosphere, or increased inorganic sediment deposition could result in stable or increasing downcore trends, whereas exceptional OC preservation linked to the recalcitrance of e.g., seagrass tissues37 could also explain stable trends at time scales of 100 to 1000 years. Acknowledging that a scenario with constant OC inputs is only theoretical, we hypothesize that given a sufficiently long profile with absence of OC inputs from the rhizosphere or disturbances, the downcore trend will always be an exponential decay. A continuous soil profile with a decay OC trend would be indicative of good preservation of the stratigraphic record, not of mixing.

The relatively high OC content in the top 25 cm of mangrove, seagrass and tidal marsh cores with a decreasing trend (5.9 ± 0.3%, 0.9 ± 0.1% and 5.8 ± 0.2% OC, respectively) compared to those with stable and increasing trends (3 ± 0.2%, 0.7 ± 0.1% and 2.9 ± 0.7% OC, respectively) could indicate that stable depositional environments promote constant soil carbon inputs at the soil surface (i.e., deposition) and/or subsurface (root productivity), where decay is the main factor controlling OC accumulation (Fig. 1A–C). Although it was not possible to explore the role of landscape configuration on OC preservation due to the lack of information in this regard associated with the data compiled, previous studies showed that the habitat extent and patchiness influence the depositional environment. Smeaton et al. showed that the main driver of soil OC stocks in British tidal marshes is their extent38, thus larger tidal marshes have higher stocks per unit area. This is consistent with higher OC accumulation in depositional stable areas, as protected areas within the tidal marsh will have lower hydrodynamic energy than at their edges. The higher the fragmentation of the tidal marsh, the higher its risk of erosion or burial by sediment deposition on top of the soil. Indeed, high tidal marsh areas that are only affected by tidal inundation at spring tides, have been reported to have higher OC stocks, as they have high productivity, work as a sink for allochthonous OC, and grow in a low hydrodynamic environment as reflected by a higher mud content39. Similarly, larger tidal marsh plant species promote both OC accumulation and soil stability throughout a higher productivity of both aboveground and belowground organs40,41 and by lowering hydrodynamic energy and reducing erosion41,42,43. Likewise, OC accumulation in mangrove forests is highly dependent of the seascape configuration44,45, with lower OC stocks in areas more exposed to hydrodynamic energy46,47. Although fine sediments are usually accumulated in low energy environments, in our dataset soil OC decreasing trends in tidal marshes and mangroves are not related to higher mud content (Fig. 1D, F). Tidal marshes and mangroves mostly inhabit coastal deltas and estuaries, which receive relatively higher loads of fine sediments compared to open coastlines48. The median mud content in the compiled database was 67 ± 27% for tidal marshes and 60 ± 31% for mangroves, with no significant differences among cores with decreasing or no decreasing OC soil profiles. Which could indicate that the high content of fine sediment in mangroves and tidal marshes could be masking the resuspension of fine sediment post deposition in no stable areas. The effect of extreme storm events could explain the significantly higher number of stable or increasing OC trends in cores from high sea water temperature areas, as the occurrence of hurricanes, typhoons or cyclones is related to high sea water temperatures49. Typhoons promote deposition of sediments from adjacent mudflats and marine sediments on tidal marsh and mangrove habitats, lowering the %OC of that sediment layer as tidal flats can have lower OC concentration50,51,52. In addition, higher temperatures at lower latitudes result in increased productivity and a larger production of root- derived OC at the subsurface of the soil53, which can disrupt the trend of decreasing OC content with depth. Lastly, human disturbances over mangrove forests and tidal marshes could also lead to the formation of stable or increasing trends with depth, as they impact the relatively shallow rhizosphere and subsoil carbon pools, potentially lowering OC content and disturbing the depth profile54,55,56.

Fig. 1: Differences between decreasing and no decreasing cores.
figure 1

Boxplots of organic carbon (A–C) and mud (D–F) contents (%), sediment accretion rates (G–I; cm yr−1), sea water temperature (J–L, °C), and δ13C (M–O ‰) in the top 25 cm of soil in cores showing decreasing or stable plus increasing downcore OC trends (i.e., non-decreasing). Data in A–C were log-transformed. Boxplot horizontal lines indicate median values; box, 25th to 75th percentiles; whiskers: highest and lowest value excluding outliers (Q3 + 1.5×IQR to Q1−1.5×IQR, IQR: interquartile range); black dots, potential outliers. Significant differences between cores with decreasing and stable or increasing soil OC downcore trends within each ecosystem (Wilcoxon test), p-value: * 0.05– 0.01, ** 0.01–0.001, *** 0.001). Numbers above x-axis labels indicate the number of cores per category.

Full size image

Seagrass cores with decreasing OC trends also showed significant higher OC content in the top 25 cm, but also higher mud content, average δ13C signal, and sea water temperature (Fig. 1C, F, L, O), suggesting that either naturally stable depositional environments and/or the presence of large and persistent seagrass species that stabilize the substrate can result in decreasing soil OC trends57,58. The higher percentage of decreasing OC trends in cores collected from persistent seagrass meadows (60%) compared to more ephemeral meadows formed by relatively small and early-colonizing species (48%), could be explained by their higher productivity and sediment accumulation rates59,60,61. The higher δ13C values in cores with decreasing OC trends also suggests a higher content of seagrass-derived organic matter in these cores pointing to seagrass primary production as the main source of soil organic matter in stable depositional environments62. Indeed, larger species also tend to form highly resilient and continuous meadows59, which stabilize the substrate reducing resuspension and enhancing sedimentation63. Previous studies reported higher OC storage within the central areas of seagrass patches compared to their edges owing to differences in the exposition to hydrodynamic energy64,65. Human or climate impacts affecting seagrass species composition and physiology, like shading, overgrazing, eutrophication, plant die-off or soil erosion, have been reported to affect soil OC fluxes66,67,68. For instance, the replacement of mangroves by seagrasses due to changes in sea level69,70, or the tropicalization of temperate seagrass meadows71,72, could explain some of the increasing OC trends with soil depth found in seagrasses. Contrary to tidal marshes and mangroves, the significantly higher mud content in seagrass cores with decreasing OC trends (Fig. 1D–F) could indicate the formation of decreasing OC trends mostly in stable depositional environments, where fine sediments are deposited and the presence of the seagrass prevents resuspension58. Seagrass meadows usually grown over sandy substrates73. The median mud content in the studied cores was 11.8 ± 21%, six times lower than the mud content of mangroves and tidal marshes cores. The initially low mud content in seagrass meadows may explain why the absence of resuspension is apparent in seagrass cores but not in tidal marsh or mangrove datasets. Contrary to expectations, seagrass cores with decreasing OC trends were associated with lower seawater temperatures but higher OC content—opposite to the anticipated pattern of increased OC under warmer conditions due to enhanced plant productivity74. In contrast to mangroves and tidal marshes, the lack of significant differences in sea water temperature for seagrass soil cores with decreasing and non-decreasing (i.e., stable or increasing) trends (Fig. 1L) suggests that seagrass meadows may be less sensitive to the effects of tropical cyclones. This could be because seagrass meadows are typically located in more sheltered areas, where storm surges may temporarily elevate water levels, placing them in deeper waters75.

Centennial and millennial decay rates

Chronological models were only available for 5% of the 3733 soil profiles compiled from seagrass, tidal marsh and mangrove ecosystems (106, 67 and 52 cores, respectively). Soil OC content followed a decreasing trend with soil age in 66% of these cores (149 cores), being more common in seagrass and tidal marsh (71 and 69%, respectively) than in mangrove (54%) ecosystems. The percentage of decreasing trends in seagrass was higher in the 106 cores with age-depth models than in the remaining 467 seagrass cores compiled. This may be due to a high number of chronological models available from Posidonia genus cores (52 out of 106 seagrass chronological models), which form stable meadows with relatively higher sediment and OC accumulation capacity12. From those cores with a coherent chronological model and a decreasing trend in OC content with age, a decay model could only be fitted in 57% of the cores for any of the eight time frames tested (83 cores).

The average short-term soil OC decay rates (last 80 to 100 years of accumulation) estimated in this study ranged from 0.017 yr−1 in mangroves forest to 0.034 yr-1 in seagrass ecosystems (mean ± SE across ecosystems = 0.023 ± 0.002 yr-1; Table 1). The decay rates over the past 80 to 100 years relate to a dynamic environment where OC decays under a mix of oxic and anoxic conditions within the rhizosphere, with rapid turnaround of labile OC compounds and the interaction of multiple biotic and abiotic factors influencing OC accumulation and preservation, including bioturbation, erosion and OC inputs from multiple allochthonous and autochthonous sources. The 80–100 yr decay values fall within the same order of magnitude of decay rates previously reported for the rhizosphere of seagrass Posidonia oceanica (0.028 ± 0.014 yr-1)10, and tidal marsh Spartina spp. (0.01 to 0.024 yr-1)19. Organic carbon decay rates over the past 80–100 yr were significantly lower in tidal marsh and mangrove forests than in seagrass ecosystems (Fig. 2A; Wilcoxon test p < 0.05). Differences in soil OC decay rates among ecosystems could be explained by the biogeochemical conditions influencing the preservation of OC upon burial, and include the proportion of labile and recalcitrant organic matter, sediment grain-size, redox conditions, the formation of organo-mineral complexes, and microbial activity76,77,78. Previous studies showed that mangroves forests and tidal marshes have the highest content of recalcitrant compounds (i.e., composed of complex aromatic molecules like lignin or humified organic matter) among CBCEs25,79,80 and thereby, should have the lowest soil OC decay rates. However, studies from terrestrial ecosystems showed that old soil organic matter is not necessarily recalcitrant, because, under specific environmental conditions, labile organic matter can be preserved81,82. Furthermore, microbial organic matter utilization with time can enrich the soil in microbial-derived compounds, which are accumulated in the long-term organic matter pool83,84. There was a negative correlation between mud content and decay rates at 100 years when all cores are considered (r = −0.65, p < 0.01), with tidal marsh and mangrove cores showing the highest mud contents and the lowest decay rates and seagrass cores the lowest mud contents and highest decay rates (Fig. S2A). However, the lack of significant relationships between mud content and decay rates within each individual ecosystem (p > 0.05; Fig. S2B) likely indicates that the observed global pattern is spurious. Factors beyond mud content can influence decay rates, including soil OC saturation85,86, and the partitioning of OC in particulate versus mineral-associated fractions87,88. There was no significant effect of seawater temperature in the soil OC decay rates estimated over the past 80–100 years in the whole dataset or within each ecosystem (Fig. 2D supplementary material material). Indeed, temperature is not the main driver of organic matter long term decay in coastal wetlands89. There are likely synergistic effects among several factors—such as mud content, temperature, and the quality of organic matter—on decay rates. However, the limited data available from several cores has hindered the application of multivariate analysis to investigate these interactions. Further studies are needed to better understand the synergistic biogeochemical processes driving differences in soil OC decay rates among blue carbon ecosystems over the past 100 years.

Fig. 2: Decay rates by ecosystem and time frame.
figure 2

A Boxplot of soil organic carbon decay rates (k; yr−1) estimated over 80–100 yr and 500–1000 yr time frames for mangrove forest, seagrass meadows and tidal marsh ecosystems. Boxplot horizontal lines indicate median values; box, 25th to 75th percentiles; whiskers: highest and lowest value excluding outliers (Q3 + 1.5×IQR to Q1 − 1.5×IQR, IQR: interquartile range); black dots, potential outliers. Different lower-case letters indicate significant differences among ecosystems tested by Wilcoxon tests (p-value 0.05). B Exponential regression of soil organic carbon decay rates with age for mangrove forests, seagrass meadows and tidal marshes. Circles indicate k values estimated in this study, whereas triangles show k values from peer-reviewed literature (10, 19–21, Table 2). Vertical lines stand for the standard error of the mean (SEM).

Full size image
Table 1 Soil organic carbon decay rates (k) estimated from mangrove, seagrass and tidal marshes by modeling decreasing organic carbon mass with age over 80–100 and 500–1,000 yr time frames
Full size table

The up to 4-fold decrease in soil OC decay rates 500–1000 yr upon burial in CBCEs compared to the 80–100 yr time frame (Fig. 2B) could be explained by the combination of a depletion with age of easily accessible organic matter compounds and the exhaustion of high energy electronic acceptors13, switching microbial metabolism to less efficient OC mineralization paths. The soil OC decay rates estimated for 500–1000 yr include both surface and deep soil layers. Anaerobic microbial activity predominates in deep soil layers and the remaining organic matter is more scarce and richer in recalcitrant compunds68,90,91,92. There were no significate correlations between decay rates at 500–1000 years and mud, temperature or OC content.

Experimental incubation of Zostera muelleri tissues showed a three-phase microbial breakdown decay26, suggesting that seagrass degradation may be composed of more than one decomposition phase even when the environmental conditions remain constant. In addition, previous studies showed that decay rates of soil OC in CBCEs can be modeled based on different decay rates for labile-young (fast decay) and recalcitrant-old organic matter (slow decay) to account for the depletion of labile organic matter with time10,19,93,94. Therefore, it seems likely that two or three decay phases simultaneously occur within soil OC, spanning different temporal scales ranging from days to millennia. However, the lack of information about the depth or duration of these phases prevents their individual modeling in this study.

Baseline estimates and permanence of OC towards the implementation of blue carbon projects

The role of CBCEs restoration and conservation in offsetting anthropogenic emissions rely on the capacity of these ecosystems to accumulate and stabilize the OC derived from their primary production over meaningful time scales for climate change mitigation (typically agreed as >100 yr95). As most of that OC in CBCEs is stored in the soil, its long-term permanence under distinct conservation status or environmental settings are key factors to decipher to understand their potential to act as net carbon sinks or sources. Based on the simple one pool model used here, the residence time of the OC can be estimated as the inverse of the decay rate96,97. Therefore, the larger the time frame used in the model, the lower the decay rate, and the larger the OC permanence in the soil. This highlights the importance of establishing a standard time frame to estimate soil OC decay rates under business-as-usual conditions in blue carbon crediting certification standards98,99. For instance, 9% of the initial OC buried will remain in the soil after 80–100 yr upon burial based on the 0.023 yr-1 decay rate estimated across all three CBCEs, whereas only 0.1% of the initial OC buried will remain after 500–1000 years (0.007 yr−1; Table 1). It has been suggested that OC cannot be considered truly sequestered until it is below a “mineralization depth”100,101, where its concentration stabilizes. However, we argue that carbon crediting methodologies should embed the natural OC dynamics and fluxes in the soil OC since burial, adjusting the accounting based on the temporal scale involved in the assessment. Albeit the soil OC within the top section of blue carbon soils is more unstable, it still constitutes a standing stock that should be accounted for. Thus, the results presented in this study can be used to holistically estimate the soil OC permanence under business-as-usual scenarios. The soil OC decay rates reached stability between 500 and 1000 yr upon burial (Fig. 2B), which could be used as a conservative estimate of the long-term OC sink capacity of CBCEs.

Incubation studies under different inundation and oxygen availability conditions have shown that soil organic matter in CBCEs has a higher decomposition rate under oxic than under anoxic conditions26,102,103. Because part of the OM will be exposed to a more oxic environment and rapid decomposition after soil disturbances104, ecosystem degradation would enhance the mineralization of soil OC stock. Previous models estimating CO2 release after the loss of CBCEs included decay rates of soil OC under oxic conditions at 0.18 yr-1, an order of magnitude higher than that buried under anoxic conditions (0.018 yr-19). Indeed, soil OC decay under natural conditions estimated in previous studies for seagrass and tidal marsh ecosystems ranged from 0.01 to 0.03 yr-1 over ~100 yr time scales, and from 0.0001 to 0.004 yr-1 over ~500–1000 yr time scales (Table 2). The mean soil OC decay rates over 80–100 yr time frames estimated in this study under natural conditions across mangrove forests, tidal marsh and seagrass ecosystems (0.023 ± 0.002 yr-1) fall within previous estimates. However, the soil OC decay rates estimated over 500–1000 yr time scales (0.007 ± 0.0007 yr-1) were one order of magnitude higher than many previous estimates. This highlights that natural soil OC decay can be higher than initially thought under baseline conditions, but also that there exists high variability in soil OC decay across and within CBCEs and time scales, illustrated by the wide range of decay rates estimated (from 0.0015 to 0.066 yr-1 for the last 80–100 years and 0.00018 to 0.021 yr-1 over the last 500–1000 yr).

Table 2 Soil organic carbon decay rates reported in the literature for blue carbon and terrestrial peatbog ecosystems
Full size table

Notably, soil OC decay rates at 1 m soil depth, with an average age of 1,471 ± 271 yr across the cores studied, were estimated at 0.00005, 0.0002 and 0.0005 yr-1 for mangroves forests, tidal marshes and seagrasses, respectively (Eqs. 1, 2 and 3 Supplementary material). These decay rates of old OC under natural conditions are three orders of magnitude lower than previous decay rate estimates used to model the decay at the top meter of CBCEs10,93,105, showing that the permanence of soil OC in CBCEs may be much higher than initially thought9. Carbon crediting schemes in CBCEs should account for the decay of soil OC under natural conditions (i.e., business as usual scenarios) to then estimate carbon avoided emissions (i.e., additionality) linked to conservation and restoration actions that maintain the preservation of soil OC stocks under anoxic conditions. The results obtained provide baseline data to estimate CO2 emissions derived from the destabilization of CBCEs soils following their degradation or loss, as well as a mean of estimating the enhanced sequestration in restored ecosystems or the avoided emissions in conserved ecosystems towards their inclusion in national and international emissions accounting.

Understanding the permanence of OC in crediting schemes is crucial to constrain current estimates of carbon emissions and removals and thereby, to estimate the climate benefits of blue carbon projects. The soil OC decay rates reported in this study can inform baseline scenarios under natural conditions for Monitoring, Reporting and Verification (MRV) frameworks. They can be used as default values to monitor carbon emissions and removals upon the implementation of conservation and restoration activities. For example, the soil OC decay rates estimated over the past 80–100 years can be used to nuance the estimates of enhanced soil OC sequestration upon restoration, whereas the soil OC decay rates under natural conditions estimated over the past 1000 years should be deducted from the estimates of avoided emissions from soil OC upon the conservation and/or restoration of vegetation.

Limitations of the study

The retrospective assessment of decay rates relies on the assumptions that habitat characteristics including climatic and anthropogenic drivers of OC accumulation, and the quantity and nature of organic and inorganic inputs remained constant over the period of accumulation reconstructed. If this holds true, the accumulated OC mass with depth should show a concave profile depicting decay106. The studied profiles are the result of decades to hundreds of years of OC accumulation. Environmental changes along this time frame can influence the relative proportion of inorganic and organic inputs to the soil, impacting microbial metabolism as well as other allogenic and autogenic factors that may affect the formation of different OC accumulation patterns with soil depth. Therefore, caution must be used when making inferences from fitting OC decay models to sediment-core data in ecosystems dominated by vascular plants32,107. Although multiple factors besides OC decay likely influenced the downcore trends observed in our cores, selecting cores with coherent chronostratigraphic sequences and OC decreasing profiles, and gathering as many cores of this kind from as many different locations and environmental settings as possible, allowed the robust estimation of average OC decay rates in anoxic conditions in CBCEs. However, the estimates provided for mangrove forests only rely on nine data, highlighting the need for further research of soil OC dynamics in this ecosystem. Indeed, applying this method to mangrove forests may present specific challenges, as it assumes that organic matter only gets buried on the top of the soils and ages with depth, dismissing root inputs that can reach deep layers of the soil, with up to a third of root production produced between 30 and 70 cm depth53. However, the subsurface addition of new OC from plant roots or non-plant sources such as soil macrofauna and microbes may result in overestimating OC decay rates in natural conditions108.

The studied cores contain varying proportions of both autochthonous and allochthonous organic matter. The origin of the organic matter significantly influences its quality and likely affects its degradability109. Further studies should aim at differentiating the decay rates of multiple OC pools in order to decipher their trends10. Allochthonous organic carbon, such as aged terrestrial material, may be more resistant to degradation and thus dominate the preserved fraction over time. As a result, using cores with a high allochthonous content could lead to an underestimation of the degradability of autochthonous organic matter.

The fitted models showed a residual distribution that reflects the influence of other processes in our cores. Although a two or three-component exponential decay model would likely better fit soil OC decay dynamics in CBCEs8,10,26, the lack of information on the specific composition of the organic matter of the cores, the depth of their rhizosphere or changes in soil physicochemical characteristics that influence biological or physical protection of organic matter pools110, precludes the application of more complex modeling approaches. A one-component decay model was hence used to avoid overfitting the data and obtain a general equation to estimate decay rates in CBCEs at different time scales96. Furthermore, the low number of modeled cores, and the lack of geochemical and geomorphological information for most of them, prevented the evaluation of how climatic and geomorphic conditions may affect OC decay.

The Posidonia genus is by far the most represented seagrass ecosystem among the age-depth modeled cores. Posidonia is one of the most studied genera in seagrass soil research73 and numerous age-depth models have been published due to their use as environmental paleo archives36,111,112 and their large OC stock compared to other genera61. Furthermore, Posidonia is known to grow preferentially in sandy soils73. This limited diversity across seagrass genera could influence the representativity of the seagrass results presented in this study.

Conclusions

Initiatives to include CBCEs in national carbon inventories and Nationally Determined Contributions (NDCs) towards meeting carbon emission reduction targets are flourishing around the globe. However, OC dynamics in its major reservoir, the soil, are still poorly understood. This study shows that around 9% of the OC buried can remain stabilized in the soil after 80–100 yr and around 0.1% after 500–1000 yr. It also highlights the importance of standardizing time scales when modeling soil OC decay rates and permanence, pointing to the need to establish clear time frames relevant to climate change mitigation when assessing the permanence of soil OC. This study provides baseline estimates of soil OC decay under natural conditions for short- and long-term periods, both relevant under IPCC guidelines, which could also be used to assess carbon abatement and additionally following conservation and restoration actions. Furthermore, the equations provided to estimate decay rates at different time frames for the three main CBCEs (i.e., mangroves forests, tidal marshes, and seagrasses; Supplementary Material, Eqs. 1, 2 and 3) allow the adaptation of these estimations to specific time frames linked to different legislative and blue carbon projects. There are still gaps in our understanding of soil OC dynamics that need to be filled to allow us to develop better OC decay models specific to multiple geomorphic and biologic settings, including the fitting of multi-pool models and gathering data from CBCEs cores from multiple regions and geomorphic settings. Specific studies addressing the interconnected role of organic matter quality and microbial interactions together with a detailed classification of the depositional environment would highly improve our understanding of soil stabilization among different settings.

Methods

The dataset compiled comprised 3761 core data from mangroves forests, tidal marsh, and seagrass soils worldwide (Fig. 3). The data were either directly provided by the authors from published manuscripts10,11,113,114,115 and unpublished datasets, or gathered from published databases linked to peer reviewed manuscripts39,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157. Furthermore, due to the low number of mangrove cores with available chronological models, a specific literature review was performed using the search terms “mangrove” and “sediment accretion rate” on the Web of Science database (accessed February 17, 2025). The cores gathered had to be at least 25 cm long, and have five or more samples with OC or/and organic matter content data, and dry bulk density data along the core profile. Furthermore, the cores had to have been collected in a vegetated area of any of the three studied ecosystems with information about the site location.

Fig. 3: Location of the sampling stations.
figure 3

Geographic distribution of the whole core data collection (n = 3773).

Full size image

Field methods varied across published and unpublished studies, but overall, soil cores were collected by manually pushing PVC tubes into the soil, using piston corers or vibracores. Soil samples were dried at 60–70 °C until constant weight to estimate the dry bulk density. Organic carbon content was measured using an elemental analyzer that was coupled in some instances with a mass spectrometer, after digesting carbonates with 4% HCl. Some studies estimated OC content from the concentration of organic matter assessed by Loss on Ignition technique (combustion at 450 °C for 4 h12), or using published equations for seagrass12 and tidal marshes134. For mangrove studies that only reported organic matter data, the linear relation between organic carbon and organic matter was estimated from the compiled dataset (R2 = 0.06 and 0.9; Supplementary Material Eq 4 and 5).

Statistical analyses and graphical representations were performed in software R using libraries ggplot2, maps, tidyr, dplyr, magrittr, gridExtra, nlstools, MASS, reshape, mosaic and rbacon158,159,160,161,162,163,164,165,166,167,168,169. Soil cores were classified based on decreasing, increasing, and stable downcore soil OC trends based on the correlation between OC content (%) and soil depth or time using the Spearman correlation method, as the data were not normally distributed and not expected to show a linear correlation.

Comparisons between OC trends with depth and geochemical variables (OC content, mud content, sediment accretion rates, and δ13C values; Fig. 1) were run for the top 25 cm of the cores only. The 25 cm standardization was chosen to reduce potential confounding effects linked to historic environmental changes including sediment facies deposited under other ecosystems, and to embrace a larger number of cores as this was the minimum length to add a core to the database. However, it’s important to note that this top soil section that also encompasses the soil OC decay rates estimated over the past 80–100 years, can be highly dynamic due to multiple OC sources entering from the surface and subsurface, and mixed oxic-anoxic conditions within the soil53,68,170. Depth was used to standardize core selection instead of soil age because most (95%) of the cores did not have age-depth models. Mean surface sea water temperature (°C) and water velocity (m s−1) for the 2000–2010 decade were obtained from the Bio-ORACLE database (v3.0)171,172. In those cores located in areas considered “land” under BIO-ORACLE databases, owing to their location close to the coastline or owing to the classification of mangrove forest and tidal marshes as land areas, the software QGis173 was used to get values from the closest “marine” area using “join attributes by nearest neighbor” function.

The chronological models fitted to calculate short-term accumulation rates (last 80–100 yr) based on excess210Pb measurements consisted of either the Constant Flux–Constant Sedimentation (CF:CS) model or the Constant Rate of Supply (CRS)23. A combination of 210Pb and radiocarbon dates were used to calculate long-term accumulation rates (last 500–1000 yr). The cumulative OC mass of the cores was used instead of soil depth to build the chronological models (age-accumulated mass models) to partially compensate for the effect of soil compaction during coring or diagenesis, similar to how 210Pb-based age-depth models are calculated23. This improved the fitting of 210Pb and radiocarbon ages particularly in cores that showed changes in dry bulk density with soil depth.

Decay rates were estimated from those cores that showed both a coherent chrono-stratigraphy and decreasing OC contents with soil depth over the time frame studied, and when available, also stable δ13C values downcore (i.e., 83 cores, Fig. 4). Soil OC decay rates were estimated using a model proposed for peat bogs and previously used in seagrass soils20,106:

$$M=\left(\frac{p}{k}\right)* (1-{e}^{-k* t})$$
(1)

Where M is the OC accumulated mass, t is the age, and OC production (p) and decay (k) are the parameters to be estimated. The normal distribution of the decay rates estimated was tested by the Shapiro–Wilk test, and possible significant differences among ecosystems were assessed by the Wilcoxon test. The R code, raw data and fitted models can be accessed at the GitHub repository (see Supplementary Material). Decay rates were estimated following this methodology for eight different time frames by fitting an exponential decay curve from the top of the soil to the desired age. The maximum sample age for each time frame was set between 80 and 100, 100 and 150, 150 and 300, 300 and 500, 500 and 1000, 1000 and 1500, 1500 and 2000, and >2000 years. Age-depth models for 80–100, 100–150, 150–300 and 300–500 yrs timeframes were built with 210Pb dates, whereas the models for the remaining time frames were built using both 210Pb and radiocarbon dates. The number of cores fitted for each time frames modeled was variable due to differences in core length, number of samples per core and dating methods.

Fig. 4: Location of decay rates sampling stations.
figure 4

Geographic distribution of the cores in which soil OC decay rates were estimated (n = 83).

Full size image

The % soil OC remaining after 100 and 1000 yr upon burial was estimated as:

$$A\left(t\right)=A0* {e}^{k* t}$$
(2)

Where A(t) is the fraction of OC remaining after time t, A0 is the initial mass (in our case 100 to estimate the percentage remaining), k is the decay rate estimated for that time frame, either 100 or 1000 years, and t is the time (either 100 or 1,000 years).

Reporting summary

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