Mangrove forests are crucial carbon sinks, but historical deforestation, coupled with their high carbon density, has led to substantial carbon emissions. Over the past 50 years, extensive mangrove deforestation has occurred due to conversion to aquaculture, agriculture, and urban development. While conservation and restoration programs have increased in recent decades, the net impact of mangrove land cover change (LCC) on global carbon stocks remains unclear. Previous studies quantified carbon emissions from deforestation but lacked data on carbon gains from reforestation or afforestation and the proportion of carbon lost after conversion. This study aims to address this gap by quantifying net changes in the global mangrove carbon stock between 1996 and 2016, considering both mangrove loss and gain, and the dynamic nature of carbon accumulation and loss during these processes. Mangrove forestation, including both natural regeneration and anthropogenic restoration efforts, has the potential to significantly offset carbon losses from deforestation. Additionally, the proportion of mangrove carbon remaining in the ecosystem after deforestation varies, with some carbon lost rapidly and some remaining for extended periods. Therefore, a comprehensive assessment requires accounting for both gains and losses and the temporal dynamics of carbon stocks.

Existing research on mangrove carbon stock change primarily focused on quantifying carbon at risk of emission due to deforestation by overlaying carbon stock maps with areas of forest loss. However, these studies lacked information on gains in mangrove carbon stock due to reforestation/afforestation and the proportion of mangrove carbon lost after conversion to other land covers. The dynamic nature of mangrove ecosystems, with natural forestation and deforestation occurring alongside anthropogenic activities (afforestation and reforestation programs), necessitates a more comprehensive approach. The accumulation of carbon in regenerating or newly established mangroves is not immediate; it progresses over time, potentially taking decades to reach the stock levels of established forests. Similarly, the proportion of mangrove carbon lost after deforestation is not uniform; some is lost rapidly, while some may remain for a significant duration. These complexities highlight the need for a study incorporating these factors to accurately estimate net changes in mangrove carbon stocks.

This study quantified net changes in the global mangrove carbon stock between 1996 and 2016 using recently released datasets describing mangrove forest extent. It combined data on mangrove deforestation and forestation from the Global Mangrove Watch (GMW) datasets with information on the proportion of carbon stocks lost after conversion (from a literature synthesis) and the rate of carbon stock accumulation during forestation (from a meta-analysis of blue carbon ecosystem restoration). The study quantified the carbon at risk of loss due to LCC (D), the potential carbon gain from forestation (F), the proportion of carbon lost due to deforestation over time (rt), and the proportion of carbon accumulated by foresting mangroves over time (at). A simulation framework estimated the net loss in carbon stock (D − rtF at) between 1996 and 2016. Uncertainty was addressed using a bootstrap simulation method with 1000 replications to generate median estimates and 95% confidence intervals. The simulation incorporated uncertainties in mangrove area change, carbon stock density, deforestation/forestation dates, and proportional carbon loss/accumulation rates. Several sensitivity analyses were conducted to assess the impact of methodological decisions on the results, focusing particularly on the Southeast Asia region where more detailed data was available. These analyses explored different methods for estimating carbon stock changes. Specifically, the study considered four scenarios: (1) carbon at risk of loss due to deforestation, (2) net loss of carbon assuming 100% carbon loss and gain, (3) loss of carbon due to deforestation accounting for carbon loss rates, and (4) net loss in mangrove carbon stock accounting for both forestation and carbon loss/accumulation rates. The GMW datasets, derived from ALOS PALSAR and Landsat satellite data, were used to map mangrove extent in 1996 and 2016. Mangrove-related LCC was defined as conversion between mangrove and other land or water covers. Spatial patterns in mangrove carbon densities were quantified using published datasets of soil carbon and above-/belowground tree biomass carbon. Uncertainty in carbon stock was modeled as a normally distributed random variable using the reported RMSE from the original studies. Temporal losses of soil and biomass carbon were modeled based on meta-analyses, with uncertainty incorporated by simulating the dates of deforestation/forestation and accounting for regression model errors. The accumulation of carbon in foresting mangroves was estimated using a meta-analysis of blue carbon ecosystem restoration. Spatial variability in net gains and losses was mapped by aggregating simulated values to a global grid.

The total area of mangrove forest decreased from 142,865 km² in 1996 to 136,717 km² in 2016, with approximately 8050.4 km² of deforestation and 2243.3 km² of forestation. The net loss of mangrove extent was about 5807.2 km² (4% of the 1996 area). The mangrove carbon stock in 1996 was estimated at 8627.6 Mt (95% CI: 4274.6–13,767.0 Mt). The carbon stock at risk of loss due to deforestation (D) was 462.5 Mt (95% CI: 209.7–758.6 Mt). The maximum potential carbon stock gain due to forestation (F) was 118.2 Mt (95% CI: 47.1–200.8 Mt). Without accounting for carbon loss and gain rates, the net loss was 344.3 Mt (95% CI: 9.0–711.6 Mt). Accounting for carbon loss and gain rates reduced the net loss to 158.4 Mt (95% CI: −156.8 to 525.9 Mt), equivalent to 1.8% of the original carbon stock. The lower confidence interval suggests a potential net gain in mangrove carbon stocks. Substantial spatial variation existed in carbon stock changes, with significant losses in the Caribbean and Southeast Asia and gains in scattered regions across Africa, South Asia, and Central America. Detailed analysis showed high variability within regions, with localized net losses in parts of Mexico, Borneo, and Papua, and net gains in parts of Mexico and Myanmar. The contribution of mangrove forestation in reducing net carbon loss was substantial, highlighting the importance of accounting for both deforestation and forestation when assessing changes in carbon stocks. Similarly, accounting for the proportion of carbon lost following LCC significantly reduced the estimated net carbon loss. The study estimates that if the net loss in mangrove carbon stock were released into the atmosphere as CO2, the total emission would be 580.4 Mt (95% CI: −574.6 to 1927.0 Mt), a relatively small contribution to global CO2 emissions (0.6% of emissions from LCC and less than 0.1% of total global CO2 emissions). The relatively low mangrove deforestation rate (0.3% per year) coupled with substantial forestation resulted in a low net loss rate (0.2% per year). A sensitivity analysis indicated that correcting for potential overestimation of soil organic carbon stocks in past research could lead to 35% lower estimates of net mangrove carbon stock loss in Southeast Asia.

The findings highlight the importance of accounting for mangrove forestation and the temporal dynamics of carbon loss and accumulation when assessing net changes in mangrove carbon stocks. The significant reduction in estimated net carbon loss after incorporating forestation and proportional carbon loss/gain rates underscores the importance of considering both aspects of LCC. The relatively small net change in mangrove carbon stock, despite deforestation, suggests that conservation and restoration efforts have yielded some success, although further investigation is needed to pinpoint the turning point in mangrove deforestation rates. The study distinguishes between natural mangrove regeneration and anthropogenic restoration, recognizing that natural processes such as range expansion (due to climate change) and land abandonment may contribute to carbon gains. Anthropogenic reforestation, while showing mixed success due to factors like poor planning, has contributed in certain locations, such as in southern China. The results emphasize that while mangroves hold high carbon densities, their small geographic extent limits their overall contribution to global carbon fluxes, making it essential to avoid overestimating their role in climate change mitigation. Further research is needed to determine the relative contributions of natural regeneration and anthropogenic restoration to the observed carbon stock changes.

This study provides the first comprehensive estimate of net changes in global mangrove carbon stock due to LCC, considering both deforestation and forestation and the dynamic nature of carbon loss and accumulation. The findings reveal relatively low net losses, suggesting some success for conservation and restoration efforts. However, considerable uncertainties remain, highlighting the need for continued research to refine methodologies and data collection. Future studies should incorporate additional carbon pools (e.g., inorganic carbon in sediments and dead biomass), account for mangrove degradation, and improve the accuracy of carbon stock density estimates. Furthermore, a more detailed understanding of carbon fluxes associated with mangrove LCC is needed for a complete accounting of carbon emissions.

Several sources of uncertainty affect the accuracy of the estimates. First, the study focused on soil organic carbon and biomass carbon, neglecting other carbon pools. Second, only changes in mangrove cover were considered, overlooking smaller-scale degradation processes such as timber extraction and pollution. Third, the estimates rely on existing mangrove carbon research, which may be subject to biases. Fourth, the use of a general blue carbon restoration curve to estimate carbon accumulation in foresting mangroves may not perfectly capture mangrove-specific dynamics. Fifth, false-negative classification errors in mangrove maps were not included in the uncertainty analysis, potentially impacting the results. Finally, the sensitivity analysis focused primarily on Southeast Asia due to data limitations, limiting the generalizability of these findings to other regions. The uncertainties are quantified through confidence intervals and sensitivity analysis.