The inclusion of Amazon mangroves in Brazil's REDD+ program

Brazil’s greenhouse gas emissions are strongly driven by deforestation, forest degradation, and conversion to agriculture and pastures. The country’s Intended Nationally Determined Contribution aims to eliminate emissions from current Land Use, Land Use Change, and Forestry by 2030, with REDD+ envisioned as a key financing mechanism. Despite their high carbon stocks and vulnerability to emissions upon conversion, mangroves have not been included in Brazil’s REDD+ framework, likely due to their smaller area relative to upland biomes and limited national recognition of their high carbon density. The study addresses this gap by quantifying Amazon mangrove total ecosystem carbon stocks (TECS), developing IPCC Tier 2 emission factors for typical land-use conversions (pastures, shrimp ponds), and establishing deforestation reference levels to support inclusion of mangroves in Brazil’s REDD+ and NDC accounting.

Prior work shows mangroves possess some of the highest total ecosystem carbon stocks among forests, often at least twice those of upland evergreen forests. Reported TECS for Brazilian mangroves range from 413 to 1851 Mg C ha−1, exceeding stocks in Brazilian upland biomes (Amazon rainforest ~463 Mg C ha−1; Caatinga ~74; Cerrado ~49). Global syntheses indicate broad TECS ranges (79–2208 Mg C ha−1) driven by environmental and physiographic gradients. Deforestation of mangroves for aquaculture and pastures can release 58–90% of TECS, but field-based emission factors are sparse in many regions, leading to reliance on models with uncertainties due to limited in-situ data. Regional assessments from NE Brazil, the Caribbean, and Indonesia highlight substantial TECS losses upon conversion. Global models have also overestimated SOC density and soil C stocks for Brazilian mangroves in some cases, underscoring the need for robust field datasets to refine national and global estimates used in climate policy.

Study area encompassed mangroves across the Brazilian Legal Amazon, including estuarine, delta, and open-coast hydrogeomorphic settings (e.g., Sucuriju, Araguari/Bailique, Curuçá, Maracanã, Bragança). Field sampling included over 190 forest plots and 900+ soil samples. At each site (mangrove and land-use converted areas: shrimp ponds, pasture), six plots were placed 20 m apart along a 100 m transect perpendicular to the shoreline or estuary edge. Control undisturbed mangrove plots were paired near converted areas to derive emission factors (IPCC Tier 2 stock-change approach). Biomass: Tree species included typical Brazilian mangrove taxa (Rhizophora mangle, Avicennia germinans, Laguncularia racemosa) and freshwater-associated species in the delta (e.g., Euterpe oleracea, Pterocarpus sp.). Tree diameters were measured at breast height (1.3 m) within 7 m radius plots for trees ≥5 cm DBH and within 2 m radius nested plots for smaller stems. Rhizophora diameters were measured 30 cm above the highest prop root. Aboveground biomass was estimated using species- and region-appropriate allometric equations; belowground root biomass followed Komiyama et al. Carbon content factors of 0.48 (aboveground) and 0.39 (belowground) were applied. Standing dead trees were included with decay-class adjustments (97.5%, 80%, 50% of live tree biomass for classes I–III, respectively). Downed wood: Estimated via planar intersect method using four 14 m transects per plot with size thresholds and condition classes (sound/rotten); wood mass converted to carbon with a 0.50 factor. Soils: Fixed-volume cores (6.4 cm radius) were collected at plot centers, sectioned by depth (0–15, 15–30, 30–50, 50–100, and >100 cm when applicable). Maximum soil depth was probed; C pools were integrated to 3 m when soil exceeded that depth. Samples were oven-dried (60 °C) for bulk density and analyzed for total carbon (elemental analyzer). A total of 914 soil samples were analyzed. Soil salinity (handheld refractometer) and pH were measured in borehole water at each plot (n=6 per stand). Stable isotopes: Acidified subsamples were analyzed for δ13C and δ15N via IRMS to infer organic matter sources. Emission factors and uncertainty: Emission factors (EFs) for conversion to pastures and shrimp ponds were computed as TECS differences between converted sites and paired pristine mangroves (stock-change, Tier 2), integrating soils to 3 m or full profile. Uncertainty in TECS estimates propagated from component confidence intervals following GOFC-GOLD. Activity data and annual emissions: Annual mangrove deforestation (2016–2021) and land-use classes (e.g., pastures, grasslands) were derived from MapBiomas LULC products to establish Forest Reference Emission Levels (FREL) and compute annual emissions (deforested area × EF), reported as CO2e using 3.67 conversion. Statistics: Differences among hydrogeomorphic settings and land-use effects (impacted vs pristine) were tested via PERMANOVA with post hoc comparisons (α=0.05; data log-transformed as needed). Multiple linear regression assessed influences of precipitation, soil salinity, tidal range, latitude/longitude, and tree basal area on TECS; multicollinearity led to removal of longitude and tidal range. Model selection employed AIC with backward stepwise procedure. Analyses were run in R using stats, vegan, MASS, and ggplot2.

• Total ecosystem carbon stocks (TECS): Amazon mangroves average 468 ± 67 Mg C ha−1 (range 181–903). Aboveground biomass 29–335 Mg C ha−1; total soil C 90–541 Mg C ha−1. TECS in estuarine and delta settings are similar (486 ± 21 and 478 ± 212 Mg C ha−1), while open-coast mangroves hold ~21% less C (377 ± 160 Mg C ha−1; PERMANOVA F=30.72, p=0.01). Total soil C is higher in estuarine (329.3 ± 19 Mg C ha−1) than delta (259 ± 124) and open-coast (207 ± 111; F=6.93, p=0.02). Surface (0–1 m) SOC represents only 19–26% of 0–3 m soil C. Surface SOC density is higher in estuarine/delta (15.9 ± 1 and 16.8 ± 3.9 mg C cm−3) than open coast (12.1 ± 3.4; F=13.7, p=0.01). Downed wood is higher in estuarine settings (F=4.21, p=0.03). - Drivers of TECS: Limited east–west gradients in precipitation (2000–2300 mm yr−1), tidal range (4–6 m), and latitude were not significant predictors. Tree basal area correlates with TECS; best-fit model including basal area and soil salinity explains 36% of TECS variability (F=8.158, Adj R2=0.364). - Isotopes and sources: Soil δ13C/δ15N indicate greater allochthonous freshwater contributions in Amazon Delta mangroves; mixed freshwater and mangrove plant sources east of the river mouth; marine sources minor in open-coast mangroves but evident in shrimp ponds. - Land-use change impacts: Converted sites (shrimp ponds and pastures; N=4) average 123 ± 79 Mg C ha−1 vs 457 ± 82 Mg C ha−1 in paired pristine mangroves (PERMANOVA F=19.59, p=0.01). Conversion results in complete loss of aboveground stocks and 71% (shrimp ponds) to 83% (pastures) decreases in TECS. Mean TECS loss is 351 Mg C ha−1. Surface (0–1 m) SOC density declines from 20.1 ± 3.5 to 5.5 ± 1.2 mg C cm−3 (F=21.96, p=0.04), with bulk density increasing from 0.8 ± 0.23 to 1.7 ± 0.22 g cm−3 (F=19.29, p=0.02). - Emission factors and emissions: Mean potential emissions from conversion are 1228 ± 146 Mg CO2e ha−1, exceeding emissions from conversion in upland Brazilian biomes (Amazon rainforest ~331 Mg CO2e ha−1; Cerrado/Caatinga 55–131). - Deforestation and reference levels: Legal Amazon mangroves cover 795,637 ha. Mean annual deforestation (2016–2021) is 751 ± 248 ha yr−1, >94% to pastures/grasslands. Estimated annual emissions from these conversions are ~0.9 Tg CO2e yr−1. - Mitigation potential: Avoiding mangrove loss could prevent 9.2 ± 0.11 Mt CO2e over 10 years and is equivalent annually to net C accumulation in ~82,400 ha of secondary forests. TECS are ~40% lower than the global mangrove average (856 Mg C ha−1) but within global ranges. Field-based SOC densities are lower than model estimates, implying prior models overestimated stocks; Tier 2 EFs (71–78% TECS loss) are higher than earlier Brazil-wide estimates.

The study establishes robust, field-based TECS and Tier 2 emission factors for Amazon mangroves, directly addressing the data gap impeding their inclusion in Brazil’s REDD+ and NDC frameworks. Findings show that although Amazon mangroves have TECS below the global mangrove average, their per-area emissions upon conversion are the highest among Brazilian biomes, highlighting a strong mitigation leverage from protecting these ecosystems. Variations in TECS are partly explained by soil depth, salinity, and forest structure (basal area), underscoring the role of hydrogeomorphic context in carbon storage. Isotopic evidence indicates substantial allochthonous inputs in deltaic systems, with potential implications for carbon stabilization dynamics. The lower observed SOC densities relative to model outputs refine global estimates, reducing overestimation risks in policy models. Demonstrated annual emissions from recent deforestation (~0.9 Tg CO2e yr−1) and the predominance of pasture conversion (>93%) identify clear management targets. Incorporating mangroves into the national FREL and REDD+ could unlock avoided emissions equivalent to large areas of secondary forest growth, with added co-benefits for biodiversity, fisheries, coastal protection, and resilience. The results thus substantiate the inclusion of mangroves as priority components of Brazil’s climate mitigation portfolio.

This work provides a comprehensive, field-based accounting of Amazon mangrove carbon stocks and Tier 2 emission factors, suitable for integration into Brazil’s REDD+ and NDC reporting. Amazon mangroves average 468 ± 67 Mg C ha−1 and experience 71–83% TECS loss when converted to shrimp ponds or pastures, driving per-area emissions (~1228 ± 146 Mg CO2e ha−1) that exceed upland biome conversions. With recent deforestation rates of ~751 ha yr−1, including mangroves in REDD+ could avoid ~0.9 ± 0.3 Tg CO2e yr−1, or ~9.2 Mt over a decade, with substantial co-benefits. Future research should focus on: expanding field assessments of converted sites across geomorphic settings; quantifying carbon accumulation rates, particularly in the Amazon Delta; elucidating mineral–organic interactions and decomposition controls on SOC; and strengthening monitoring to avoid double counting while improving REDD+ project evaluation and outcomes.

• Limited number of land-use change sites sampled within the Legal Amazon (four shrimp ponds and one pasture), primarily in estuarine settings, constraining the breadth of EF estimates. - Upscaling from sampled plots to regional estimates introduces uncertainty, though emission factor ratios align with global assessments. - Activity data-driven FREL relies on 2016–2021 MapBiomas deforestation rates and assumes business-as-usual trends; future changes in drivers/policies could alter emissions. - Comparison with national FRELs for upland biomes is complicated by differing carbon pool accounting (Brazil’s FREL excludes soil C in upland biomes). - Some model explanatory power for TECS is modest (best model Adj R2 ≈ 0.36), indicating additional unmeasured drivers (e.g., mineralogy, sea-level dynamics) affect variability.