Carbon dioxide equivalents (CO_2eq) for both carbon stock and carbon sequestration rate values were estimated using a CO₂:C stoichiometric ratio of 3.67 (i.e., CO₂/C = 44/12 = 3.67), which is used as a multiplying factor to convert carbon atoms to CO₂ molecules. Potential CO_2eq emissions were computed on a “stock-difference” basis (sensu Kauffman et al., 2017) using published mangrove biomass and soil carbon stock estimates (based on Rovai et al., 2018, 2021b as detailed above) and carbon sequestration rates (from the literature review). Further, we coupled degradation-specific carbon emission factors (after Sasmito et al., 2019: Erosion AGB = 1, SOC = 1; Clearing AGB = 0.7, SOC = 0.21; Settlement AGB = 1, SOC = 0.66; Extreme weather AGB = 0.31, SOC = 0.14; Agriculture and aquaculture AGB = 0.83, SOC = 0.52) with a high-resolution map of drivers of mangrove forest loss (covering the period 2000–2016; after Goldberg et al., 2020) to determine the dominant historical cause of mangrove degradation for each Brazilian state. While some mangrove loss drivers may change over time, dominant degradation causes, particularly those driven by climate (e.g., erosion caused by sea level rise and extreme weather events, which affects 85% of the country’s mangrove coverage; Goldberg et al., 2020), are likely to remain as a result of global climate change. Likewise, agriculture or aquaculture and clearing may be harder to reduce in Brazil in the years to come due to increasing relaxation of environmental regulations. Once determined, dominant state-level emission factors were multiplied by carbon stocks in AGB and in soils (top 1 meter) separately and then summed to compute ecosystem-level potential CO_2eq emissions for each mangrove cell in the gridded dataset (that is, AGB and SOC density estimates combined from Rovai et al., 2018, 2021b).

All raster and vector manipulations and geospatial analyses were performed using R (R Core Team, 2020) packages ‘geobr’ (Pereira and Goncalves, 2021), ‘raster’ (Hijmans, 2020), and ‘rgdal’ (Bivand et al., 2020).

Based on recent global estimates (Table 1), Brazil holds on average 8.5% (or 0.32 PgC) of the world’s mangrove organic carbon stocks, partitioned among AGB (0.05 PgC or 6.9% of global stocks), BGB (0.02 PgC or 6.1% of global stocks) and soils (0.21 PgC or 7.0% of global stocks). On a per-area basis, Brazilian mangroves store on average 66, 33, and 241 MgC ha^–1 in AGB, BGB and soils, respectively (from Rovai et al., 2018, 2021b for AGB and BGB, and SOC, respectively). Standardized to the same mangrove forest coverage, these values are comparable to and often more conservative than other studies’ estimates. However, our ecosystem-level carbon stock estimate for Brazilian mangroves is 36% lower than that reported in Hamilton and Friess (2018) due to overestimated SOC density estimates for Brazil (from Jardine and Siikamäki, 2014) used in that study.

Over 80% of all mangrove carbon stocks in Brazil are found in the states of Maranhão (91.3 TgC), Pará (61.2 TgC) and Amapá (47.3 TgC), reflecting extensive coverage which amounts to more than 80% of the country’s total mangrove area (Table 2).

Largest per-area AGB values are also found in these three states (215.5, 205.3, and 166.7 Mg ha^–1 in Amapá, Pará and Maranhão, respectively) as well as in Piauí (143.4 Mg ha^–1), where mangroves develop in nutrient-rich deltaic systems. In contrast, lowest per-area AGB was found in Santa Catarina (56.8 Mg ha^–1), near the austral distribution limit for mangrove forests in the Southwestern Atlantic (Schaeffer-Novelli et al., 1990; Soares et al., 2012). AGB was also lower in São Paulo (84.2 Mg ha^–1) and Rio de Janeiro (83.1 Mg ha^–1), where extensive mangrove areas have been impacted by industrial activities and urban expansion (Soares, 1999; Ferreira and Lacerda, 2016; Moschetto et al., 2021). AGB values <100 Mg ha^–1 were also found in Paraíba, Sergipe, Pernambuco, Ceará and Alagoas where shrimp farming has compromised the structural and functional integrity of Brazil’s drier-climate mangrove forests (Lacerda et al., 2021). AGB values >100 Mg ha^–1 were found in Espírito Santo and Bahia mangroves where the multidecadal stability of more than 70% of the mangrove coverage suggests that the integrity of core areas have been maintained over time (Diniz et al., 2019). Predicted median AGB for Rio Grande do Norte mangroves was also >100 Mg ha^–1 despite mangroves developing in a semi-arid climate and historical damage from shrimp farming (Lacerda et al., 2021). However, this result is likely due to the small number of observations used to constrain biomass predictions for that region (only two AGB values available for Rio Grande do Norte at the time Rovai et al., 2021b study was conducted; Supplementary Table 1). Regarding SOC stocks, deltaic mangroves in Piauí, Amapá, Pará and Maranhão states had lower soil carbon density due to higher inorganic-to-organic ratio per soil volume characteristic of coastal deltaic floodplains when compared to predominantly estuarine or lagoonal mangroves (Rovai et al., 2018; Sanderman et al., 2018; Jennerjahn, 2020; MacKenzie et al., 2020) found in other Brazilian states (Table 2). When summed, carbon stocks in biomass (AGB+BGB) and soils across Brazilian mangroves averaged 341 MgC ha^–1 (range: 297–397 MgC ha^–1), showing little variation among states (e.g., maximum difference of 23% or ∼80 MgC ha^–1) (Table 2). This relatively small variability in per-unit area carbon stocks reflect mangrove plants’ resource partitioning strategies in response to broad geographical gradients (Rovai et al., 2021b), chiefly the role of coastal geomorphology in controlling the ratio between inorganic and organic matter in mangrove soils (Twilley et al., 2018; Jennerjahn, 2020).

Comparatively, on a per-area basis mangroves store between 2.2 and 4.3 times more carbon in the top meter of soil relative to other Brazilian vegetated biomes (Figure 1). Regarding mean carbon stocks in biomass (AGB and BGB combined), mangroves are second only to the Amazon forest, and 2.7–4.7 times higher than other Brazilian vegetation formations.

Currently, only two studies in Brazil report on carbon sequestration in mangrove woody AGB (Table 3). From these studies, carbon sequestration in Brazilian mangroves’ woody AGB was estimated at 3.18 MgC ha^–1 yr^–1, consistent with values reported for a diversity of coastal typologies worldwide (Table 3). Thus, we used this reference value to produce a first order country-level estimate of annual carbon sequestration in Brazilian mangrove AGB, which totals 2.44 TgC yr^–1, equivalent to 10% of all carbon sequestered in mangroves AGB globally.

Long-term carbon sequestration rates (mostly ²¹⁰Pb-dated cores) in Brazilian mangrove soils was estimated at 2.81 MgC ha^–1 yr^–1 (Table 4). While there were no differences (P > 0.05, results not shown) across the coastal geomorphic types found along Brazil’s shoreline, this value is 15–30% higher than recent global estimates (e.g,. 1.94–2.39 MgC ha^–1 yr^–1; Jennerjahn, 2020; MacKenzie et al., 2020; Wang et al., 2020), likely due to the predominance of minerogenic coastlines (deltaic, which accounts for >80% of the country’s mangrove area, and meso- and macrotidal estuarine systems) where deposition of both autochthonous (mangrove detritus) and allochthonous (terrestrial and marine detritus) sediments are amplified (Adame et al., 2010; Kusumaningtyas et al., 2019; Cragg et al., 2020). Importantly, when this national median value is multiplied by the country’s mangrove area coverage, annual carbon sequestration in Brazilian mangroves soils was estimated at 2.14 TgC yr^–1, corresponding to about 13.5% of the total amount of carbon buried annually in the world’s mangroves.

Highest potential CO_2eq emissions (>900 MgCO_2eq ha^–1) resulting from loss of existing mangrove forests were estimated for Rio de Janeiro, Alagoas, Piauí, Pará, Amapá, and Maranhão states driven by the dominance of erosion (Figure 2 and Table 5) where eventually all carbon stored in soils (here based on top 1 meter) and in AGB is lost to the atmosphere. It should be noted, however, that while eroded SOC is rapidly mineralized in aerobic estuarine waters (Sapkota and White, 2021), carbon release back to the atmosphere from biomass loss is not immediate given slow decomposition rates of downed wood in mangrove forests (Romero et al., 2005). Notably, when considering only the top 1 meter of soil to compute such estimates, these values are amongst the highest CO_2eq emissions reported in the literature for other mangrove sites worldwide (Kauffman et al., 2017; Alongi, 2020; Adame et al., 2021). Further agriculture/aquaculture- and settlement-based losses (emission factors of 0.83 and 1.00 for AGB and 0.52 and 0.66 for SOC, respectively) were also anticipated to cause high potential CO_2eq emissions (>500 MgCO_2eq ha^–1) in Espírito Santo, Pernambuco, Rio Grande do Norte, São Paulo, and Santa Catarina states as these activities represent a considerable loss of both aboveground and soil compartments (Figure 2). Lowest potential CO_2eq emissions were linked to episodic extreme weather events that have the potential to release smaller fractions on carbon stored in AGB (31%) and soils (14%) followed by clearing, which can remove substantial carbon stocks in aboveground (70%) and soil (21%) compartments. These estimates are conservative considering only carbon stored in AGB and soils (but not in BGB, since emission factors for this plant compartment have not been established yet) were used to compute potential CO_2eq emissions resulting from mangrove forest loss.

The loss of carbon sequestration potential after mangrove forests are degraded was assumed here to be 100% considering that soil and vegetation loss represent either acute stressors, which ceases mangrove production altogether, or chronic stressors that leave the system more susceptible to eventually collapse (Lugo et al., 1981; Lewis et al., 2016; Krauss et al., 2018). Further, there are currently no consistent degradation-specific emission factors available to estimate loss of carbon sequestration potential as there is for change in carbon stocks resulting from distinct mangrove deforestation causes (e.g., Sasmito et al., 2019).

Based on the current reference value of 2.81 MgC ha^–1 yr^–1 (Table 4), we estimated an annual loss of 10.31 MgCO₂ ha^–1 yr^–1 that would otherwise be buried in mangrove soils. Combined with loss of carbon sequestration potential in woody biomass, based on the current reference value of 3.18 MgC ha^–1 yr^–1 (Table 3), foregone carbon sequestration in Brazilian mangroves annually could total 22 MgCO₂ ha^–1 yr^–1, in line with estimates reported for other mangroves worldwide (23-254 MgCO₂ ha^–1 yr^–1as reviewed in Alongi, 2014).