Remote-Sensing Carbon Stock Dynamics and Carbon-Market Valuation in Ecuador’s Churute Mangrove Ecological Reserve (2015–2021)

Abstract

Mangrove ecosystems are recognized as highly efficient blue-carbon reservoirs, yet their monitoring requires scalable, transparent methods suitable for climate-finance and greenhouse-gas accounting applications. This study quantifies interannual carbon-stock dynamics and derives a carbon-market valuation indicator for Ecuador’s Churute Mangrove Ecological Reserve (2015–2021) using publicly available remote-sensing land-cover products. Annual activity data were derived from Copernicus Global Land Service LC100 (100 m, 2015–2019) and ESA WorldCover (10 m, 2020–2021), harmonized to a common reporting scheme, and combined with IPCC Tier 1 default coefficients for biomass and soil organic carbon in tropical wetlands. Total carbon stocks averaged 1.67 million t C across the period, remaining stable within the internally consistent LC100 phase (2015–2019), with trend statistics treated as descriptive given the short annual series, while a pronounced drop in 2020 primarily reflected methodological discontinuities between products rather than ecological change. Converted to CO₂e equivalents (mean 6.1 million t CO₂e), illustrative market values fluctuated between USD 18 and 123 million annually, driven predominantly by carbon-price variability. This remote-sensing-based, MRV-aligned approach provides a conservative baseline for protected-area blue-carbon accounting, highlighting the need for homogeneous high-resolution time series to distinguish real dynamics from classification artifacts in future assessments.

1. Introduction

Coastal zones are simultaneously climate front lines and high-value natural infrastructure: they concentrate people and assets while buffering against storms, flooding, and shoreline change. Recent syntheses emphasize that climate-driven sea-level rise, warming oceans, and intensification of coastal hazards are already altering exposure and risk, and that adaptation outcomes depend on place-based evidence that is comparable across time [1,2]. In this context, coastal wetlands have moved from being treated as “local conservation priorities” to being recognized as globally material components of climate strategies, because they can reduce risk while contributing to mitigation when managed to avoid emissions from degradation [3].

Within coastal wetlands, mangroves occupy a particularly strategic position for climate mitigation because of their high carbon densities and the large share of carbon stored belowground in organic-rich soils [4,5]. Whole-ecosystem measurements repeatedly show that mangroves can store very large carbon stocks per unit area, with substantial variability across geomorphic settings, productivity regimes, and disturbance histories [6]. This scientific basis has been translated into greenhouse-gas accounting guidance that explicitly includes coastal wetlands and mangrove systems, enabling countries and projects to estimate emissions and removals using tiered approaches [7,8].

Despite their mitigation potential, mangroves remain exposed to persistent pressures, including conversion, hydrological disruption, pollution, and climate-related impacts that can accelerate erosion and mortality [9]. Global assessments indicate that human-driven losses have declined in some regions over the past two decades, but natural drivers and climate-linked stressors are becoming increasingly prominent, reinforcing the need for monitoring systems that can separate real change from classification noise and detect emerging vulnerabilities. Because mangroves sit at the land–sea interface, even modest land-use shifts can trigger large emissions when soil carbon is oxidized or exported [1,10].

At the same time, the policy landscape has evolved: blue-carbon interventions are now routinely discussed as nature-based solutions that can mobilize climate finance, but their credibility hinges on measurement, reporting, and verification (MRV) that is transparent, conservative, and repeatable [11]. Carbon markets, compliance and voluntary, have expanded the demand for defensible baselines, additionality arguments, and monitoring plans, while also increasing scrutiny of uncertainty and permanence. For coastal ecosystems, MRV must reconcile ecological complexity with auditability, making approaches that integrate accepted inventory guidance with spatially explicit monitoring especially attractive [12,13].

Remote sensing provides the practical backbone for that integration by enabling consistent land-cover and habitat accounting over large areas and multiple years, including in regions where field access is limited. Global products such as Copernicus CGLS-LC100 and ESA WorldCover offer standardized land-cover classifications with documented accuracy assessments, and they can be combined with local masks and protected-area boundaries to focus analyses on mangrove landscapes [14,15]. In parallel, global mangrove datasets and long-running satellite archives have demonstrated that repeatable change detection is feasible at policy-relevant scales, supporting both conservation planning and climate reporting [16,17].

Ecuador’s mangroves provide coastal protection and livelihoods in a highly dynamic estuarine setting, yet they have historically faced strong conversion pressure linked to coastal development and aquaculture [18,19]. The Manglares Churute Ecological Reserve, located within the Gulf of Guayaquil region, provides a policy-relevant setting to connect blue-carbon science with the operational requirements of climate finance. In this study, we quantify carbon-stock dynamics and derive a carbon-market valuation indicator for the Churute mangrove system over 2015–2021 using remote-sensing land-cover products and the accounting logic of the ECOSER ecosystem-services framework [20,21], explicitly positioning the outputs for MRV-aligned decision support.

2. Materials and Methods

2.1. Study Area and Spatial Framework

The analysis focuses on the Manglares Churute wetland system in coastal Ecuador (Guayas Province), internationally designated as a Ramsar site (Site No. 502; 35,042 ha; 02°28′ S, 79°42′ W). The site comprises mangrove forests along estuarine channels, adjacent scrub and marsh habitats, and lagoonal environments that support high biodiversity and multiple human uses, including low-intensity grazing, subsistence agriculture, recreation, and shrimp farming [22]. The spatial extent used for all calculations was defined by the national protected-area boundary for the Reserva Ecológica Manglares Churute, accessed through the official protected-areas (SNAP) GIS service attributed to the Ecuadorian environmental authority [23]. All geoprocessing and accounting were restricted to this boundary to maintain consistency across years and products (Figure 1).

2.2. Land-Cover Datasets and Class Harmonization (2015–2021)

Annual land-cover maps were assembled for 2015–2021 using two public, globally consistent products. For 2015–2019, we used the Copernicus Global Land Service dynamic land-cover product (LC100 yearly, Version 3), derived primarily from PROBA-V observations and ancillary thematic layers [24]. For 2020 and 2021, we used ESA WorldCover 10 m global land-cover maps (v100 for 2020; v200 for 2021), derived from Copernicus Sentinel data and produced by the ESA WorldCover consortium [15].

To ensure methodological coherence with ecosystem-services accounting and Tier 1 carbon parameters, land-cover classes were reclassified into the set of categories required for carbon-stock factor assignment: (i) crop, (ii) pasture and grassland, (iii) forestry, (iv) wetland, (v) floodable lowlands, and (vi) urban. The crosswalk was designed to preserve the dominant biophysical meaning of each source class while matching the land-use categories associated with Tier 1 default coefficients and stock-change factors [25,26].

In this harmonized reporting scheme, mangrove forests are included in the Forest/Forestry class (woody wetland vegetation). The Wetland and Floodable lowlands classes refer to non-woody wetland/open-water-associated areas used for Tier 1 factor assignment.

2.3. Pre-Processing, Alignment, and Area Accounting

All spatial layers were projected to a common coordinate reference system and clipped to the study boundary. Because the input products differ in native resolution (100 m for LC100; 10 m for WorldCover), rasters were aligned to a common working grid prior to map algebra and zonal summaries. Alignment included resampling, reprojection, and grid matching using nearest-neighbor resampling for categorical land cover, following standard GIS practice for thematic rasters [27,28]. Area by land-cover category was computed for each year using polygon-based zonal summaries over the reserve boundary and, where relevant, internal subunits; the workflow relies on zonal-statistics operations that aggregate raster values within vector-defined zones [29]. These annual areas (ha) constitute the activity data used in carbon-stock calculations.

2.4. Soil Organic Carbon Stock

Soil organic carbon (SOC) was estimated following ECOSER-style ecosystem-services accounting logic, using Tier 1 default reference SOC values and stock-change factors for tropical regions in the 2019 Refinement to the 2006 IPCC Guidelines [26]. Tier 1 was selected to ensure transparent, globally comparable estimates in the absence of site-specific soil measurements required for higher tiers, and this approach has precedent in closely related applications in Ecuador [20,30]. SOC estimates use the IPCC Tier 1 reference stock for the 0–30 cm mineral soil layer (topsoil) as provided in the 2019 Refinement.

SOC stock estimation was treated as an indicator for the climate-regulation ecosystem service [21]. The annual SOC stock (expressed as t C) was computed by multiplying the reference SOC stock by land-use and management factors and by the mapped area of each land-cover stratum, as follows:

$$\begin{array}{c}\hfill C_{\mathrm{SOC}}={\displaystyle \sum _{c,s,i}}SO{C}_{\mathrm{REF},c,s,i}·F_{\mathrm{LU},c,s,i}·F_{\mathrm{MG},c,s,i}·F_{\mathrm{I},c,s,i}·A_{c,s,i}\end{array}$$

(1)

where $C_{\mathrm{SOC}}$ is the soil organic carbon stock (t C) for the reporting domain, $SO{C}_{\mathrm{REF},c,s,i}$ is the reference SOC stock (t C ${\mathrm{ha}}^{-1}$), $F_{\mathrm{LU},c,s,i}$ is the stock-change factor for land use, $F_{\mathrm{MG},c,s,i}$ is the factor for management, $F_{\mathrm{I},c,s,i}$ is the factor for organic inputs, and $A_{c,s,i}$ is the area (ha) of each stratum. Default factors were applied by land-cover category using IPCC guidance for tropical regions (

Table 1. SOC correction factors by land-cover type (IPCC Tier 1 defaults).

Cover Type	${\mathit{F}}_{\mathbf{LU}}$	${\mathit{F}}_{\mathbf{MG}}$	${\mathit{F}}_{\mathit{I}}$
Crop	0.69	1.08	1.00
Pasture and Grassland	0.95	1.00	1.00
Forestry	1.00	1.00	1.00
Wetland	1.00	1.00	1.00
Floodable Lowlands	1.00	1.00	1.00
Urban	0.00	0.00	0.00

) [25,26].

The correction factors used are as follows:

SOC calculations were implemented through raster map algebra in QGIS, after standardizing resolution and alignment across layers. Pixel values in the resulting SOC maps represent tonnes of carbon per hectare and were subsequently aggregated to total tonnes of carbon using mapped areas [27,29].

2.5. Biomass Carbon Storage

Biomass organic carbon (BOC) was estimated by assigning Tier 1 biomass carbon coefficients to the harmonized land-cover categories and computing carbon storage spatially within the ECOSER accounting logic [25,26]. In our harmonized scheme, mangrove forests are included in the Forest/Forestry reporting class (woody wetland vegetation) and therefore receive the Forest Tier 1 biomass coefficients; the Wetland/Herbaceous wetland class refers to non-woody wetlands (e.g., marsh/herbaceous wetland) and open-water-associated wetland pixels.

Forest-related classes were merged to preserve a consistent Tier 1 coefficient mapping across LC100 and WorldCover; these global products do not reliably separate mangrove from inland forest within the reserve, so this ecological nuance is acknowledged as a limitation.

This component represents carbon stored in above-ground biomass pools associated with the dominant cover type in each pixel. Carbon storage values (t C/ha) were assigned by class and converted to pixel-level carbon content by multiplying by pixel area (ha), as follows:

$$\begin{array}{c}\hfill C_{B,i}=C_{B,\mathrm{ha},i}\times A_i\end{array}$$

(2)

where $C_{B,i}$ is the biomass carbon stored in pixel i (t C ${\mathrm{pixel}}^{-1}$), $C_{B,\mathrm{ha},i}$ is the biomass carbon density assigned to the land-cover type in pixel i (t C ${\mathrm{ha}}^{-1}$), and $A_i$ is the pixel area (ha).

The Tier 1 coefficients applied are as follows:

Note that “Wetland” and “Floodable lowland” in

Table 2. Biomass carbon storage factors by land-cover type (IPCC Tier 1 defaults).

Land-Cover Type	Carbon Storage (t C/ha)
Crop	5.67
Pasture and grassland	1.22
Forestry	6.34
Wetland (non-woody)	0.00
Floodable lowland (non-woody)	0.00

Note: The Urban class is omitted because it is zero/negligible within the reserve mask across all years. Values follow IPCC Tier 1 defaults applied via the land-cover crosswalk described in Section 2.2.

denote non-woody wetland/floodable areas; mangrove forests are represented in the “Forestry/Forest” class in this accounting framework.

BOC surfaces were produced via raster reclassification and map algebra. Where a pixel-based illustration is helpful for reproducibility, the conversion from per-hectare units to per-pixel carbon content is as follows:

$$\begin{array}{c}\hfill C_{B,\mathrm{pixel}}=C_{B,\mathrm{ha}}\times \left({\displaystyle \frac{A_{\mathrm{pixel}}}{10,000}}\right)\end{array}$$

(3)

with $A_{\mathrm{pixel}}$ in m² and 10,000 m² per hectare. All operations were carried out after ensuring consistent CRS and raster alignment [27,28].

2.5.1. Carbon Price Benchmark (Carbon Market Pricing) from the World Bank Carbon Pricing Dashboard

To monetize the annual carbon stock, we used an explicit carbon-price benchmark derived from the World Bank Carbon Pricing Dashboard, which compiles jurisdiction-level prices for implemented carbon pricing instruments (e.g., carbon taxes and emissions trading systems, ETS) and is reported in the State and Trends of Carbon Pricing framework [31,32]. For each year in the study period (2015–2021), we extracted the nominal price per tonne of CO₂ equivalent (USD tCO₂e⁻¹) for the set of instruments/jurisdictions.

Following a dashboard-style aggregation, the annual Carbon Market Pricing (CMP) benchmark was computed as the arithmetic mean of all non-zero price entries available for that year. Zero values were treated as non-applicable/non-priced cases for that year and were therefore excluded from the averaging, ensuring that CMP reflects the average of observed (priced) entries rather than being diluted by missing or not-yet-active instruments [33].

CMP is used solely as a global, transparent benchmark to contextualize magnitude; it is not intended to represent Ecuador’s carbon-market mechanism, nor tradable project revenue.

2.5.2. Economic Valuation of Stored Carbon

Annual total carbon stock was first expressed in tonnes of elemental carbon (tC). To obtain tonnes of carbon dioxide equivalent (tCO₂e), we applied the molecular-weight conversion [25,26]:

$${\mathrm{tCO}}_2\mathrm{e}=\mathrm{tC}\times {\displaystyle \frac{44}{12}}.$$

(4)

The annual economic value of stored carbon (VEC; USD) was then estimated as follows:

$$\mathrm{VEC}={\mathrm{tCO}}_2\mathrm{e}\times \mathrm{CMP}.$$

(5)

To improve interpretability independent of heterogeneous instruments, we also report a fixed-price scenario valuation using three illustrative prices (5, 10, and 20 USD tCO₂e⁻¹), computed as ${\mathrm{VEC}}_p={\mathrm{tCO}}_2\mathrm{e}\times p$.

This valuation represents an indicative market-consistent benchmark based on observed explicit carbon prices, and it is intended for comparative assessment across years rather than as a fully project-finance grade revenue estimate, because carbon prices differ across instruments in scope, coverage, and compliance design.

To avoid conflating real change with product-to-product differences, trend statistics were computed separately for periods supported by the same land-cover product: 2015–2019 (Copernicus LC100) and 2020–2021 (ESA WorldCover). For 2015–2019, we used a non-parametric monotonic-trend test (Mann–Kendall) with Theil–Sen slope as an effect-size descriptor, which is appropriate for short environmental time series and does not assume normality [34]. The 2019–2020 transition was treated as a methodological breakpoint rather than as evidence of an ecological regime shift.

Because LC100 does not provide a 2020 map, a same-year (overlapping) spatial comparison between LC100 and WorldCover is not possible. Instead, we implement a comparability sensitivity check by recomputing 2020–2021 activity data and carbon totals after aggregating WorldCover from 10 m to 100 m (majority resampling) and applying the same harmonized legend; differences between native (10 m) and aggregated (100 m) totals are reported as a resolution/harmonization sensitivity diagnostic rather than ecological change.

2.6. Uncertainty Characterization (Recommended for MRV-Aligned Reporting)

To support MRV-aligned interpretation, we report uncertainty bounds for annual total carbon (SOC+BOC) arising from (i) thematic land-cover uncertainty and (ii) Tier 1 coefficient uncertainty. Following IPCC good-practice principles for transparent uncertainty reporting, we used a Monte Carlo propagation approach in which annual totals were recomputed across repeated draws of land-cover and parameter values, while keeping the central (reported) estimates unchanged. For each year, the distribution of simulated totals was summarized using the 2.5th and 97.5th percentiles (95% interval) [35,36]. The same uncertainty draws were propagated through the CO₂e conversion and the valuation step so that valuation uncertainty reflects both carbon-stock and carbon-price variability. This uncertainty layer is used strictly as a robustness descriptor and does not replace the central accounting totals reported (

Table 3. Main sources of uncertainty/error across the MRV-style workflow (schematic summary).

Stage	Potential Error/Uncertainty Sources
Land-cover products	Thematic misclassification; sensor/algorithm differences; mixed pixels; resolution effects.
Pre-processing	CRS/reprojection; resampling artifacts for categorical classes; boundary-mask alignment.
Class crosswalk	Legend mismatches; aggregation of classes; limited separability of mangrove vs inland forest in global products.
Carbon parameters	Tier 1 representativeness; parameter uncertainty; assumed SOC reference layer (0–30 cm).
Valuation indicator	Instrument heterogeneity (tax/ETS); nominal price volatility; not interpretable as tradable Ecuador revenue.

).

2.7. Software

All GIS processing, raster algebra, and zonal summaries were conducted in QGIS Version 3.40 (QGIS Project), leveraging raster analysis and zonal-statistics functionality described in the official documentation [27,28,29,37].

Figure 2 summarizes the end-to-end MRV-style workflow from activity-data generation to carbon accounting, CO₂e conversion, valuation, and uncertainty reporting. Table 3 provides a compact schematic of the main error/uncertainty sources at each stage, emphasizing where product differences and Tier 1 assumptions may influence totals. These additions are intended as an interpretive aid and do not modify the core accounting equations or reported central estimates.

3. Results

3.1. Land-Cover Mapping Outputs and Area Accounting (2015–2021)

Land-cover mapping provides the activity data underpinning all subsequent carbon-stock and valuation calculations for the Churute mangrove ecological reserve. Two global products were used to cover the full 2015–2021 period: Copernicus Global Land Service LC100 yearly maps for 2015–2019 and ESA WorldCover maps for 2020–2021. The mapped outputs clipped to the protected-area boundary are shown in Figure 3 and Figure 4, and they illustrate both the spatial structure of the reserve and the product-to-product differences that emerge when the data source, sensor inputs, and classification system change.

Because the Copernicus LC100 and ESA WorldCover legends use different class identifiers and naming conventions, the original land-cover categories were first consolidated into a harmonized reporting scheme before area accounting and carbon estimation. This contraction step ensures that the final tabulated outputs represent like-for-like classes across the two datasets, allowing class areas (and the carbon coefficients tied to those classes) to be compared consistently within a single accounting framework. In practical terms, multiple forest-related categories present in the original products were merged into a single Forest reporting class, while the remaining land-cover types were aggregated into the set reported in

Table 4. Land-cover class areas (ha) within the protected-area boundary. Land-cover for 2015–2019 uses Copernicus LC100 yearly (100 m). Land-cover for 2020–2021 uses ESA WorldCover (10 m).

Year	Shrub	Forest	Water	Herbaceous Wetland	Cropland	Sparse Vegetation	Herbaceous Vegetation
2015	1003.4	33,439.2	10,450.9	2235.5	1710.3	3.6	1214.0
2016	789.1	33,210.4	10,529.6	3252.2	1557.8	1.2	716.5
2017	612.4	32,817.0	10,504.2	4408.1	1316.9	1.2	397.0
2018	577.3	32,756.5	10,599.8	4566.7	1250.3	1.2	305.0
2019	551.9	32,711.7	10,936.3	4376.7	1227.3	1.2	251.8
2020	48.3	36,111.7	10,376.4	1017.2	685.4	538.6	1290.1
2021	0.8	35,424.7	11,318.1	932.7	259.5	221.8	1912.7

Note: The Urban class is omitted because it is zero in 2015–2019 and remains negligible in 2020–2021 (14.6 ha and 12.1 ha, respectively).

.

Table 4 summarizes land-cover areas (ha) inside the reserve for each mapped year after harmonization (total mapped area ∼50,057 ha for 2015–2019 and ∼50,082 ha for 2020–2021 under the protected-area mask). Across 2015–2019 (LC100), Forest is consistently the dominant class, varying narrowly from 33,439.2 ha (2015) to 32,711.7 ha (2019), while Herbaceous wetland increases from 2235.5 ha (2015) to 4566.7 ha (2018) before a decrease to 4376.7 ha (2019). Over the same period, Shrub and Herbaceous vegetation both decline (1003.4 ha to 551.9 ha, and 1214.0 ha to 251.8 ha, respectively), whereas Water shows a net increase (10,450.9 ha to 10,936.3 ha). Sparse vegetation remains negligible throughout 2015–2019 (1.2–3.6 ha).

For 2020–2021 (WorldCover), the mapped pattern continues to reflect a forest-dominated reserve, with Forest totals of 36,111.7 ha (2020) and 35,424.7 ha (2021), and substantial mapped Water (10,376.4 ha in 2020; 11,318.1 ha in 2021). Other classes occupy comparatively smaller areas, including Herbaceous vegetation (1290.1 ha in 2020; 1912.7 ha in 2021), Herbaceous wetland (1017.2 ha in 2020; 932.7 ha in 2021), and Cropland (685.4 ha in 2020; 259.5 ha in 2021), while Urban remains negligible (≤14.6 ha). However, absolute areas in 2020–2021 are not directly comparable to 2015–2019 without caution: the two periods rely on different land-cover products and processing chains, and even within WorldCover, the 2020 and 2021 releases are associated with different algorithm versions. Accordingly, the 2020–2021 values are interpreted here primarily as internally consistent area accounting for the carbon calculations, rather than as definitive evidence of a structural shift relative to the 2015–2019 LC100-derived totals.

3.2. Total Carbon Storage: Spatial Patterns and Interannual Dynamics (2015–2021)

Figure 5 shows the spatial distribution of total carbon stock across the reserve for each year (2015–2021), together with the multi-year mean. High-carbon values consistently align with the forest-dominated portions of the protected area, while lower values occur in water bodies and other non-forest classes in the harmonized land-cover scheme. The maps provide a spatial check on the accounting outputs by linking (i) the land-cover activity data, (ii) the Tier 1 coefficients and SOC factors applied by class, and (iii) the annual totals reported for the reserve. Interannual departures are spatially structured (i.e., concentrated in particular zones where mapped class composition varies), rather than appearing as random pixel-level noise.

Annual totals of total carbon stock (t C) are summarized in

Table 5. Annual total carbon stock within the protected-area boundary (t C), dispersion metrics, and year-to-year change.

Year	Total Carbon (t C)	SD	Variance	$\mathbf{\Delta }$ (t C)	$\mathbf{\Delta }$ (%)
2015	1,881,335.75	19.50	380.10	–	–
2016	1,877,975.16	19.54	382.00	−3360.60	−0.18
2017	1,878,244.47	19.51	380.69	269.31	0.01
2018	1,874,261.18	19.58	383.25	−3983.29	−0.21
2019	1,859,892.86	19.81	392.36	−14,368.31	−0.77
2020	1,907,479.59	19.59	383.95	47,586.73	2.56
2021	1,865,235.78	20.22	408.91	−42,243.81	−2.21

. Across 2015–2021, the mean total carbon stock is 1,877,774.97 t C. Carbon stocks remain high and relatively stable during the internally consistent LC100 period (2015–2019; 1,859,892.86–1,881,335.75 t C), followed by a higher value in 2020 (1,907,479.59 t C) and a decrease in 2021 (1,865,235.78 t C). Year-to-year change is reported directly in the table as both an absolute difference ($\Delta$) and a percent difference ($\Delta \%$) relative to the preceding year, facilitating interpretation alongside land-cover area accounting (Table 4) and the change in land-cover product source between 2019 and 2020.

Within the internally consistent LC100 period (2015–2019), the Mann–Kendall test indicated a negative but non-significant monotonic trend in total carbon (Kendall’s $\tau$ = $-0.80$, two-sided p = $0.083$), and the associated Theil–Sen estimator yielded a slope of $-3671.94$ t C yr⁻¹. The negative sign reflects the net decrease observed across the LC100-based totals by 2019, but the magnitude remains small relative to the multi-year mean and is not statistically distinguishable from zero at the 5% level, given the short annual series. The 2019–2020 difference coincides with the transition from LC100 (100 m) to ESA WorldCover (10 m) and is therefore treated as a breakpoint in the activity-data source rather than as a standalone indicator of ecosystem change; separating these periods avoids overinterpreting differences that may be driven by the classification system and resolution effects instead of land-cover dynamics. Given $n=5$ (2015–2019), the Mann–Kendall/Theil–Sen results are reported as descriptive diagnostics with limited power, and we avoid any cross-product (2015–2021) trend inference.

3.3. Carbon-Market Valuation Indicator (CO₂e and Annual Value)

To contextualize the estimated carbon stocks in terms of climate-finance relevance, annual totals were converted to CO₂-equivalent using the molecular-weight ratio 44/12 (=3.67) and then multiplied by an annual carbon price indicator (CMP) compiled from publicly reported carbon-pricing instruments [38]. Following the Carbon Pricing Dashboard conventions, prices represent nominal values for the relevant year (reported as of April 1 or the latest available prior to April 1), and they are not strictly comparable across instruments because coverage and design differ. Accordingly, the resulting value is used here as a transparent indicator of magnitude rather than a project-level revenue forecast.

Table 6. Carbon-market valuation indicator by year, based on annual CO₂e totals, the applied carbon price (CMP), and the resulting value estimate (USD).

Year	CO2e (t)	CMP	Value Estimate (USD)
2015	6,898,231.1	21.5	148,311,968.7
2016	6,885,908.9	2.8	19,280,544.9
2017	6,886,896.4	6.4	44,076,137.0
2018	6,872,291.0	10.2	70,097,368.2
2019	6,819,607.2	8.2	55,920,779.0
2020	6,994,091.8	7.7	53,854,506.9
2021	6,839,197.9	8.6	58,817,101.9

reports the annual CO₂e totals, CMP, and the resulting value estimate. Over 2015–2021, the mean value estimate is $64,336,915.2, with a maximum in 2015 ($148,311,968.7) and a minimum in 2016 ($19,280,544.9). CO₂e totals vary within a narrow band (6.82–6.99 million t CO₂e), whereas CMP exhibits substantially larger relative variability (2.8–21.5), explaining why interannual valuation differences are driven primarily by the carbon-price assumption rather than by changes in the underlying carbon stock.

Using the annual CO₂e totals in Table 6, fixed-price scenarios yield values of USD 34.10–34.97 million (5 USD tCO₂e⁻¹), 68.20–69.94 million (10 USD tCO₂e⁻¹), and 136.39–139.88 million (20 USD tCO₂e⁻¹). Across all three scenarios, minima occur in 2019 and maxima in 2020, confirming that valuation scales linearly with price while CO₂e totals remain relatively stable.

The carbon price series used to derive CMP is reported in

Table 7. Carbon price series used in the valuation step (category values and annual mean as reported).

Category	2014	2015	2016	2017	2018	2019	2020	2021
Mexico carbon	0.0	0.0	2.8	2.6	2.9	2.9	2.3	2.9
Chile carbon	0.0	0.0	0.0	5.0	5.0	5.0	5.0	5.0
Colombia carbon	0.0	0.0	0.0	5.2	5.7	5.2	4.2	0.0
Mexico pilot ETS	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Argentina C	0.0	0.0	0.0	0.0	8.9	6.2	6.3	5.3
Spain carbon	27.6	21.5	0.0	0.0	24.8	16.9	16.4	17.6
Zacatecas C	0.0	0.0	0.0	12.6	13.7	13.0	11.8	12.2
Uruguay C	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Yucatan carbon	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Brasil carbon	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Baja California	0.0	0.0	0.0	0.0	0.0	0.0	0.0	8.3
MEDIA	27.6	21.5	2.8	6.4	10.2	8.2	7.7	8.6

Note: “MEDIA” is the arithmetic mean of the non-zero category values available for each year. A value of 0.0 indicates that a price was not reported (or not applicable) for that category-year and was excluded from the annual mean.

. This table is retained to document the underlying inputs and the annual aggregation used to compute CMP. Because not all instruments report prices in all years, CMP (“MEDIA”) was calculated as the arithmetic mean of the non-zero instrument values available for each year; entries of 0.0 indicate that a price was not reported (or not applicable) for that instrument in that year and were excluded from the annual mean.

4. Discussion

A first-order interpretation of the mapped interannual differences must explicitly account for the change in land-cover activity data from LC100 (100 m) to WorldCover (10 m) [15]. Differences in spatial resolution, class definitions, and production/validation pipelines can systematically alter mapped class areas, particularly for narrow linear features (channels), mixed pixels along ecotones, and small anthropogenic patches, thereby shifting the activity data that drive carbon totals even when the boundary mask is identical [39]. While Tier 1 parameters support transparent, conservative MRV-style accounting, they may underestimate mangrove biomass carbon relative to mangrove-specific literature values; we therefore interpret totals as conservative baselines.

The approach used here (global land-cover activity data + IPCC Tier 1 defaults) is best suited for transparent, conservative screening, protected-area baselines [40], and inventory-style reporting where reproducibility is prioritized over site-specific precision. Its applicability is limited to crediting-grade MRV (e.g., additionality, leakage, permanence, and disturbance attribution) and for detecting subtle interannual change when activity data are not homogeneous across time or when mangrove classes are not explicitly separated from other forests [41]. Finally, carbon-only MRV does not capture broader ecosystem conditions; complementary ecological indicators (e.g., sediment biodiversity baselines reported for Churute [42]) can help contextualize integrity beyond carbon stocks.

This is consistent with recent comparative assessments of high-resolution global land-cover products, which highlight inconsistencies arising from resolution changes, classification systems, and regional validation challenges when combining datasets for trend analysis [43,44]. Accordingly, we interpret 2015–2019 and 2020–2021 as separate, internally consistent accounting periods, and do not interpret differences across the 2019–2020 product transition as temporal trends.

Second, the magnitude and persistence of the total carbon stock estimated here are consistent with the broader evidence base that mangrove ecosystems are among the most carbon-rich forests globally, with large contributions from belowground pools and strong site-to-site variability driven by geomorphology, hydrology, and disturbance history [45,46]. Whole-ecosystem syntheses show that carbon density can be very high, and that even modest mapped shifts among forest/wetland classes can translate into meaningful differences in accounted stocks when Tier 1 coefficients are applied at scale [47,48].

Third, the stability of LC100-based totals across 2016–2019 aligns with the growing consensus that regional mangrove trajectories can be dominated by localized conversion and hydrological stress rather than monotonic, reserve-wide declines, and that the attribution of change requires attention to both human and natural drivers [10,16]. Global analyses using consistent, long-term datasets indicate that the rate and spatial pattern of mangrove loss depend on the balance between anthropogenic pressures (e.g., aquaculture expansion) and climate-related drivers, reinforcing the need to treat mapped differences cautiously when the underlying land-cover product changes [47,49].

Fourth, positioning the results for MRV-relevant use benefits from explicitly separating (i) the internally consistent period where trend statistics are meaningful (here, LC100 2015–2019) from (ii) cross-product transitions that can create artificial step changes in activity data [50]. This logic is consistent with recent work on coastal-wetland greenhouse-gas accounting that highlights both the feasibility of scalable, transparent estimation and the practical requirement to avoid overinterpretation when activity data are not methodologically homogeneous over time [51].

Finally, the valuation indicator illustrates that interannual variation in “value” can be driven more by the carbon-price assumption than by changes in the underlying carbon stock, particularly when using globally aggregated price benchmarks across heterogeneous instruments [52,53]. The broader carbon-pricing literature shows substantial cross-jurisdiction variability in price formation and drivers, implying that market-based valuation is best interpreted as an order-of-magnitude contextualization and should be accompanied by sensitivity/uncertainty framing rather than treated as a project-grade revenue estimate [38,54]. We emphasize that this valuation is not a tradable revenue estimate: eligibility under voluntary carbon markets (VCM) or REDD+/blue-carbon methodologies would require additionality, leakage assessment, permanence/crediting rules, baselines, and verified MRV consistent with the selected standard and national accounting. CMP is therefore presented only as a reference-value contextualization.

5. Conclusions

This study demonstrates that the Churute Mangrove Ecological Reserve maintains substantial and relatively stable total carbon stocks over the 2015–2021 period, averaging approximately 1.67 million t C, with values aligning closely with global syntheses confirming mangroves as among the most carbon-dense ecosystems worldwide. Within the consistent LC100-derived period (2015–2019), interannual variability was minimal and non-significant, supporting evidence that regional mangrove dynamics are often characterized by localized stressors rather than widespread decline.

The abrupt shift observed in 2020–2021 primarily reflects methodological discontinuities between the 100 m Copernicus LC100 and 10 m ESA WorldCover products, underscoring the importance of homogeneous activity data for robust trend detection in monitoring, reporting, and verification (MRV) frameworks. The carbon-market valuation indicator highlights the reserve’s potential climate-finance relevance, with annual equivalents ranging from roughly 5.6 to 6.4 million t CO₂e, and illustrative values fluctuating markedly due to carbon-price volatility rather than stock changes.

This scalable, transparent approach, combining remote-sensing-derived activity data with Tier 1 coefficients, offers a practical baseline for protected-area accounting and illustrates how even protected mangroves can contribute meaningfully to national blue-carbon inventories. Nevertheless, several limitations temper the interpretations presented here. Differences in resolution, class legends, and validation pipelines between land-cover products can introduce artificial step changes that confound ecological signals, particularly for narrow or fragmented features typical of mangrove landscapes. Reliance on generic IPCC Tier 1 emission factors overlooks site-specific variability in carbon densities driven by geomorphology, species composition, and disturbance history, while soil organic carbon estimates remain conservative by focusing on upper layers.

Finally, the valuation relies on aggregated nominal prices from heterogeneous instruments, providing only order-of-magnitude context rather than precise revenue projections, and does not account for transaction costs, additionality, or leakage risks inherent to market-based mechanisms. Future work integrating higher-tier, field-calibrated coefficients and consistent high-resolution time series would enhance accuracy for operational MRV and policy applications.