The Uaymil Flora and Fauna Protected Area, established on November 17, 1994, is located in the municipalities of Felipe Carrillo Puerto, Bacalar, and Othón P. Blanco in Quintana Roo, Mexico (18° 59′ 13′′ N, 87° 48′ 25′′ W) (Figure 1). This area covers 89,118 ha and is part of the Sian Ka’an Biosphere Reserve Complex.

Uaymil is comprised of 11 distinct ecosystem types, as identified by the Instituto Nacional de Estadística y Geología [INEGI] (2018) land use and cover map VII series (Figure 1a): this include ADV with 4.13 ha (Areas devoid of vegetation: surface where vegetation was removed by human activities); PT with 701.7 ha (wetland forests: island of sub-evergreen vegetation within mangrove, Halophytic vegetation or swamp area, between 8 and 12 m); SBQ with 8330.9 ha (sub-evergreen low thorn forests: sub-evergreen vegetation between 5 and 11 m, within lowland flood prone areas); SBQP with 35 517.5 ha (sub-evergreen low forest between 4 and 14 m, within lowland or plane surface with some degree of flooding); SMQ with 13 574.7 ha (sub-evergreen medium forests between 15 and 25 m, hence with a seasonal loss of 20 to 50 % of foliage); VM with 9545.1 ha (mangroves vegetation between 1 to 30 m, located on the margins of lagoons estuaries, mouths of rivers and streams); and VT with 20 444.8 ha (marshes: aquatic plant community mostly belonging to Typhaceae, Cyperaceae, Poaceae families within lagoons, freshwater/saltwater lakes with little depth, swampy areas, canals and backwaters of rivers).

Additionally, Uaymil includes secondary vegetation types such as VSA/SBQ with 597.5 ha (secondary tree vegetation of low sub-evergreen thorny forests: sub-evergreen low forest result of the regeneration process due to human activities); VSa/SBQ with 356.2 ha (secondary sub-evergreen low thorn forests with spiny shrubby vegetation: sub-evergreen low forest dominated by thorny shrub species result of the regeneration process due to human activities); VSa/SBQP with 25.6 ha (secondary sub-evergreen low forests with shrubby vegetation: sub-evergreen low forest dominated by shrub species result of the regeneration process due to human activities); and VSA/SMQ with 39.0 ha (secondary sub-evergreen low forests with arboreal vegetation: sub-evergreen medium forest result of the regeneration process due to human activities.

The elevation in Uaymil ranges from -19 meters in mangrove and marsh ecosystems to 20 meters in areas with sub-evergreen lowland thorn and medium forests, according to ALOS PALSAR DEM data (Figure 1b). These ecosystems are crucial for conservation, forming a mosaic that protects the upper basin of Espiritu Santo Bay. Additionally, Uaymil serves as a vital habitat for species such as jaguars, ocelots, pumas, white-tailed deers, and anteaters (Secretariat of Environment and Natural Resources [SEMARNAT], 2016).

This study uses three different MODIS (Moderate Resolution Imaging Spectroradiometer) satellite products (Wang et al., 2022): MOD13Q1 that contains the Normalized Difference Vegetation Index (NDVI), with a 16-day 250 m spatial resolution (Didan and Barreto-Muñoz, 2019); MOD15A2H that contains values of the Leaf Area Index (LAI), with an 8-day 500 m spatial resolution (Myneni et al., 2015); MOD17A2 that contains data on Gross Primary Productivity (GPP), with an 8-day 500 m spatial resolution (Running and Zhao, 2019).

The MODIS products were acquired from the NASA LAADS DAAC (http://ladsweb.modaps.eosdis.nasa.gov/search/order) for the years 2000, 2005, 2011, 2015, and 2023. A comprehensive examination of satellite images was conducted to ensure that the best images were used for an annual analysis. Images with distorted pixels within the study area were discarded, resulting in the retention of 19 MODIS images of each product, totalling 57 MODIS images, with path and row 09/07. Each year included approximately four images, except for 2023 where only three images were obtained (Table 1).

MODIS Terra data utilized in this study was first converted from MODIS HDF data to GeoTIFF format. Subsequently, the satellite image product was reprojected to WGS 1984 UTM Zone 16N. Due to differences in resolution from the MODIS products (LAI and GPP: 500 m; NDVI: 250 m), resampling (nearest-neighbour method) was done to 250 m resolution for a better integration of images, to maintain consistency and for a finer resolution before conducting the EQI composite analysis (Santhosh-Baboo and Renuka-Devi, 2010; Bartmiński and Siłuch, 2020). The next step involved applying multiple scale factor for radiometric calibration (scale factor LAI: 0.1; NDVI: 0.0001; GPP: 0.0001) (Didan and Barreto-Muñoz, 2019; Running and Zhao, 2019). After image processing, the annual mean of NDVI, LAI, and GPP were obtained. Lastly, the study area was extracted using the masking tool of ArcMap 10.8.

The Ecological Quality Index (EQI) developed by the Chinese Ministry of Ecology and Environment (MEE) utilizes three primary ecosystem spectral indices: Leaf Area Index (LAI), Gross Primary Production (GPP), and Fractional Vegetation Cover (FVC) (Ren et al., 2022; Wang et al., 2022). LAI and GPP values are obtained from MODIS products, while FVC is calculated using MOD13Q1 NDVI data, with the dense vegetation mosaic pixel model by Gutman and Ignatov (1998) (1).

Where NDVI is the Normalized Vegetation Index; NDVI_soil is the NDVI value of pure bare soil; NDVI_veg is the maximum NDVI value.

For calculating the annual EQI, the initial step involved to determine the Relative Vegetation Index (RVI) of ecological indicators by ecosystem type, using the Instituto Nacional de Estadistica y Geologia [INEGI] (2018) land cover maps, and to identify the vegetation types within the PA. LAI, GPP, and FVC values were extracted for each vegetation type using masking techniques, and RVI was calculated using the Chinese Ministry of Environment [MEE] (2021) formula (2):

Where RVI_{i, j, k} is the relative density of ecological parameters for vegetation type k in Zone j in year i; F_{i, j, k} is the value of the ecological parameter; F max_{i, j, k} is the maximum value of the ecological parameter.

These RVI values were then merged, and all indices were normalized using Chinese Ministry of Environment [MEE] (2021) formula (3):

Where X is the normalized index; x is the original index; min/max are the minimum and maximum values.

Finally, the EQI was calculated for the analysed years using the Chinese Ministry of Environment [MEE] (2021) formula (4):

To assess temporal changes in the quality of Uaymil ecosystems, EQI values were categorized into five classes: very low, low, moderate, high, and very high (Chinese Ministry of Environment [MEE] (2021). A very high EQI indicates natural vegetation in a favourable condition, while a very low EQI suggests an unfavourable condition. The EQI for each of the five years was classified using the mean natural break method, a common GIS technique that identifies natural groupings within data and provides a suitable representation of real-world conditions (Brewer and Pickle, 2002; Chen et al., 2013). The resulting EQI categories were very low (0-0.26), low (0.26-0.49), moderate (0.49-0.75), high (0.75-0.89), and very high (0.89-1.0).

Before calculating the annual EQI, the ecological indicators (LAI, GPP, and FVC) were normalized (Formula 3) to ensure they were on a comparable scale, allowing for equal contribution to the EQI calculation (MEE, 2021). Although all indicators equally influence the EQI, their complex interrelationships may affect EQI changes differently. To explore these dynamics, we performed a Pearson correlation analysis, among indicators and EQI. Additionally, to evaluates how each indicator integrates with the EQI, with the EQI as the dependent variable, we conducted multiple and stepwise linear regression analyses. (Zhang et al., 2022; Zhao et al., 2023).

The spectral ecological indices (LAI, GPP, and FVC) demonstrate a strong positive correlation with EQI, with Pearson correlation coefficients ranging from 0.75 to 0.97 and a significance level of p = 0.001 (Figure 2). Furthermore, the combination of these spectral ecological indicators significantly enhances the explanatory power of EQI (GPP: R² = 0.95, GPP + LAI: R² = 0.98, GPP + LAI + FVC: R² = 0.99). This relationship has also been reported by Zhang et al. (2022) in mountainous areas of China, where LAI, GPP, and FVC showed strong positive correlations. Similarly, Wang et al. (2023) reported a positive correlation between NDVI and EQI, with higher EQI located within dense vegetation. This relationship illustrates a coupled response and recurring pattern in the ecosystem’s dynamics, where enhancements or declines in one indicator are systematically reflected by changes in the others.

The temporal patterns of ecological indicators reveal clear fluctuations over the study period, which are reflected in their spatial distributions within the Uaymil PA (Table 3, Figures 3 and 4). There were notable changes in mean FVC over time. From 2000 to 2005, the FVC mean decreased by 0.18, indicating a reduction in vegetation coverage. This decline was followed by an increase of 0.17 from 2005 to 2011. A subsequent decrease of 0.15 was observed from 2011 to 2015, with a recovery of 0.13 from 2015 to 2023. These fluctuations are mirrored in the spatial distribution maps (Figures 3 and 4), where high FVC values were predominantly located in forest ecosystems. However, between 2005 and 2015, a shift towards lower and moderate FVC categories, particularly in marsh and mangrove areas.

The GPP mean exhibited minimal fluctuations compared to FVC. There was a slight decline of 0.01 from 2000 to 2005, stability at 0.62 from 2005 to 2011, a notable decrease of 0.05 from 2011 to 2015, and an increase of 0.08 from 2015 to 2023. This trend was reflected in the spatial distribution, with marsh and mangrove ecosystems showing consistently low GPP, while forest ecosystems exhibited higher GPP values, except for a decrease in very high GPP categories in 2015, which dropped to high and low ranges in medium to lowland sub-evergreen forests.

LAI demonstrated a significant increase from 0.49 in 2000 to 0.62 in 2023, with periodic increases and decreases. LAI showed consistent high values in forest ecosystems, while marsh and mangrove areas retained lower LAI categories. The spatial distribution maps revealed an increase in high LAI coverage from 2005 to 2023, indicating a broader expansion of foliage in forested regions. EQI showed reductions in 2005 and 2015, with a mean change rate of 0.04, and an improvement from 2011 to 2023 with an increase of 0.06. EQI increased from 0.64 in 2000 to 0.68 in 2023. Spatially, the distribution of EQI categories highlighted low EQI coverage in mangrove, marsh, devoid of vegetation, wetland forests, and secondary sub-evergreen lowland forests. In contrast, other forest ecosystems showed an increase in high to very high EQI categories, with a notable rise in very high EQI by 2023.

Moreover, a clear pattern was observed where higher EQI categories corresponded to greater vegetation cover, leaf cover, and GPP, while mangrove and marsh ecosystems were often associated with low EQI and ecological indicator categories. In contrast, forest ecosystems typically demonstrated high EQI and ecological indicator values. These findings are consistent with research conducted by Wei et al. (2023), which noted that regions such as Australia, Africa, and Asia generally exhibit low EQI due to low vegetation coverage and productivity, particularly in wetlands, deserts, and shrublands. The consistent observation of low EQI and ecological indicator categories in mangrove ecosystems across different studies suggests that low EQI does not necessarily indicate poor conservation. Mangrove ecosystems frequently exhibit low canopy density, vegetation coverage, and water presence, which can lead to variability in the results of spectral indices and potentially result in misinterpretations (Tran et al., 2022). According to Moore et al. (2015), these ecosystems also tend to exhibit low productivity due to above mentioned factors. However, despite these factors, mangrove ecosystems can still be ecologically valuable and play crucial roles in coastal ecosystems.

Between 2000 and 2005, the low and high EQI categories dominated the study area, accounting for almost 70-80% of the total surface area. In 2000, the high EQI category, encompassing both high and very high EQI categories, covered 61% (53 011.62 ha of 89,118 ha) of the Uaymil Pa surface area. However, in 2005 there was an increase from 7% to 8% (6425.0 to 7118.6 ha) in surface covered with very low and high EQI category, while for moderate, low and very high EQI categories decreased, respectively. By 2011, there was a slight increase in surface covered by EQI categories: 2525.29 hectares in low; 460.8 hectares in moderate, and 2354.2 hectares in very high. In 2015, the area in Uaymil classified under very low and moderate EQI categories increased from 3% to 12% (3004 to 6761.6 ha). Conversely, the surface area of low, high, and very high EQI categories decreased by 3%, 2%, and 5%, respectively. In 2023, there was a significant cover of surface in very high EQI category with approximately 52% (45 450.1 to 89 118 ha) of the PA. Overall, during the entire study period, low EQI categories covered 35% (31 191.3 ha) of the PA, while optimal EQI categories (high and very high) had the greatest coverage, accounting for 63% (596 144.34 ha) of the PA (Table 4).

Most ecosystems in the Uaymil PA show high ecosystem quality index. However, 35% of ecosystems within the PA showed low EQI. Throughout the study period, mangrove and marsh ecosystems consistently fell within the very low to low EQI category, covering about 76.2% of the area classified as low EQI. This equates to 23 777.1 ha out of a total of 31 191.3 ha. The forest vegetation exhibited a substantial coverage of optimal EQI, with most of the forest surface (63%) classified as having excellent ecosystem quality. In contrast, the low EQI categories had a minimal coverage, accounting for 7414.1 ha, which represents 23.7% of the total area reported as having low EQI (Figure 5). Wetland forests are patches of vegetation within wetland ecosystems; these ecosystems showed 50% of poor coverage and 50% of excellent quality ecosystem index. On the other hand, areas devoid of vegetation, as well as secondary subevergreen low forests characterized by spiny shrubby vegetation, secondary subevergreen low arboreal vegetation, and subevergreen low thorn forests, consistently showed a higher coverage in the low and very low EQI categories throughout the year. Despite some areas showing high EQI, approximately 7% of the coverage was of poor quality (Figure 5), indicating potential impacts and suggesting the need for a restoration plan.

Notably, the coverage of very low EQI categories increased during certain years, particularly in 2005, 2011, and 2015. While this study did not focus on analyzing the factors contributing to these changes, several reports and studies have identified potential causes. For example, the Comisión Nacional para el Conocimiento y Uso de la Biodiversidad [CONABIO] (2024) from the Forest Fire Early Warning System (2000-2021) recorded fires in 2003, 2007, 2011, 2013, and 2016 within Uaymil. Velasco-Tapia (2009) identified fires as a major cause of deforestation and forest transformation within the Sian Ka’an-Uaymil Biosphere Reserve. Hurricanes have also been significant drivers of change. For instance, Whigham et al. (2003) and Pat-Aké et al. (2018) reported that hurricanes caused significant reductions in canopy cover and basal area, increased tree mortality and tree damage, with effects lasting up to five years. Between 2007 and 2017, six hurricanes impacted Mahahual, Felipe Carrillo Puerto, Bacalar, and Quintana Roo municipalities where Uaymil protected area is located (INEGI, 2016). Studies by Zheng et al. (2023), Ren et al. (2022), and Zhang et al. (2023) indicated that high elevation, temperature, and precipitation adversely affect EQI. In Quintana Roo, climatic variables have been on the rise since 1986 (De la Barreda et al., 2020). These factors, along with others mentioned, could be primary drivers contributing to the changes and patterns observed in Uaymil as reported by this study.

Despite these challenges and the presence of deforestation and degradation drivers, Uaymil’s ecosystems have shown resilience. Increases in the mean values of each ecological indicator, maintenance of high vegetation coverage, productivity, leaf cover, and an increase in the surface area of high to very high EQI categories demonstrate the ecological quality and resilience of the area. This resilience is essential for biodiversity conservation and the provision of ecosystem services.

EQI has highlighted the overall high resilience, and the current state of the vegetation types, by identifying areas for further analysis and potential restoration activities. The EQI allows for monitoring and identifying areas that have experienced vegetation loss and those with no vegetation cover, which require further analysis. In addition, the temporal analysis of changes in EQI could help us determine influencing factors that contribute to the degradation of ecosystem quality. For example, Zhang et al. 2023 used EQI to evaluate degradation within forests, grasslands, wetlands, and farmlands. They identified that grasslands, farmlands, and wetlands presented low quality and a reduction in EQI over the last 20 years, due to an increase in economic growth and an over exploitation of fishery resources. This resulted in increased pollution, habitat destruction, and overuse of natural resources. In contrast, forest ecosystems presented the opposite trend, with forest management and returning farmland to forest improving quality. Other studies also emphasize the potential of EQI for the evaluation of ecosystems. Wang et al. (2022) and Zheng et al. (2023) noted that EQI, when combined with other spectral vegetation indicators through a pressure-state-response analysis, is a fundamental approach for evaluating ecosystems and identifying degradation factors. In addition, it aids in ecological zoning of key conservation areas, and potential restoration zones.

Nevertheless, EQI has several limitations. For instance, vegetation indices may not adequately account for the quality and productivity of diverse ecosystems such as mangroves, savannas, and wetlands. This is related to poor vegetation coverage, low LAI characteristics, and the presence of water, which can lead to underestimations and misinterpretations due to the variability in conditions in these ecosystems influenced by climatic, soil, and water levels (Gao, 1996; Tran et al., 2022). Secondly, ecological indicators (LAI, GPP, and FVC) may exhibit distinct variations in response to distinct vegetation type, and climatic, seasonal or degradation factors. For example, Zhang et al. (2023) found 33.2% to 35.0% mismatch phenological difference between indicators, which means each ecological indicator responds to climatic or environmental factors differently. Additionally, they reported a 10-day difference in identifying phenological changes using these indicators. This indicates that each indicator may detect seasonal changes at different times, showing changes up to 10 days earlier or later than others. Thirdly, using low spectral resolution images excludes finer details and small patches of degradation areas within these ecosystems (Xu et al., 2021). For instance, Figure 6 shows image D4, which represents patches of degraded areas where construction material extraction occurred, however these areas were classified as having a moderate EQI. This output could be due to the low resolution of the used images, and the weighted reflectance average of the small degraded patch and its surrounding vegetation.

Other limitations would be obtaining high-quality images, especially in the tropics where there is a presence of high cloud cover throughout the year. Additionally, the categorization of each EQI level has not been standardized to reflect the conditions observed in the field, which may affect the assessment accuracy. However, these limitations could be reduced by considering the characteristics of each ecosystem type analysed and their response to different environmental, social, and economic influencing factors. Furthermore, using high resolution satellite images for analysing ecosystem quality, integrated with field evaluation or utilizing UVA (unmanned aerial vehicles), could be an excellent tool used to accurately determine traits of degradation within ecosystem types.

Future research directions should be aim at developing or improving EQI tailored to distinct ecosystems like mangroves, savannas, and wetlands. In addition, establishing standardized frameworks for EQI categorization through field validation and local calibration will ensure precision and relevance.