Tropical cyclone effects on vegetation resilience in the Yucatan Peninsula , México , between 2000-2012

The resilience capacity of vegetation in the Yucatan Peninsula is influenced by the winds and rains of tropical cyclones. There are no recent long-term studies on cyclonic impacts on natural vegetation in the region despite their significant effects on infrastructure and biodiversity. The objective of this study was to identify the area impacted by 21 tropical cyclones between 2000 and 2012 and to quantify the recovery capacity of the vegetation by using standardized anomalies of the normalized vegetation index (aNDVI). MODIS images from NASA’s “Terra and Aqua” satellites were used to calculate the damaged areas by analyzing the frequency of pixels corresponding to each type of vegetation per impact zone. The results showed that in 67% of the tropical cyclones, the impacts on vegetation were negative —a decrease in aNDVI—but in 33% of the cyclones, positive effects were recorded —an increase in aNDVI—. The lapse rate of vegetation recovery varied in 52% of the cases; vegetation recovered between two and three weeks after each cyclonic event, while 38% of the cases recovered within four to five weeks of the cyclone landfall. Tropical forests suffered the most significant effects, followed by hydrophilic vegetation. The most destructive hurricanes were Emily, Wilma, and Dean. The rate of recovery laps ranged from 4 to 10 weeks after the hurricane hit. The results could improve assessments of vegetation vulnerability against severe hydrometeorological events and establish priority zones for prompt inspection.


Introduction
Resilience and recovery are two crucial concepts in the ecological service literature. Holling (1996) defines ecological resilience as the magnitude of the disturbance that a system can absorb before changing its structure by modifying the variables and processes that control its behavior. Oliver et al. (2016) argued that the concept of ecological resilience recognized the existence of multiple stable states and their ability to resist "regime changes" between alternate states. Tropical cyclones are considered regime changes in ecosystems. Vegetation recovery after the strike of a hurricane is essential to maintaining ecological protection services. To understand resistance and resilience, it is necessary to perform studies from different spatiotemporal and dimensional scales to grasp the significance of the processes that lead to an ecosystem suffering the impairment of its dynamics and functional performance (Ghazoul & Chazdon, 2017).
Mexico receives direct impacts from tropical cyclones because it is located between two cyclogenetic zones: the Northeast Pacific and the Northern Atlantic (Rosengaus-Moshinsky, 2010). The National Oceanic and Atmospheric Administration (NOAA) estimated an annual average of 12.1 tropical storms and 6.4 hurricanes in the Atlantic basin and 16.6 tropical storms and 8.9 hurricanes in the northeast and central Pacific (NOAA, 2014). During the second half of the 20th century, the most intense hurricanes, categories 4 or 5 according to the Saffir-Simpson scale, occurred in the Yucatan Peninsula region, mainly in Quintana Roo state, since it is on the track of Atlantic cyclonic systems (Ihl & Frausto-Martínez, 2014).
High-intensity winds and heavy rains associated with cyclones can cause disasters in regions with human settlements, such as damage to infrastructure, cultivation areas, and human losses and injuries (Rosengaus-Moshinsky, Jiménez-Espinosa, & Vázquez-Conde, 2002). In addition, these extreme weather systems also affect the natural environment by altering biodiversity patterns and ecosystem services (Solow, 2017;Van de Pol, Jenouvrier, Cornelissen, & Visser, 2017). In contrast, hurricanes can create favorable conditions for the regeneration of certain species capable of withstanding hurricane-force winds (Snook, 1993;Vink & Ahsan, 2018).
Ecosystem services depend directly on the conservation status of the vegetation cover. In the Yucatan Peninsula case, vegetation is mainly represented by tropical rainforests; medium and low deciduous forests are located predominantly in Yucatan state, while tall and medium subevergreen primarily occur in Campeche and Quintana Roo states (Sánchez Aguilar & Rebollar Domínguez, 2016). However, this type of forest suffers from deforestation and illegal logging (Rebollar, De la Paz-Pérez Olvera, & Quintanar, 1993) and land-use change to urban areas or cattle (Rosengaus-Moshinsky, 2010). Additionally, vulnerability to tropical cyclones is increased due to low relief, which does not decrease wind flow (Boose, Foster, & Hall, 2003).
The multitemporal analysis of the impacts of cyclones on vegetation is relevant. According to Buma and Wessman (2011), much of the knowledge about vegetation resilience comes from the investigation of disturbances caused by singular events. Various studies on the Pacific coast of Mexico consider tropical cyclones to be one of the prime causes of disturbances in ecosystems (Bhaskar et al., 2018;Tapia-Palacios et al., 2018). Some studies, such as those of Parker, Martínez-Yrízar, Álvarez-Yépiz, Maass, & Araiza (2018), quantified the change in vegetation by applying the normalized vegetation index (NDVI) and LIDAR images (laser imaging detection and ranging) after the passage of hurricanes Jova (2011) and Patricia (2016). Examples of large-scale studies applying sampling techniques are those of Jimenez- Rodríguez et al. (2018), who identified the relationship between the structure and composition of the forest, recovery capacity, and damage magnitude.
Another example is provided Martínez-Yrízar et al. (2018), who compared two hurricanes of distinct categories to estimate the Chamela dry forest litter production by analyzing the ecosystem' s short-term Tropical cyclone effects on vegetation resilience in the Yucatan Peninsula, México, between 2000-2012 response. However, most of the research has focused on the evaluation a single cyclone' s effects on vegetation or the comparison of just a few cyclonic events. Long-term analysis of hurricane impact on vegetation in the Yucatan Peninsula is still scarce.
To understand the dynamics of the ecosystem recovery processes, it is necessary to have data before and after each extreme event. However, the field samples and measurements may be restricted due to damage and restricted access to sites (Holm et al., 2017). Therefore, indirect methods, such as satellite-derived images, present an alternative to estimating the impacts' spatiotemporal scope. Remote sensing techniques with products designed to record different electromagnetic spectrum ranges have allowed the algebraic combination of spectral bands to detect vegetation cover variations (called "green indexes"). Yengoh et al. (2016) argued that the NVDI is the most widely used technique in detecting spatiotemporal vegetation changes. As Ghazoul & Chazdon (2017) indicated, the implementation of the NDVI on a smaller scale allows the identification of the landscape' s degradation and recovery through indirect measures of treetop or forest cover densities.
The objectives of this work are a) to estimate the impact area of 21 tropical cyclones that hit the Yucatan Peninsula between 2000-2012; b) to estimate the vegetation recovery capacity after the landfall of each cyclone through the NDVI anomaly value; and c) to identify whether the effects of cyclones on ecosystems are related to their trajectory, their category, or their landfall.

Study area
This study considers the Yucatan Peninsula physiographic province (INEGI, National Institute of Geography and Statistics, 2001). The Yucatan Peninsula is located between 17.8° and 21.6° north latitude and 86.7° and 92.4° west longitude ( Figure 1) and has an extension of 126,547.38 km 2 and a vegetation cover of 98%. It is characterized by a large karst platform and low altitude (0-300 m). The region largely consists of two climates: a warm wet climate with an average temperature between 24 and 26°C distributed from north to south, and a very warm climate with temperatures between 26 and 28°C toward the state of Campeche. The hurricane season occurs from June to November. May to October is the rainy season, with the highest rainfall (> 200 mm) in September due to the presence of tropical cyclones.

Data
Data and supplies for the research development came from the following free remote sensing databases and cartographic repositories: a) Cyclone trajectory from the library of the Best Track Archive for Climate Stewardship (IBTrACS, 2016) version "v03r09" by NOAA for the period 2000-2012, b) Vegetation types from the National Continuum of Land Use and Vegetation (LUV), scale 1: 250,000 series IV (INEGI, 2009), c) NDVI time series from the data library of the Research Institute for Climate and Society (IRI, 2017) of the USGS LandDAAC MODIS version_005 Southern North America series, with a temporal resolution of "16-day composites", at a spatial resolution of 250 m/pixel (Huete et al., 2002). Both cartographic products and satellite images were processed using map algebra functions integrated with the geographic information system (GIS) ArcGIS 10.3.
An intersection geoprocess was used on a map overlapping the vector layers of the cyclone tracks formed into the North Atlantic basin and the peninsula polygon to identify the 21 cyclones that made landfall on the Yucatan Peninsula between 2000 and 2012. To simplify vegetation types, we performed cartographic reclassification from LUV to eliminate the polygons corresponding to the following categories: human settlements, bodies of water, other vegetation, and without vegetation. The remaining polygons were merged to create a new classification integrated only by tropical forest, grassland, farming, and hydrophilic vegetation.
The condition of vegetation is characterized at any given time by the historical mean of the period, in such a way that positive values represent an increase in photosynthetic activity -surplus-, while negative values are stated lower than expected by the mean -deficit-, where the "normal condition" is given by amounts between -1 and +1. The NDVI was calculated as the normalized difference of two spectral bands -red and near-infrared-, whose standardized variation was within the range of -1 to +1; the mathematical function for the computation is shown in Equation

Methods
The study had two analysis stages: a) estimation of the cyclone impact area and b) estimation of the vegetation resilience after every cyclone.

Estimation of the cyclone impact area
To distinguish between seasonal phenological changes in vegetation and those associated with a tropical cyclone passage, we defined three "neighborhoods (Nh)" (buffer areas) for each cyclone track: Nh-100 km, Nh-200 km, and Nh-300 km. For the neighborhoods for each cyclonic trajectory, we considered the impact zone (strike zone), described by the National Hurricane Center (NHC, 2019). According to the track, the strike zone represents the typical extension of the hurricane winds category, centered on the hurricane´s eye ( Figure 2).

NDVI anomalies (aNDVI)
To determine the annual variation of vegetation phenology in the NDVI values from the sudden changes associated with a cyclonic path, we calculated the NDVI anomalies (aNDVI), where aNDVI is the difference between a sample value of a 16-day NDVI composite and the average NDVI value for the study period. The concept of climatic anomalies proposed by Wilks (2011) was adapted to calculate the aNDVI. The aNDVI is used widely for different purposes in agriculture, pests, forest management, water deficit, and drought assessment, allowing measuring changes in terms of the photosynthetic activity of vegetation -greenery-in its interannual shift (Aboud, Bias, Brites, & Santos, 2018;Meroni, Fasbender, Rembold, Atzberger, & Klisch, 2019;Nanzad et al., 2019;Zewdie & Csaplovics, 2015). The average and extreme values of aNDVI were quantified under the assumption that, due to the seasonal variation in events, anomalies provide more information about the magnitude of data series values by removing the influences of dispersion (Wilks, 2011 To calculate the aNDVI, we applied Equation 2 at the pixel level for each annual set of images. The operation was performed using the ArcGIS "Cell Statistics" function. The NDVI and aNDVI were plotted to compare the landfalling cyclones with the time series to highlight observed phenological variations associated with cyclonic events, given that the interannual variation in NDVI is an indicator of vegetation phenology (Gómez-Mendoza, 2007). Sánchez-Rivera, G. and Gómez-Mendoza, L. Investigaciones Geográficas, in press.

Resilience estimation
The estimation of ecological resilience capacity (Holling, 1996) was obtained by calculating the recovery rates of photosynthetic activity measured as a function of the aNDVI in periods of 4 weeks after each tropical cyclone had landfall (Rouse, Haas, Schell, & Deering, 1973). The damaged area for each composite was calculated through the frequency of pixels concerning each neighborhood's vegetation type. We defined the impacts caused by each cyclone in five stages: a. For each cyclone, six composites were selected: one prior -16 days before-, one during the event, and four after landfalling -84 days later-. b. We recalculated the aNDVI values for each of the three cyclone-neighborhood combinations for each of the six composites selected per event. This operation was replicated for every one of the 21 cyclones. Subsequently, the results of each composite were subclassified by each vegetation type. c. The areas damaged by cyclones were estimated in percentages divided into two classes: moderate and extreme deficits. d. The estimation of the aNDVI per vegetation type was processed from the earlier step' s images using the ArcGIS "Zonal Histogram" function. e. The weekly recovery rates were obtained, calculating the differences in aNDVI values between the previous composite concerning cyclone landfall and the subsequent composites.
Map algebra processing and operations were performed with Python programming language at a pixel resolution.
The impacts of each cyclone were classified using the IRI, UNESCO, FAO, and Ministry of Agriculture of Chile & Center Water for Arid Lands (2021) scale of interpretation of aNDVI values. This scale classifies the data dispersion around the mean value, measured in terms of the standard deviation, where values between -1 and 1 represent normal conditions, values greater than 2 represent moderate and extreme surplus, and less than -2 represent moderate deficit and extreme (Table 1).

Tropical cyclones
Thirteen of the twenty-one cyclones studied in the period 2000-2012 reached the Saffir-Simpson hurricane category, and only five of them maintained that category when landing: Isidore (2002) (Table 2).

Vegetation type classification
Tropical forests are the predominant vegetation type in the entire region, covering more than three-quarters of the peninsula (78%), followed by 9.6% grasslands, concentrated mainly in northern Yucatan State and some small areas in Campeche and southern Quintana Roo (Figure 3). Farming (cropland) land use accounts for 5% of land cover, mainly near Yucatan and Campeche states and small areas in Quintana Roo' s southern zone. Finally, hydrophilic vegetation covers 5.2% of the region, where it is distributed throughout the peninsula' s coastal zone, from south of Quintana Roo in the Caribbean Sea, through northern Yucatan to south of Campeche.

Impact zone
The aNDVI values were statistically diluted, given the territorial extension of the areas directly impacted against the peninsula' s surface not damaged by the winds and rains associated with the cyclones. Figure 4 shows an example of the NDVI and aNDVI values for hurricane Wilma (2005) during its track through the peninsula. Figure 5 shows the hurricane Wilma (2005) time series, wherein for the first composite  before the presence of the hurricane, the aNDVI values were in the range of -1 to +1 -considered as a condition of normality-. The second image , depicts the cyclone' s landfall, during which the aNDVI increased dramatically toward the light and moderate deficit classes but only in a small area adjacent to its trajectory. It was approximately the fifth and sixth weeks (20-2005), after the cyclone had dissipated, that the most significant adverse effects on vegetation occurred with increases to a negative aNDVI, mainly toward the extreme deficit class. Subsequently, the anomaly decreased remarkably close to the values before the cyclone track. Signs of recovery appeared between weeks 7 and 8 (23-2005). However, six weeks after the hurricane made landfall , there was a slight uptick in the aNDVI values toward the moderate and extreme deficit classes on the south-southwest portion of the peninsula not directly associated with the passage of the cyclone. When calculated on a peninsular scale, such differences prevent the ability to accurately quantify the recovery rates of vegetation that suffered direct damage from cyclonic impacts.
In cases of very intense hurricanes such as Emily (2005, H4), Wilma (2005, H4), and Dean (2007, H5), the values did not decrease below 0.5 units ( Figure 6). Furthermore, the existence of ascending and descending peaks in some aNDVI that did not correspond to the arrival of any cyclone could be the result of other types of extreme disturbances, such as drought, forest fires, El Niño-Southern Oscillation (ENSO), or land-use changes. Such is the case for composites 2-2005, 6-2009, and 8-2011. However, the explanation of such phenomena is outside the scope of this study. The effects of cyclonic tracks on vegetation greenery loss could not be determined using aNDVI values at the peninsula level. Based on the earlier results, the neighborhoods were delimited by each cyclonic event to isolate the areas damaged by precipitation and associated winds, which allowed the quantitative estimation of the vegetation' s resilience in the face of such phenomena.

Impact zones
Cyclonic systems typically cross the peninsula in an east-west direction, either through the north or south, without showing a particular trend (Figure 7). That is, the number of cyclones that passed through the south was 9 in each case, except for three tropical depressions that crossed the peninsula diagonally, from southeast to northwest (Gordon, 2000;Bill, 2003 andCindy, 2005). The distances traveled overland varied; the smallest corresponded to tropical storm Richard (2010), with 57.5 km, and the maximum corresponded to hurricane Ernesto (2012, H2), with 342 km. Of the five cyclones that made landfall in the Yucatan Peninsula in the hurricane category, three entered through the northeast (Isidore, 2002;Emily and Wilma, 2005), and two entered through the southeast (Dean, 2007 andErnesto, 2012; Figure 7). Concerning the track length (total), Dean (2007) stands Tropical cyclone effects on vegetation resilience in the Yucatan Peninsula, México, between 2000-2012 out with a maximum distance of 7,700 km, and Ernesto (2012) stands out with a minimum distance of 5,500 km (Table 2). Thus, Quintana Roo' s state suffered the most considerable cyclonic disturbances, according to the aNDVI changes observed.

Vegetation resilience to cyclonic impacts
We classified the maximum damaged area and the recovery rates -resilience-into two groups (Table 3): a) those which had deficit values -loss of greenness-and b) those which favored photosynthetic activity -surplus-. Seven cyclones (33.3%) that reached the tropical depression or tropical storm categories caused an immediate increase in photosynthetic activity. Such were the cases of Karl (2010), Rina (2011), and Larry (2003), when after 2 or 4 weeks, the aNDVI reached levels higher than those before the cyclones had landfall (Figure 8). Tropical forests suffered the most significant damage from cyclones. In nine of the 14 cases (64%), negative aNDVI was observed. Hydrophilic vegetation was also damaged negatively in five cases (36%). In 11 cases (52%), recovery needed four to five weeks; that is, the aNDVI values returned to extremely close levels before the cyclone landfall at a lapse rate. In 38% of the cases, the recovery lapses were 2 or 3 weeks. The case of hurricane Wilma (2005) stands out; its damages were the most extreme, with 70% negative impacts. The most prolonged recovery corresponded to Hurricane Emily (2005) and tropical storm Claudette (2003). It took more than ten weeks in the case of Hurricane Emily, while after Claudette, it took more than thirteen weeks (Figure 9).
The impact of cyclones on vegetation was classified based on the maximum differences in aNDVI values, and as time passed, the aNDVI returned to precyclone levels. The results by category are as follows: • Negative impacts: 67% (14 cyclones) -Little significant impact: in ten cases, the differences in the aNDVI did not stand for an increase of more than 10%, most of them corresponding to depressions and tropical storms, except for hurricane Ernesto (

Comparative statistical analysis
The multiple correlations showed a highly significant relationship between the impacted area and aNDVI (R 2 = 0.99); the greater the area impacted, the more significant the decrease in NDVI values. A moderately significant relationship between the cyclonic categories and the aNDVI was obtained (R 2 = 0.63). A higher intensity of sustained winds was found in 60% of cases; the damaged area was higher, and the aNDVI differences were more significant. In contrast, there was not a correlation between the cyclone categories and the length of their tracks overland (R 2 = 0.16) and with the recovery interval of the aND-VI (R 2 = -0.07), and between the distance traveled and the maximum damaged area (R 2 = -0.12) and the difference in the aNDVI (R 2 = -0.17). This pattern is likely due to the fact that the cyclone tracks crossed the mainland very near the coast but did not make landfall. The results of the correlations between the maximum damaged area and the difference in the aNDVI are consistent with those obtained through linear regression (Pearson R 2 = 0.97), which indicates a directly proportional relationship between the maximum impacted area and the difference in the aNDVI (Figure 10).

Discussion
Resilience is a recurrent concept in different fields and disciplines, which has led to the formulation of various definitions of the term (Hosseini, Barker, & Ramirez-Marquez, 2016). To study the resilience and adaptation of natural systems, the ecological resilience definition proposed by Holling (1996) is still valid. Resilience estimations can be quantitative or qualitative (Wang, Nistor, & Pickl, 2017). For the present study, the change in photosynthetic activity, measured through aNDVI, was assumed to be a quantitative indicator of vegetation recovery capacity. We found that vegetation recovery rates -resilience-after a cyclone crossing took an average of 8 to 9 weeks for the Yucatan Peninsula. Tropical forest covers approximately 78% of the peninsula' s surface and shows greenery recovery in periods less than 8 to 9 weeks after high-intensity hurricanes cross the region. In contrast, low-intensity hurricanes presented decreases in the values of the aNDVI in percentages of less than 10% of the surface with recovery rates between 2 and 4 weeks. Seven of the cyclones identified as tropical depression and tropical storm categories damaged the vegetation but also allowed for rapid regeneration.
The two most intense cyclones in the Yucatan Peninsula from 2000 to 2012 were Wilma (2005) and Dean (2007). The first reached the peninsula near Quintana Roo state, in an area known as the "Riviera Maya." Some of the most significant land-use changes have occurred in this area due to tourist activities and urbanization. Wilma wreaked havoc in the cities of Cozumel, Playa del Carmen, and Cancun. Among the damages are the destruction of plant communities and urban systems, which means significant economic losses for the region (Zenteno Casas, Avelar Frausto, &Reinoso Angulo, 2006 andRivera-Monroy et al., 2020). The second case was hurricane Dean (2007), which made landfall in the southwestern part of Quintana Roo and led to the highest number of uprooted and broken trees (Navarro-Martínez, Durán-García, & Méndez-González, 2012). Mangrove and medium-stature forests in Mahahual were defoliated to varying degrees; however, within a month, the medium-stature forest had recovered -foliation-by nearly 80% (Islebe, Torrescano-Valle, Valdez-Hernández, Tuz-Novelo, & Weissenberger, 2009). In previous work, Sánchez & Islebe (1999) emphasized that mangroves begin to recover after five to seven months. Although our findings were based on remote sensing products, the vegetation types damaged by cyclonic activity coincide with those reported in the literature.
Some studies on resilience to cyclones have covered the Mexican Pacific coast. Bhaskar et al. (2018) researched the resilience of the dry forest to the impact of Hurricane Jova (2011, H2) in the Chamela-Tropical cyclone effects on vegetation resilience in the Yucatan Peninsula, México, between 2000-2012 Cuixmala Biosphere Reserve in the state of Jalisco, Mexico. Their results revealed a decrease in the total basal area. The most frequent damage is uprooted trees and the loss of small branches. This is why the most dense vegetation, such as mangroves and tropical forests, must be conserved and protected against illegal logging.
As in the case of the dry forest of the Chamela-Cuixmala Biosphere Reserve, along the coast of the Mexican Pacific and the Gulf of Mexico, different types of forest ecosystems exposed to the impact of tropical cyclones exist (Rosengaus-Moshinsky, Jiménez-Espinosa, & Vázquez-Conde, 2002;INEGI, 2017), where the resistance and resilience capacities could be estimated by implementing the methodology presented in this study.

Conclusions
A methodology based on remote sensing products and techniques, combined with vegetation indexes, such as the NDVI and the aNDVI, allowed us to estimate the impacted areas and vegetation resilience capacity between a period of less than 8 and 9 weeks after each of 21 registered tropical cyclones that hit the Yucatan Peninsula between 2000 and 2012.
This study shows that satellite images and the green index are helpful for detecting areas where the vegetation status or conditions after the cyclone impact require priority attention and fieldwork and studies to identify the exact causes of the minimum recovery rates. The techniques used made it possible to identify relationships between each cyclone' s impacts given their trajectory, category, and permanence by neighborhood.
The rapid recovery of photosynthetic activity following the passage of tropical cyclones suggests that the vegetation on the peninsula is well adapted to the interaction with such hydrometeorological phenomena. However, of the 21 cyclones analyzed only in those cases considered extreme events in terms of their wind intensity and inland permanency, the vegetation suffered severe damage that required long periods for their recovery.
Although the effects of high-intensity tropical cyclone landfall on vegetation have already been reported in the scientific literature, the study results allow us to compare the resilience capacity differences between four types of vegetation against the impact of cyclones of varying intensities. Our findings could contribute to assessing the vulnerability of the peninsula' s vegetation to severe hydrometeorological events and establishing priority areas for their prompt inspection and, where appropriate, designing and carrying out more detailed and higher-scale studies in specific areas.
These research findings can also help to estimate recovery costs, apply conservation measures, and sustainably manage farming and ecotouristic locations in the region. Likewise, they allow the ranking of the areas with the most significant impacts, easing the establishment and delimitation of regions or priority areas for their attention.
To advance the understanding and knowledge of the processes of adaptation and recovery of the pe-ninsula´s forest ecosystems, investigations must be conducted that consider the effects caused not only by a single phenomenon and in isolation but also by the integration of multiple natural and extreme anthropogenic events at different spatiotemporal scales. This could also be considered to design different uses and management schemes that identify the effects and responses based on the four types of vegetation present in the region.
Finally, we propose to consider other resilience models, such as those based on a bivariate approach of resistance and resilience applied under a framework of spatial analysis techniques. Adding additional analysis variables, such as precipitation volumes and the flood zones associated with each cyclone studied, would allow future studies to better estimate the intrinsic attributes and extrinsic environmental factors for an ecological unit from observing long-term state changes.

Funding
The author thanked the National Autonomous University of Mexico, the Faculty of Philosophy and Letters, and Postgraduate in Geography for the scholarship granted to carry out the Master of Geography studies, as well as to the University of Quintana Roo, the Observation Laboratory and Space Research, the Sustainable Development Division and CONACYT, for the scholarship granted (597620) to carry out doctoral studies in Sustainable Development in the Cozumel Academic Unit.