Regionalization of the Mexican Gulf of Mexico: A synthetic approach for multipurpose sensitivity analysis

Ernesto Vega-Peña; Rigel Alfonso Zaragoza-Álvarez; Gabriela Reséndiz-Colorado; Edward Michael Peters; Juan Carlos Herguera

doi:10.1371/journal.pstr.0000213

Abstract

The Gulf of Mexico is an ocean basin with high environmental and economic importance where, inevitably, natural and anthropic disasters will take place. Therefore it is relevant to assess and regionalize its main biological, environmental and socioeconomic characteristics, for both management and theoretical purposes. In order to characterize the variation of this system, we divided the Gulf of Mexico’s exclusive economic zone of Mexico into 2682, 256.2 km² hexagonal cells grouped within four zones: oceanic, coastal, insular, and inland. In each cell, we assessed 32 biological, ecosystem, and socioeconomic variables. We made a principal component analysis (PCA) for each zone using the standardized variables to detect cell clusters. We found heterogeneity wtihin the four zones, each having significantly different regions. The coastal zone was the most complex because its regions combine environmental and socioeconomic attributes. Using network analysis with PCA results we identified groups of synergistic and antagonistic variables in each zone. In general, we observed that the synergistic variables are proportionally more connected than the antagonistic ones. However, in the oceanic zone, the connectivity of the antagonistic variables was slightly higher than in the other three zones. This study offers a new integrative view of a complex region with high biological and socioeconomic relevance in a global context. These findings can be useful both for applied and academic aims. The link between PCA and network analysis offers a novel approach for identifying the relative importance of regions and finding not obvious connections between variables. This approach can be used in any socioecological system, whether marine or terrestrial, large or small.

Author summary

It is essential to characterize simultaneously biotic and socioeconomic attributes in order to achieve a better understanding of a region, whether for theoretical or sustainable management purposes. Also it is important to identify potential interactions (synergistic or antagonistic) between the variables analyzed. These approaches lead us to consider that resource management should include aspects that can be neglected at first glance. The Gulf of Mexico is a basin of great biological and economic importance, where three countries share interests (Cuba, México and the US). So it is important to fully understand the spatial variation and the relative importance of biological and socioeconomic variables involved and how they may interact with each other. In this study we divided the Gulf of Mexico’s exclusive economic zone of Mexico into 2682 hexagonal cells classified in four zones (oceanic, coastal, insular, and inland), where me measured 32 biological and socioeconomic variables. We linked PCA and network analysis in order to characterize spatial heterogeneity and interactions between variables. We found different regions within zones and that variable importance and their interactions differ between zones. This is a proposal that may help to foster a shared approach for the management and conservation of this region.

Citation: Vega-Peña E, Zaragoza-Álvarez RA, Reséndiz-Colorado G, Peters EM, Herguera JC (2025) Regionalization of the Mexican Gulf of Mexico: A synthetic approach for multipurpose sensitivity analysis. PLOS Sustain Transform 4(12): e0000213. https://doi.org/10.1371/journal.pstr.0000213

Editor: Jose Carlos Báez, Spanish Institute of Oceanography: Instituto Espanol de Oceanografia, SPAIN

Received: January 18, 2024; Accepted: December 1, 2025; Published: December 19, 2025

Copyright: © 2025 Vega-Peña et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the manuscript and its Supporting information files.

Funding: The author(s) received no specific funding for this work.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Ocean basins and their coastal regions can be viewed as complex socio-ecological systems, characterized by multi-scale (spatial and temporal) interactions between biological and socio-economic components.

They are worldwide subjects of pressure from human activities that include fisheries, the oil industry, deforestation, erosion, land use change, chemical pollution, and climate change [1–3], of which the Gulf of Mexico basin (GOM hereafter) is a typical example. The GOM region (nearly 1,500,000 km²) encompasses high environmental (mangrove forests, reefs, coastal lagoons, river mouths, and seagrass) and biological diversity [4–6]. The region’s ecological complexity adds upon its socioeconomic complexity due to the region sharing three contrasting national interests between the U.S., Mexico, and Cuba. The exclusive economic zone of Mexico in the GOM is limited to the east with that of Cuba, and to the west, with Mexico and U.S. coastlines. According to [7], the Mexican GOM zone extends to nearly 837 162 km² with a mean length of 865 km, a width of 1,350 km, and a maximum depth of nearly 3,900 m (S1 Fig).

The intense oil industry, fisheries, tourism, and commodity transportation activities taking place within the GOM make the occurrence of accidents unavoidable, like the Ixtoc and Deepwater Horizon oil spills [8,9]. The occurrence of such environmental disasters in a basin with high socioeconomic importance in the context of global climatic change makes it essential to have a socioenvironmental characterization of the GOM that allows identifying the zones potentially most sensitive to possible disturbances.

Sensitivity analysis is the most frequently cited approach reported in the scientific literature for assessing which parts of a system are most likely to be disrupted during an accident [10–13]. They frequently use compound indexes that synthesize the information from a set of variables, that allow for establishing a hierarchy of susceptibility to disasters between the regions that make up the study area [14,15]. However, such an approach does not allow us to identify the relationships between the variables included in the synthetic indexes adopted. It also does not allow us to compare the variables’ relative contribution to the value of the synthetic indexes or to identify groups of variables having similar relative importance or relationships. Recently, the GOM has been studied with different approaches that are important for understanding the possible consequences of climatic change [16,17] and how they can be assessed through adaptive management on a local scale [18,19]. However, the hierarchization of the importance of its components and regions remains a necessity, both theoretically and in terms of management.

Disturbance events like oil spills or shifts of main environmental drivers might have unexpected consequences on basin’s attributes and processes such as biodiversity, ecosystem services, or tourism, among others [20–23]. This stems from their multiple interconnected components [24], so it may be a good idea to understand how they are related. To address these issues it is useful to use different approaches simultaneously. It is necessary to understand the general patterns of variation throughout the system and then to identify the relative importance of variables and the interactions between them. The first issue is accomplished with multivariate techniques like PCA, which synthesize data variation and identify clusters of variables and observations in a system. For the second one network analysis [24] is appropriate as they can easily detect the synergistic and antagonistic relationships between variables. The link between these two approaches is achieved through the variable loadings matrix (eigenvectors of PCA). Linking PCA and network analysis allows for simultaneous analysis of both global patterns of variation and relationships between a specific set of variables. This analysis proposal can be very useful for decision-making, and may be used in other socio-ecological systems, in addition to marine and coastal ones. This approach may also offer a new perspective for characterizing such systems.

Our general goal in this work was to make an environmental characterization of the exclusive economic zone of Mexico within the GOM. In particular, we attempted to identify and characterize zones based on the importance of their descriptive variables and to understand the interrelationships between them in terms of their synergy or antagonism. This double-goal approach

Materials and methods

Study area

Mexico has jurisdiction over about 47% of the GOM, including five states with a total population of 4,359,765 inhabitants of 11,503 settlements located near the coastline. The population in the GOM region is involved in diverse economic activities using the variety of natural resources available in its territory, the most prominent of which are commercial fisheries, tourism, maritime transportation, recreational boating, and oil and gas extraction [25].

Spatial characterization of the Mexican Gulf of Mexico

For its characterization, the Mexican exclusive economic zone of the GOM was divided into 2,682 hexagonal cells, each with a surface of 256.2 km² (S1 Fig). The use of hexagonal cells minimizes the edge effect between cells, for which it was previously used by [26] for studying the land surface of Mexico.

Each cell was assigned to one of four zones providing a general characterization of the environmental variation present in the GOM: inland zone (N_ont = 265), coastal zone (N_coastal = 236), insular zone (N_islands = 21), and oceanic zone (N_sea = 2160). The inland zone included about 30 km inland from the coastline covering 20 settlements with populations above 50,000 [27].

Selection of the databases

For each cell, biological, fishery, socioeconomic, and infrastructure data were searched online in 11 governmental databases (S1 Table). All of them are free access, have some degree of review by experts and are updated periodically, so they can be considered as reliable. A total of 32 variables were used to characterize each cell (Table 1). Only variables with values in all cells were selected. Each of them represent an aspect of GOM relevant for management, also susceptible to being analyzed with multivariate methods. Each variable was individually processed according their particular attributes. Detailed information on data processing is found elsewhere (S2 Table).

Download:

Table 1. Variables used in sensitivity characterization of the exclusive economic zone of Mexico in the Gulf of Mexico.

https://doi.org/10.1371/journal.pstr.0000213.t001

Data standardization

The R platform was used for data analyses [28], while final maps were made with ArcGis pro - ArcPy. 2.1 [29]. The selected variables were double standardized using the decostand function in the vegan package [30]. First, the columns values were divided by their corresponding maximums so that the values range between 0 and 1, so all the variables were in the same scale although they have different original measure units and represent completely different aspects of the system. After that, the values in each row were divided by the total of variables used in the database. In this way a row sum shows the importance of a cell (ranging from 0 to 1), while a column sum represents the importance of a variable (ranging from zero to 2682). By this simple procedure, it is possible to compare qualitatively different variables and cells. This work has a strong quantitative focus, due to the treatment given to the variables and the methods used.

Identification and characterization of variable groups in the Mexican Gulf of Mexico

For all the variables in each cell and for all zones, principal component analysis (PCA) was used to generate multivariate and synthetic descriptors using the dudi.pca function in the ade4 package [31]. Cell clusters in each zone were looked for based on the scores of the first two principal components of the PCA using the NbClust function in the NbClust package [32]. Because these cell clusters are spatially aggregated, they were designated as regions within the zones. To find out if the previously found regions (cell clusters) were different, a permutational multivariate analysis was run using the adonis2 function in the vegan package [30]. Finally, using the pairwise.adonis function in the funfuns package [33], a pairwise comparison between regions was made to know if the observed differences were statistically significant.

Influence of variables in the Mexican Gulf of Mexico

Relative importance of variables.

To better categorize the regions inside each zone, the standardized variables within each cell were ranked and the four higher rank variables (4-tuples) having a cumulative importance equal or higher than 80% of the total cell were determined. After identifying the importance of variables within regions, alluvial plots were generated with the geom_santkey function in the ggsankey package [34]. Last, the five most frequent 4-tuples were identified for each region, therefore improving the characterization of the internal variation within each region.

Variables’ system-wide synergistic or antagonistic influence.

Briefly explained, PCA combines variables into new axes (principal components) using weights that maximize the explained variance. Each principal component shows the relative importance of the variables in explaining the variation of the system. There are as many principal components as there are system variables. The variation in the system is explained by all the principal components. The first of them explains the largest proportion of the variance, followed by the second, and so on. One of the fundamental results of PCA is the biplot,which shows on a 2D plane the relative importance of the variables in the first two components. In the biplot, the arrows represent the contribution and importance of each variable in the components while the points show the observations projected into that new space [35].

The link between PCA and network analysis is achieved with a matrix whose columns are the first two components (that is, the one used to represent the variables in the biplot). This matrix shows the coordinates of the variables in the plane. And since all the variables (or arrows) have the same origin (0,0) in the plane, it’s possible to calculate the angles between all pairs of variables. To understand the system-wide influence of variables, the vector angle matrix of the PCA biplot (Θ_PCA) was used. For example, in a PCA made with 5 variables, there will be 10 angles to represent all variable pairs ({n²-n}/2 = 10; n = 5). The matrix Θ_PCA contains the angles between all pairs of the variables used in the analysis, classified according to their magnitude. If the ϕ_a,b angle between variables a and b is less than 90°, then there is a positive correlation, or synergy, between the pair of variables, and if ϕ_a,b > 90°, there is a negative correlation, or antagonism, between the a and b variables. Therefore, the proportion of > 90° and < 90° angles can be known for a set of variables. For each variable, 1,000 permutations were run, for each calculating the synergistic (S) influence —that is, the percentages of less than 90° angles (S_varX = (N_s/N_tot × 100; N_s = ϕ < 90°; N_tot = total ϕ angles)— and the antagonistic (A) influence —the percentage of angles larger than 90° (A_varX = (N_a/N_tot × 100; N_a = ϕ > 90°; N_tot = total ϕ angles). The system-wide effect of a z variable was calculated as the S_varX and A_varX averages.

Clustering of antagonistic and synergistic variables.

The PCA biplot matrix (Θ_PCA) can also be used to identify clusters of variables. Clustering of variables having similar impacts might be essential to understanding how attributes and processes are potentially connected and which might be the consequence on the remaining variables of modifying one of them. The Θ_PCA matrix can be decomposed into two submatrices Θ_S and Θ_A, which, respectively, contain the lower than (synergistic connections between variables) and larger than 90° angles (antagonistic connections between variables).

With the Θ_S and Θ_A matrixes calculated for each zone, network models were generated in the igraph package [36], thus allowing identifying the network structure, that is, clusters of more akin variables either due to their connectivity or to their proximity. After an informal comparison of the various approaches to cluster identification, the multilevel optimization algorithm was chosen and performed using the cluster_louvain function [37]. The cluster_louvain algorithm identifies variable clusters (nodes) within each network (synergistic or antagonistic matrix). For each network type encountered with the combination of zones and regions, one or more groups to which variables belong can be found. An M matrix was made in which each row is a group within of the network, the columns are the variables, and the cells have a value of 1 when the variables correspond to the group, or 0 if they don’t. To identify which variables were more similar, dendrograms were made for the M_S (synergistic variables) and M_A (antagonistic variables) matrices using the hclust function in the stats package [28] with Jaccard dissimilarity and Ward’s clustering. After, a tanglegram was plotted with the tanglegram function of the dendextend package [38] to identify the coincidences between the two resulting dendrograms. The network’s structure was characterized by the most simple descriptor, the node’s degree (d), which is the number of connections to other nodes [39]. The average degree (d_avg) was calculated from the Θ_S and Θ_A submatrixes of each zone and region.

Results

Identification of regions within zones

In three of the zones (insular, coastal, and oceanic) we found that the variance explained by the first four PCA axes was relatively low (16% to 50%), but in the insular zone that value was 80% (S3 Table). The results of the clustering analyses showed five regions in the marine zone, three in the coast and inland zones, and two in the island zone (S2 Fig), all regions being significantly different between them (S4 Table).

The relative importance of variables

The results of the variable hierarchization within cells showed that regions within zones are highly variable (Fig 1), and the four most abundant variable combinations (4-tuples) suggested the presence of subregions within regions (Table 2).

Download:

Table 2. Main groups of variables within zones, regions, and subregions in the exclusive economic zone of Mexico in the Gulf of Mexico. Variables are ordered by importance value (v1 > v2 > v3 > v4). N: number of cells pertaning to subregion.

https://doi.org/10.1371/journal.pstr.0000213.t002

Download:

Fig 1. Spatial location of regions within each zone in the exclusive economic zone of Mexico in the Gulf of Mexico.

Letters identify the most frequent 4-tuples (see Table 2). Cells with a dot have unique combinations of variables. The base map was generated using SRTM data from the U.S. Geological Survey (USGS) (http://www.usgs.gov; License Policy: http://www.usgs.gov/information-policies-and-instructions/copyrights-and-credits). Coastline was sourced from Natural Earth (https://www.naturalearthdata.com, License Policy: https://www.naturalearthdata.com/about/terms-of-use/).

https://doi.org/10.1371/journal.pstr.0000213.g001

Oceanic zone.

We identified five regions within the oceanic zone (a to e). While region 1 (a) showed high complexity due to the high number of potentially important varieties, the main variables were commercial fishing of crustaceans (cCrust), fish (cFish), and mollusks (cMoll), followed by environmentally important variables and oil facilities (Fig 2a). In contrast, in the oceanic zone region 2 (b) four variables were highly important: commercial fishing (cFish), shipping routes ((ShipRt), inland oil facilities (OilFac), and oil drilling activities in ultradeep water (OilUD; Fig 2b). Region 3 of the oceanic zone (Fig 2c) was the simplest comparatively because it has only three critical variables: commercial shark and ray fishing (cShk), commercial fishing (cFish), and commercial mollusk fishing (cMoll: Fig 2c). While region 4 of the oceanic zone (Fig 2d) was similar to region 3 in the same zone because of having the same three dominant variables (cShk, cFish, and cMoll), it had a higher complexity due to the presence within it of more potentially relevant variables like cMoll and cFish (Fig 2d). Finally, region 5 of the oceanic zone was relatively simple due to the presence of few prominent variables; however, the coexistence in this region of oil drilling in shallow water (OilShll) with commercial shrimp (cShrimp), crustacean (cCrust), and fish (cFish) fishing is an interesting combination of relevant variables (Fig 2e).

Download:

Fig 2. Alluvial plots with the relative importance of variable combinations in the regions within the zones of the exclusive economic zone of Mexico in the Gulf of Mexico.

Variables are organized from left to right according to their importance value (v1 > v2 > v3 > v4). a to e: regions in the oceanic zone; f to h: regions in the coastal zone; i to k: regions in the inland zone; l and m: regions in the insular zone.

https://doi.org/10.1371/journal.pstr.0000213.g002

Coastal zone.

In the coastal zone, we identified three regions sharing the important variables commercial fishing of sharks and rays (cShk), and fish (cFish), the regions being different because of the presence of complementary important variables in each. In region 1 of the coastal zone (Fig 2f), these complementary variables were relevant to conservation by expert opinion (envXp), relevant to conservation by legal regulation (nvL), and relevant to ecosystem services (envRe), followed by socioeconomic variables like social security (SocSec) and poverty (Pov; Fig 2f). In comparison with the latter region, in region 2 of the coastal zone (Fig 2g) the above-mentioned socioeconomic variables had the highest importance and included social deprivation (SocDprv), while the environmental variables were less significant (Fig 2g). Finally, region 3 of the coastal zone (Fig 2h) displayed a more equilibrated representation of the same variables (Fig 2h).

Inland zone.

The inland zone was also made up of three regions (i, j, and k) within which the most influential variables were socioeconomic, as we expected from the high population levels found here. Among the socioeconomic variables in these three regions, social security (SocSec) was dominant. Region 1 of the inland zone (in1) was the most complex because it had more important variables than the other two regions in the same zone. These variables were, in order of importance, poverty (Pov), social deprivation (SocDprv), health services (HltSrv), and educational lag (eduLag; Fig 2j). Region 2 of the inland zone (Fig 2i) had less important variables and, comparatively, educational lag (eduLag) and relevant to conservation by expert opinion (envXp) were the most important (Fig 2i). Finally, region 3 of the inland zone (Fig 2j) resembled i but, in the former, environmental variables were less significant, and social deprivation (SocDpr) had a higher importance.

Insular zone.

Within the insular zone, we identified two regions (l and m) differing in the number of important variables present in them. Region 1 (l) of the insular zone had few relevant variables, of which, in order of importance, commercial fishing of sharks and rays (cShk), fish (cFish), and (cMoll) were the most influential (Fig 2l). In contrast, region 2 of the insular zone (m) had more important variables, commercial fishing activities remaining important, but the variable relevant to conservation by expert opinion (envXp) was also influential in the region (Fig 2m).

System-wide synergistic or antagonistic influence of variables

Inland zone.

The results of our permutation analyses showed that all the variables present in the insular zone might have synergistic or antagonistic interactions, with no variables displaying a stronger influence over the others (S3a Fig).

Oceanic zone.

Most variables in the oceanic zone have system-wide synergistic influences. The strongest synergistic variables in this zone were oil drilling activities in deep (OilD) and ultradeep (OilUD) water and commercial crustacean (cCrust) and mollusk (cMoll) fishing. In this zone, the variables with the strongest system-wide antagonistic influence were inland oil facilities (OilFac), oil drilling in shallow water (OilShll), and the presence of endangered species of mammals (Mamm), invertebrates (Inverteb), and fish (Fish) (S3b Fig).

Coastal zone.

Most of the variables present in the coastal zone had system-wide synergistic effects. The strongest of these synergistic influences were due to commercial fishing of crustaceans (cCrust), shrimp (cShrimp), and mollusks (cMoll), and the presence of endangered invertebrate species (Inverteb). The variables with system-wide antagonistic influences in the coastal zone were population (Pop), permanent dwellings (Dwell), and railways (Train) (S3c Fig).

Insular zone.

The system-wide influences of variables in the insular zone were mostly synergistic, the strongest such ones corresponded to commercial shrimp fishing (cShrimp), inland oil facilities (OilFac), and relevant to conservation by expert opinion (envXp). The variables with the strongest system-wide antagonistic effects in the insular zone were the presence of endangered species of fish (Fish) and invertebrates (Invert), and shipping routes (ShipRt) (S3d Fig).

Identification of synergistic and antagonistic groups of variables

We can interpret the average node degree (d_avg) of a network as the proportion of nodes in the network to which any other node connects. Node degree values of about 0.5 suggest that each node connects to 50% of the total nodes in the network. In general, synergistic networks are proportionately more connected than antagonistic networks (S4 Fig). However, the network connectivity in the oceanic zone was slightly higher than in the other zones. Most regions within the oceanic, coastal, inland, and insular zones displayed three variable groups in the synergistic and antagonistic networks (S5 Table); however, region 2 of the coastal zone (g) had a higher complexity with four groups in its synergistic network and three in its antagonistic network. There was little coincidence between the synergistic and antagonistic dendrograms because of differences in their node structure (entanglement = 0.69; Spearman’s cophenetic correlation coefficient = 0.3385). The cluster variables differed because they shared only four variable groups (Fig 3).

Download:

Fig 3. Tanglegram of the dendrograms of synergistic (S) and antagonistic (A) interactions between variables in the zones and regions of the exclusive economic zone of Mexico in the Gulf of Mexico.

Dashed lines represent unique nodes in dendrograms. Nodes with solid lines are present in both dendrograms. Colored lines show terminal branches present in both dendrograms. Meaning of variable acronyms: AirP, airport facilities; Amph, presence of endangered amphibian species; cCrust, presence of commercial crustacean fishing activities; cFish, presence of commercial fishing activities; cMoll, presence of commercial mollusk fishing activities; Crop, agriculture; cShk, presence of commercial shark and rays fishing activities; cShrimp, presence of commercial shrimp fishing activities; Dwell, permanent dwellings; EduLag, educational lag; envL, relevant to conservation by legal regulation; envRe, relevant to ecosystem services; envXp, relevant to conservation by expert opinion; Fish, presence of endangered fish species; HltSv, health services; Invertebr, presence of endangered invertebrate species; Mamm, presence of endangered mammalian species; OilDeep, oil drilling in deep water; OilFac, inland oil facilities; OilShll, oil drilling in shallow water; OilUD, oil drilling in ultradeep water; Plant, presence of endangered plant species; Pop, population; Port, port; Pov, poverty index; Rept, presence of endangered reptile species; SaltWk, salt mining; ShipRt, shipping route; SocDprv, social deprivation; SocSec, social security; Train, railways; vuingF, income vulnerability.

https://doi.org/10.1371/journal.pstr.0000213.g003

Discussion

In this work we made a socioenvironmental regionalization of the exclusive economic zone of Mexico in the Gulf of Mexico and assessed the relative importance and potential interactions between the variables we used. Regardless that the size of the hexagonal cells we chose for the analysis was only 256 km² (approximately equivalent to a square with 16 km sides), we observed high cell heterogeneity in the four zones (inland, coastal, oceanic, and insular). We found the coexistence of a diversity of environments, natural resource management strategies, and degree of development of the resident population within the relatively small cells, which summarizes into databases containing widely different combinations of attributes. Consequently, we considered it essential to standardize the variables for determining their relative importance. The GOM can be seen as a system with multiple stressors acting upon multiple targets or attributes, which are themselves interactive [40–42]. The occurrence within a region of two processes or attributes simultaneously does not necessarily imply that they are determined by a causal relationship; nevertheless, variable independence cannot be discarded a priori. The well-established inverse relationship of the distances between the locations in space and the similitude magnitudes of measured attributes made it reasonable to assume that the same would happen in the GOM.

We obtained synergistic and antagonistic interactions between variables based on the matrix of the angles between the main PCA vectors (Θ_PCA) for zones and regions. After, we were able to identify variable groups with the highest connectivity by sorting the elements of the angle matrix into two submatrices (Θ_S containing only values smaller than 90° and Θ_A containing only values larger than 90°). Finally, we built network models with these matrices and analyzed the networks’ structures. The results of our perturbation analysis suggested a higher importance of synergistic relationships between variables in the coastal zone, followed by those in the inland, oceanic, and insular zones. The inland zone was where the fluctuation of variables was most noticeable, followed by the oceanic, coastal, and insular zones, which suggested that the same disturbance (for example, an oil spill) would have different consequences in each zone.

Since the magnitude of the synergistic and antagonistic interactions is obtained from the PCA, it would be justified to expect an inverse relationship between them; however, the dendrograms we built with the same variables assessed with these two metrics were different. This suggests that the effects of the propagation of a disturbance in variable a on the remaining variables would differ according to the kind (antagonistic or synergistic) and intensities of the connections between variable a and other variables.

Network models and PCA have been previously used separately —for example, by [43] and [44], respectively— to evaluate the vulnerability and interactions between elements of a sociobiological system. However, to the best of our knowledge, our approach combining both techniques had not been applied before to this means. Our approach is potentially interesting for identifying unnoticeable connections between variables. The analysis of connections between coexistent variables is based on the multivariate linear model, as is PCA; therefore, if variables had non-linear or indirect interactions, the latter approach might be limited.

Commonly, the data presented in this work are applied to generate tactical information, that is, aiming at particular objectives (for example, oil spills) or for limited time and space windows [12]. However, in this work, we chose a strategic approach that allows its adequacy to diverse circumstances like hurricanes, red tides, or sargassum invasions. With a rough estimate of the spatial and temporal trajectories of these examples, authorities can easily use the maps showed here to prioritize risk regions and develop appropriate action plans for them. The results of this work can also be a useful input for land, coastal, and marine planning and for the implementation of environmental management programs.

The risk from environmental or anthropogenic disasters for private commercial activities is a little-explored aspect. Economic activities play an important role in the resilience of social systems, because of that, the damage to these activities is one of the main manifestations of environmental damage [45]. The economic activities taking place within the GOM are critical for the economies of Mexico, the U.S., and Cuba. It is pertinent to evaluate in what measure the state and private-owned enterprises take into account and are prepared for recovering and adapting to the global climate change processes and their local manifestations.

While sensitivity analyses and multi-risk assessments have been made since nearly 40 years ago, the development of these approaches has been uneven. At a global scale, landslides have been the most studied topic, while risks associated with climate change and geophysical or anthropic processes have remained less explored. The development of sensitivity and multi-risk methodologies has differed among world regions, with few reports of works made in South America and Asia [46]. The evaluations of risks have been mostly static because they have not considered time and space variations or climate change [47]. The inclusion of climate change in risk assessment studies must encompass both qualitative (interviews, expert meetings) and quantitative (geophysics, meteorology, biology, geography) methods, and keep in mind that the systems of interest have social and economic components that are connected in feedback cycles and change through time, which makes their modeling challenging [47–49].

The transdisciplinarity of risk and sensitivity studies involves global environmental change, human-nature interactions, and natural disasters [50]. In consequence, the methods and concepts adopted in these studies are diverse, which causes a lack of homogeneity of the interests and objectives of the involved parties. A multipurpose study such as this may show flaws if it is analyzed from the theoretical perspective of an academic discipline (such as Economics, Geography, or Marine Ecology). Perhaps a relevant consequence of this approach is the difficulty of testing specific hypotheses about the study system. The double standardization method used in this study can be considered an oversimplification. In this context, other variable standardization criteria could be used to correct this. However, we believe that these shortcomings may reveal aspects not initially considered, which suggest new lines of research and analysis. Descriptive studies in socioecological systems like GOM allow us to characterize and systematize the complexity of the interactions between the ecological, social and economic dimensions that occur in it. This is relevant in cases like this, where there are very few integrative articles. Also, such approach can provide important elements for hypotheses oriented research. Finally, it is essential to update the information used for the analyses because the environmental conditions are dynamic and currently subjected to rapid variations associated with climate change and socioeconomic dynamics.

Conciliating approaches, goals, and methods in risk and sensitivity studies represent an enormous conceptual and methodological challenge. The specific environmental conditions in the study sites need consideration for precisely understanding the possible impacts and making generic evaluations that are not focused on a single disturbance like fishing activities [40,51]. Another important concept in studies of the sensitivity of socioecological systems is that of tipping points, whose identification might aid in understanding and managing marine systems in which many human activities take place [50].

The spatial characterization is essential for making risk and sensitivity analyses, but that is not completely viable in marine environments. This is because the borders of environments and processes occurring in the latter systems are less easily defined than in land systems and because these environments or processes will occur in three dimensions instead of in the two dimensions analyzed in land systems. Oceanic and coastal systems are highly heterogeneous and dynamic. A characterization of risk and sensitivity in marine systems must include their seasonality [14].

The GOM regionalization proposed in this this study offers two aspects that can be considered both novel and useful. The first is the capability to identify the relative importance of all variables within a cell and how they can interact with each other; the second refers to the finding of regions within zones. This information enhances a deeper understanding of GOM spatial, environmental and socioeconomic variation, that allows

the implementation of specific management strategies, either for a cell or a group of them.

General implications of these findings on policy and management should be positive, in the long term at least. As the relative importance of variables and their antagonistic or synergistic interactions are clearly stated in each cell, the potential conflicts between contrasting management goals (like those of commercial fishing and environment conservation) can easily be detected, which is an important step in order to achieve agreed management strategies.

It may be worthwhile to address some potential limitations of this proposal. Regionalization studies like this one depend heavily on availability and quality of data. Another important issue arises in the fact that although socioecological systems like GOM are intrinsically dynamic, analyses and methods usually do not consider this, making the results and findings some sort of “static” in nature. For example, it is likely that some variables can take different values if they are measured during hurricane season, so regionalization methods may yield different results. Sample sizes and statistical tests should also be considered as issues that can undermine confidence in a work of this type. Despite the different sample sizes in each region, the reliability of the results was explored with intensive resampling and permutation methods, which are appropriate when the underlying pdfs in the variables are not known [51].

Our study is not the first sensitivity analysis made in Mexican coastal areas [52]. We consider that more work using similar approaches is needed to develop a better and more comprehensive understanding of natural systems and the possible impacts on them from human activities. We could not analyze some hexagons in the northern oceanic zone due to the lack of information. The latter region corresponds to the limits between Mexico and the U.S., because of which it might be of interest. Ideally, the studies made in national bordering regions should use methods and criteria agreed between the neighboring nations.

Recommendations

A characterization of risk and sensitivity in coastal and marine systems would be more complete if local or regional seasonal variations are taken into account.
The inclusion of climate change trends in risk characterization would make these kind of studies more versatile and of interest to a broader group of specialists interested in conservation and management.
The national bordering regions can be an interesting opportunity for binational coordinated studies that employ consensual goals and methods.
The systems involved in risk evaluation frequently are dynamic ones, so a relevant issue can be the identification of tipping points.

Supporting information

S1 Fig. Zonification of the exclusive economic zone of Mexico in the Gulf of Mexico.

Colors indicate environmental zones. The base map was generated using SRTM data from the U.S. Geological Survey (USGS) (http://www.usgs.gov; License Policy: http://www.usgs.gov/information-policies-and-instructions/copyrights-and-credits). Coastline was sourced from Natural Earth (https://www.naturalearthdata.com, License Policy: https://www.naturalearthdata.com/about/terms-of-use/).

https://doi.org/10.1371/journal.pstr.0000213.s001

(DOCX)

S2 Fig. Cell clustering in the PCA of zones in the exclusive economic zone of Mexico in the Gulf of Mexico.

Colors correspond to regions within zones. a. Insular zone, b. Oceanic zone, c. Coastal zone, and d. Inland zone.

https://doi.org/10.1371/journal.pstr.0000213.s002

(DOCX)

S3 Fig. Sensitivity analyses of each variable over the others in four zones of the exclusive economic zone of Mexico in the Gulf of Mexico.

The color of the bars shows synergistic (“S”, orange) and antagonistic (“A”, blue) effects of a single variable on the rest, and the length of the bar represents the magnitude of the system-wide interactions. (a) Insular zone. (b) Coastal zone. (c) Oceanic zone. (d) Inland zone. REFERENCE, the unmodified proportion of synergistic and antagonistic interactions in the angle matrix (ΘPCA). Meaning of variable acronyms: AirP, airport facilities; Amph, presence of endangered amphibian species; cCrust, presence of commercial crustacean fishing activities; cFish, presence of commercial fishing activities; cMoll, presence of commercial mollusk fishing activities; Crop, agriculture; cShk, presence of commercial shark and rays fishing activities; cShrimp, presence of commercial shrimp fishing activities; Dwell, permanent dwellings; EduLag, educational lag; envL, relevant to conservation by legal regulation; envRe, relevant to ecosystem services; envXp, relevant to conservation by expert opinion; Fish, presence of endangered fish species; HltSv, health services; Invertebr, presence of endangered invertebrate species; Mamm, presence of endangered mammalian species; OilDeep, oil drilling in deep water; OilFac, inland oil facilities; OilShll, oil drilling in shallow water; OilUD, oil drilling in ultradeep water; Plant, presence of endangered plant species; Pop, population; Port, port; Pov, poverty index; Rept, presence of endangered reptile species; SaltWk, salt mining; ShipRt, shipping route; SocDprv, social deprivation; SocSec, social security; Train, railways; vuingF, income vulnerability.

https://doi.org/10.1371/journal.pstr.0000213.s003

(DOCX)

S4 Fig. Average node degree (number of connections in nodes) of the synergistic and antagonistic networks within the zones and regions of the exclusive economic zone of Mexico in the Gulf of Mexico.

The scale can be interpreted as the proportion of nodes to which a node is connected to any other node in the network. Acronyms identify the studied zones and the regions within zones. co, coastal regions; in, inland regions; is, island regions; oc, oceanic regions.

https://doi.org/10.1371/journal.pstr.0000213.s004

(DOCX)

S1 Table. Databases used in the sensitivity characterization of the exclusive economic zone of Mexico in the Gulf of Mexico.

https://doi.org/10.1371/journal.pstr.0000213.s005

(DOCX)

S2 Table. Main data processes applied to each variable used in Gulf of Mexico environmental analysis.

https://doi.org/10.1371/journal.pstr.0000213.s006

(DOCX)

S3 Table. Explained variance (%) of the first five PCA axes in each zone in the exclusive economic zone of Mexico in the Gulf of Mexico.

https://doi.org/10.1371/journal.pstr.0000213.s007

(DOCX)

S4 Table. Significance test for similitude analysis (ANOSIM) of the four zones in the exclusive economic zone of Mexico in the Gulf of Mexico. a): Permutation significance tests; b): pairwise comparisons of regions within zones.

https://doi.org/10.1371/journal.pstr.0000213.s008

(DOCX)

S5 Table. Number of groups in the synergistic and antagonistic networks in the zones and regions of the exclusive economic zone of Mexico in the Gulf of Mexico.

https://doi.org/10.1371/journal.pstr.0000213.s009

(DOCX)

References

1. Visbeck M. Ocean science research is key for a sustainable future. Nat Commun. 2018;9(1):690. pmid:29449661
- View Article
- PubMed/NCBI
- Google Scholar
2. Landrigan PJ, Stegeman JJ, Fleming LE, Allemand D, Anderson DM, Backer LC, et al. Human health and ocean pollution. Ann Glob Health. 2020;86(1):151. pmid:33354517
- View Article
- PubMed/NCBI
- Google Scholar
3. Jiang D, Yang Q, Fang Y, Zhang X, Song J. A bibliometric review of environmental pollution research in major global gulfs. Water. 2025;17(10):1455.
- View Article
- Google Scholar
4. Caso M. Diagnóstico ambiental del Golfo de México. Vol. 2. México, D.F.: Instituto Nacional de Ecología; 2004.
5. Ward CH. Habitats and biota of the Gulf of Mexico: before the Deepwater Horizon oil spill. Vol. 2. Springer Open; 2017.
- View Article
- Google Scholar
6. Herzka SZ, Zaragoza-Álvarez RA, Peters EM, Hernández-Cárdenas G. Atlas de línea base ambiental del Golfo de México. Consorcio de Investigación del Golfo de México. 2021. Available from: https://atlascigom.cicese.mx/
- View Article
- Google Scholar
7. Salvador A. The Gulf of Mexico Basin. Boulder: Geological Society of America; 1991.
- View Article
- Google Scholar
8. Soto LA, Botello AV, Licea-Durán S, Lizárraga-Partida ML, Yañez-Arancibia A. The environmental legacy of the Ixtoc-I oil spill in Campeche Sound, southwestern Gulf of Mexico. Front Mar Sci. 2014;1.
- View Article
- Google Scholar
9. Beyer J, Trannum HC, Bakke T, Hodson PV, Collier TK. Environmental effects of the deepwater horizon oil spill: a review. Mar Pollut Bull. 2016;110(1):28–51. pmid:27301686
- View Article
- PubMed/NCBI
- Google Scholar
10. Laflamme A, Percy RJ. Sensitivity Mapping – With Flare! An internet approach to environmental mapping. International Oil Spill Conference Proceedings. 2003;2003(1):95–102.
- View Article
- Google Scholar
11. IPIECA. Sensitivity mapping for oil spill response. 2011. [cited 2022 Jan 1] Available from: https://www.ipieca.org/resources/sensitivity-mapping-for-oil-spill-response
- View Article
- Google Scholar
12. IPIECA. Guidelines on implementing spill impact mitigation assessment (SIMA). 2017. [cited 2022 Mar 4] Available from: https://www.ipieca.org/resources/guidelines-on-implementing-spill-impact-mitigation-assessment-sima
- View Article
- Google Scholar
13. Flores-Medina PW, Sepp-Neves AA, Coppini G, Morales-Caselles C. Strategic environmental sensitivity mapping for oil spill contingency planning in the Peruvian marine-coastal zone. Sci Total Environ. 2022;852:158356. pmid:36049685
- View Article
- PubMed/NCBI
- Google Scholar
14. Zacharias MA, Gregr EJ. Sensitivity and vulnerability in marine environments: an approach to identifying vulnerable marine areas. Conserv Biol. 2005;19(1):86–97.
- View Article
- Google Scholar
15. Hochrainer-Stigler S, Trogrlic Šakic R, Reiter K, Ward PJ, de Ruiter MC, Duncan MJ, et al. Toward a framework for systemic multi-hazard and multi-risk assessment and management. iScience. 2023;26(5):106736. pmid:37216095
- View Article
- PubMed/NCBI
- Google Scholar
16. Appendini CM, Ruiz-Salcines P, Marsooli R, Cerezo-Mota R. Assessing the effects of climate change on the Gulf of Mexico wave climate using the COWCLIP framework and the PRECIS regional climate model. Ocean Modelling. 2025;194:102486.
- View Article
- Google Scholar
17. Martínez ML, Silva R, Chávez V, López-Portillo J, Salgado K, Marín-Coria E, et al. The challenges of climate change and human impacts faced by Mexican coasts: a comprehensive evaluation. PLoS One. 2025;20(4):e0320087. pmid:40245097
- View Article
- PubMed/NCBI
- Google Scholar
18. Valdés-Rodríguez OA, Del Valle-Cárdenas B, Conde C, Zavaleta-Lizárraga L. Contributions to sustainable development in coastal communities of the Gulf of Mexico while assessing climate change: a case study. Earth. 2025;6(2):43.
- View Article
- Google Scholar
19. Dorantes-Hernández JM, Morzaria-Luna HN, Pérez-Jiménez JC, Coronado-Castro E, Molina-Rosales D. Stakeholders’ perceptions of feasibility of Fishery Refuge Zone implementation in Campeche, Southern Gulf of Mexico. Region Stud Mar Sci. 2025;90:104420.
- View Article
- Google Scholar
20. Potter RWK, Pearson BC. Assessing the global ocean science community: understanding international collaboration, concerns and the current state of ocean basin research. npj Ocean Sustain. 2023;2(1).
- View Article
- Google Scholar
21. Duarte CM, Hendriks IE, Moore TS, Olsen YS, Steckbauer A, Ramajo L, et al. Is ocean acidification an open-ocean syndrome? understanding anthropogenic impacts on seawater pH. Estuaries Coasts. 2013;36(2):221–36.
- View Article
- Google Scholar
22. Ivshina IB, Kuyukina MS, Krivoruchko AV, Elkin AA, Makarov SO, Cunningham CJ, et al. Oil spill problems and sustainable response strategies through new technologies. Environ Sci Process Impacts. 2015;17(7):1201–19. pmid:26089295
- View Article
- PubMed/NCBI
- Google Scholar
23. Paulus E. Shedding light on deep-sea biodiversity—a highly vulnerable habitat in the face of anthropogenic change. Front Mar Sci. 2021;8.
- View Article
- Google Scholar
24. Dee LE, Allesina S, Bonn A, Eklöf A, Gaines SD, Hines J, et al. Operationalizing network theory for ecosystem service assessments. Trends Ecol Evol. 2017;32(2):118–30. pmid:27856059
- View Article
- PubMed/NCBI
- Google Scholar
25. Gasca J. Actividades económicas. In: Herzka SZ, Zaragoza-Álvarez RA, Peters EM, Hernández-Cárdenas G, editors. Atlas de línea base ambiental del Golfo de México. Consorcio de Investigación del Golfo de México; 2021.
26. Koleff P, Tambutti M, March IJ, Esquivel R, Cantú C, Lira-Noriega A. Identificación de prioridades y análisis de vacíos y omisiones en la conservación de la biodiversidad de México. In: Sarukhán J, editor. Capital natural de México, vol. II: Estado de conservación y tendencias de cambio. México: CONABIO; 2009. pp. 651–718.
27. Instituto Nacional de Estadística y Geografía (INEGI). Censo de Población y Vivienda 2020. [cited 2022 Jun 9] Available from: https://www.inegi.org.mx/programas/ccpv/2020/
- View Article
- Google Scholar
28. R Core Team. R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2023. Available from: https://www.R-project.org/
- View Article
- Google Scholar
29. Esri. ArcGIS Pro (Versión 3.2). 2025. Available from: https://www.esri.com/en-us/arcgis/products/arcgis-pro/overview
- View Article
- Google Scholar
30. Oksanen J, Simpson G, Blanchet F, Kindt R, Legendre P, Minchin P, et al. vegan: community ecology package. R package version 2.6-4. 2022. Available from: https://CRAN.R-project.org/package=vegan
- View Article
- Google Scholar
31. Dray S, Dufour A-B. Theade4Package: Implementing the Duality Diagram for Ecologists. J Stat Soft. 2007;22(4).
- View Article
- Google Scholar
32. Charrad M, Ghazzali N, Boiteau V, Niknafs A. NbClust: AnRPackage for Determining the Relevant Number of Clusters in a Data Set. J Stat Soft. 2014;61(6).
- View Article
- Google Scholar
33. Trachsel J. funfuns: functions I use (title case). R package version 0.1.2. 2023. Available from: https://rdrr.io/github/Jtrachsel/funfuns/
- View Article
- Google Scholar
34. Sjoberg D. ggsankey: sankey, alluvial and sankey bump plots. R package version 0.0.99999. 2023. Available from: https://github.com/davidsjoberg/ggsankey
- View Article
- Google Scholar
35. McGarigal K. Multivariate stallstlcs for wildlife and ecology research. Springer; 2000.
36. Csárdi G, Nepusz T, Traag V, Horvát S, Zanini F, Noom D, et al. igraph: network analysis and visualization in R. R package version 1.5.1. 2023. Available from: https://CRAN.R-project.org/package=igraph
- View Article
- Google Scholar
37. Blondel VD, Guillaume J-L, Lambiotte R, Lefebvre E. Fast unfolding of communities in large networks. J Stat Mech. 2008;2008(10):P10008.
- View Article
- Google Scholar
38. Galili T. dendextend: an R package for visualizing, adjusting and comparing trees of hierarchical clustering. Bioinformatics. 2015;31(22):3718–20. pmid:26209431
- View Article
- PubMed/NCBI
- Google Scholar
39. Costa LDF, Rodrigues FA, Travieso G, Villas Boas PR. Characterization of complex networks: a survey of measurements. Advances in Physics. 2007;56(1):167–242.
- View Article
- Google Scholar
40. Hobday AJ, Smith ADM, Stobutzki IC, Bulman C, Daley R, Dambacher JM, et al. Ecological risk assessment for the effects of fishing. Fish Res. 2011;108(2–3):372–84.
- View Article
- Google Scholar
41. Holsman K, Samhouri J, Cook G, Hazen E, Olsen E, Dillard M, et al. An ecosystem-based approach to marine risk assessment. Ecosyst Health Sustain. 2017;3(1).
- View Article
- Google Scholar
42. Sijing X, Gang L, Biao M. Vulnerability analysis of land ecosystem considering ecological cost and value: a complex network approach. Ecol Indic. 2023;147:109941.
- View Article
- Google Scholar
43. Abson DJ, Dougill AJ, Stringer LC. Using principal component analysis for information-rich socio-ecological vulnerability mapping in Southern Africa. Appl Geogr. 2012;35(1–2):515–24.
- View Article
- Google Scholar
44. Mohan PS. Disasters, disaster preparedness and post disaster recovery: evidence from Caribbean firms. Int J Disas Risk Reduc. 2023;92:103731.
- View Article
- Google Scholar
45. Owolabi TA, Sajjad M. A global outlook on multi-hazard risk analysis: a systematic and scientometric review. Int J Disas Risk Reduc. 2023;92:103727.
- View Article
- Google Scholar
46. Gallina V, Torresan S, Critto A, Sperotto A, Glade T, Marcomini A. A review of multi-risk methodologies for natural hazards: consequences and challenges for a climate change impact assessment. J Environ Manage. 2016;168:123–32. pmid:26704454
- View Article
- PubMed/NCBI
- Google Scholar
47. Yang W, Dun X, Jiang X, Zhou Y, Hou B, Lang R, et al. An integrated risk assessment framework for multiple natural disasters based on multi-dimensional correlation analysis. Nat Hazards. 2023;119(3):1531–50.
- View Article
- Google Scholar
48. Eakin H, Luers AL. Assessing the vulnerability of social-environmental systems. Annu Rev Environ Resour. 2006;31(1):365–94.
- View Article
- Google Scholar
49. Simeoni C, Furlan E, Pham HV, Critto A, de Juan S, Trégarot E, et al. Evaluating the combined effect of climate and anthropogenic stressors on marine coastal ecosystems: insights from a systematic review of cumulative impact assessment approaches. Sci Total Environ. 2023;861:160687. pmid:36473660
- View Article
- PubMed/NCBI
- Google Scholar
50. Lauerburg RAM, Diekmann R, Blanz B, Gee K, Held H, Kannen A, et al. Socio-ecological vulnerability to tipping points: a review of empirical approaches and their use for marine management. Sci Total Environ. 2020;705:135838. pmid:31855803
- View Article
- PubMed/NCBI
- Google Scholar
51. Chernick MR. Resampling methods. WIREs Data Min Knowl Discov. 2012;2(3):255–62.
- View Article
- Google Scholar
52. Godwyn-Paulson P, Jonathan MP, Rodríguez-Espinosa PF, Abdul Rahaman S, Roy PD, Muthusankar G, et al. Multi-hazard risk assessment of coastal municipalities of Oaxaca, Southwestern Mexico: an index based remote sensing and geospatial technique. Int J Disas Risk Reduct. 2022;77:103041.
- View Article
- Google Scholar

[ref1] 1. Visbeck M. Ocean science research is key for a sustainable future. Nat Commun. 2018;9(1):690. pmid:29449661
View Article
PubMed/NCBI
Google Scholar

[2] View Article

[3] PubMed/NCBI

[4] Google Scholar

[ref2] 2. Landrigan PJ, Stegeman JJ, Fleming LE, Allemand D, Anderson DM, Backer LC, et al. Human health and ocean pollution. Ann Glob Health. 2020;86(1):151. pmid:33354517
View Article
PubMed/NCBI
Google Scholar

[6] View Article

[7] PubMed/NCBI

[8] Google Scholar

[ref3] 3. Jiang D, Yang Q, Fang Y, Zhang X, Song J. A bibliometric review of environmental pollution research in major global gulfs. Water. 2025;17(10):1455.
View Article
Google Scholar

[10] View Article

[11] Google Scholar

[ref4] 4. Caso M. Diagnóstico ambiental del Golfo de México. Vol. 2. México, D.F.: Instituto Nacional de Ecología; 2004.

[ref5] 5. Ward CH. Habitats and biota of the Gulf of Mexico: before the Deepwater Horizon oil spill. Vol. 2. Springer Open; 2017.
View Article
Google Scholar

[14] View Article

[15] Google Scholar

[ref6] 6. Herzka SZ, Zaragoza-Álvarez RA, Peters EM, Hernández-Cárdenas G. Atlas de línea base ambiental del Golfo de México. Consorcio de Investigación del Golfo de México. 2021. Available from: https://atlascigom.cicese.mx/
View Article
Google Scholar

[17] View Article

[18] Google Scholar

[ref7] 7. Salvador A. The Gulf of Mexico Basin. Boulder: Geological Society of America; 1991.
View Article
Google Scholar

[20] View Article

[21] Google Scholar

[ref8] 8. Soto LA, Botello AV, Licea-Durán S, Lizárraga-Partida ML, Yañez-Arancibia A. The environmental legacy of the Ixtoc-I oil spill in Campeche Sound, southwestern Gulf of Mexico. Front Mar Sci. 2014;1.
View Article
Google Scholar

[23] View Article

[24] Google Scholar

[ref9] 9. Beyer J, Trannum HC, Bakke T, Hodson PV, Collier TK. Environmental effects of the deepwater horizon oil spill: a review. Mar Pollut Bull. 2016;110(1):28–51. pmid:27301686
View Article
PubMed/NCBI
Google Scholar

[26] View Article

[27] PubMed/NCBI

[28] Google Scholar

[ref10] 10. Laflamme A, Percy RJ. Sensitivity Mapping – With Flare! An internet approach to environmental mapping. International Oil Spill Conference Proceedings. 2003;2003(1):95–102.
View Article
Google Scholar

[30] View Article

[31] Google Scholar

[ref11] 11. IPIECA. Sensitivity mapping for oil spill response. 2011. [cited 2022 Jan 1] Available from: https://www.ipieca.org/resources/sensitivity-mapping-for-oil-spill-response
View Article
Google Scholar

[33] View Article

[34] Google Scholar

[ref12] 12. IPIECA. Guidelines on implementing spill impact mitigation assessment (SIMA). 2017. [cited 2022 Mar 4] Available from: https://www.ipieca.org/resources/guidelines-on-implementing-spill-impact-mitigation-assessment-sima
View Article
Google Scholar

[36] View Article

[37] Google Scholar

[ref13] 13. Flores-Medina PW, Sepp-Neves AA, Coppini G, Morales-Caselles C. Strategic environmental sensitivity mapping for oil spill contingency planning in the Peruvian marine-coastal zone. Sci Total Environ. 2022;852:158356. pmid:36049685
View Article
PubMed/NCBI
Google Scholar

[39] View Article

[40] PubMed/NCBI

[41] Google Scholar

[ref14] 14. Zacharias MA, Gregr EJ. Sensitivity and vulnerability in marine environments: an approach to identifying vulnerable marine areas. Conserv Biol. 2005;19(1):86–97.
View Article
Google Scholar

[43] View Article

[44] Google Scholar

[ref15] 15. Hochrainer-Stigler S, Trogrlic Šakic R, Reiter K, Ward PJ, de Ruiter MC, Duncan MJ, et al. Toward a framework for systemic multi-hazard and multi-risk assessment and management. iScience. 2023;26(5):106736. pmid:37216095
View Article
PubMed/NCBI
Google Scholar

[46] View Article

[47] PubMed/NCBI

[48] Google Scholar

[ref16] 16. Appendini CM, Ruiz-Salcines P, Marsooli R, Cerezo-Mota R. Assessing the effects of climate change on the Gulf of Mexico wave climate using the COWCLIP framework and the PRECIS regional climate model. Ocean Modelling. 2025;194:102486.
View Article
Google Scholar

[50] View Article

[51] Google Scholar

[ref17] 17. Martínez ML, Silva R, Chávez V, López-Portillo J, Salgado K, Marín-Coria E, et al. The challenges of climate change and human impacts faced by Mexican coasts: a comprehensive evaluation. PLoS One. 2025;20(4):e0320087. pmid:40245097
View Article
PubMed/NCBI
Google Scholar

[53] View Article

[54] PubMed/NCBI

[55] Google Scholar

[ref18] 18. Valdés-Rodríguez OA, Del Valle-Cárdenas B, Conde C, Zavaleta-Lizárraga L. Contributions to sustainable development in coastal communities of the Gulf of Mexico while assessing climate change: a case study. Earth. 2025;6(2):43.
View Article
Google Scholar

[57] View Article

[58] Google Scholar

[ref19] 19. Dorantes-Hernández JM, Morzaria-Luna HN, Pérez-Jiménez JC, Coronado-Castro E, Molina-Rosales D. Stakeholders’ perceptions of feasibility of Fishery Refuge Zone implementation in Campeche, Southern Gulf of Mexico. Region Stud Mar Sci. 2025;90:104420.
View Article
Google Scholar

[60] View Article

[61] Google Scholar

[ref20] 20. Potter RWK, Pearson BC. Assessing the global ocean science community: understanding international collaboration, concerns and the current state of ocean basin research. npj Ocean Sustain. 2023;2(1).
View Article
Google Scholar

[63] View Article

[64] Google Scholar

[ref21] 21. Duarte CM, Hendriks IE, Moore TS, Olsen YS, Steckbauer A, Ramajo L, et al. Is ocean acidification an open-ocean syndrome? understanding anthropogenic impacts on seawater pH. Estuaries Coasts. 2013;36(2):221–36.
View Article
Google Scholar

[66] View Article

[67] Google Scholar

[ref22] 22. Ivshina IB, Kuyukina MS, Krivoruchko AV, Elkin AA, Makarov SO, Cunningham CJ, et al. Oil spill problems and sustainable response strategies through new technologies. Environ Sci Process Impacts. 2015;17(7):1201–19. pmid:26089295
View Article
PubMed/NCBI
Google Scholar

[69] View Article

[70] PubMed/NCBI

[71] Google Scholar

[ref23] 23. Paulus E. Shedding light on deep-sea biodiversity—a highly vulnerable habitat in the face of anthropogenic change. Front Mar Sci. 2021;8.
View Article
Google Scholar

[73] View Article

[74] Google Scholar

[ref24] 24. Dee LE, Allesina S, Bonn A, Eklöf A, Gaines SD, Hines J, et al. Operationalizing network theory for ecosystem service assessments. Trends Ecol Evol. 2017;32(2):118–30. pmid:27856059
View Article
PubMed/NCBI
Google Scholar

[76] View Article

[77] PubMed/NCBI

[78] Google Scholar

[ref25] 25. Gasca J. Actividades económicas. In: Herzka SZ, Zaragoza-Álvarez RA, Peters EM, Hernández-Cárdenas G, editors. Atlas de línea base ambiental del Golfo de México. Consorcio de Investigación del Golfo de México; 2021.

[ref26] 26. Koleff P, Tambutti M, March IJ, Esquivel R, Cantú C, Lira-Noriega A. Identificación de prioridades y análisis de vacíos y omisiones en la conservación de la biodiversidad de México. In: Sarukhán J, editor. Capital natural de México, vol. II: Estado de conservación y tendencias de cambio. México: CONABIO; 2009. pp. 651–718.

[ref27] 27. Instituto Nacional de Estadística y Geografía (INEGI). Censo de Población y Vivienda 2020. [cited 2022 Jun 9] Available from: https://www.inegi.org.mx/programas/ccpv/2020/
View Article
Google Scholar

[82] View Article

[83] Google Scholar

[ref28] 28. R Core Team. R: a language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2023. Available from: https://www.R-project.org/
View Article
Google Scholar

[85] View Article

[86] Google Scholar

[ref29] 29. Esri. ArcGIS Pro (Versión 3.2). 2025. Available from: https://www.esri.com/en-us/arcgis/products/arcgis-pro/overview
View Article
Google Scholar

[88] View Article

[89] Google Scholar

[ref30] 30. Oksanen J, Simpson G, Blanchet F, Kindt R, Legendre P, Minchin P, et al. vegan: community ecology package. R package version 2.6-4. 2022. Available from: https://CRAN.R-project.org/package=vegan
View Article
Google Scholar

[91] View Article

[92] Google Scholar

[ref31] 31. Dray S, Dufour A-B. Theade4Package: Implementing the Duality Diagram for Ecologists. J Stat Soft. 2007;22(4).
View Article
Google Scholar

[94] View Article

[95] Google Scholar

[ref32] 32. Charrad M, Ghazzali N, Boiteau V, Niknafs A. NbClust: AnRPackage for Determining the Relevant Number of Clusters in a Data Set. J Stat Soft. 2014;61(6).
View Article
Google Scholar

[97] View Article

[98] Google Scholar

[ref33] 33. Trachsel J. funfuns: functions I use (title case). R package version 0.1.2. 2023. Available from: https://rdrr.io/github/Jtrachsel/funfuns/
View Article
Google Scholar

[100] View Article

[101] Google Scholar

[ref34] 34. Sjoberg D. ggsankey: sankey, alluvial and sankey bump plots. R package version 0.0.99999. 2023. Available from: https://github.com/davidsjoberg/ggsankey
View Article
Google Scholar

[103] View Article

[104] Google Scholar

[ref35] 35. McGarigal K. Multivariate stallstlcs for wildlife and ecology research. Springer; 2000.

[ref36] 36. Csárdi G, Nepusz T, Traag V, Horvát S, Zanini F, Noom D, et al. igraph: network analysis and visualization in R. R package version 1.5.1. 2023. Available from: https://CRAN.R-project.org/package=igraph
View Article
Google Scholar

[107] View Article

[108] Google Scholar

[ref37] 37. Blondel VD, Guillaume J-L, Lambiotte R, Lefebvre E. Fast unfolding of communities in large networks. J Stat Mech. 2008;2008(10):P10008.
View Article
Google Scholar

[110] View Article

[111] Google Scholar

[ref38] 38. Galili T. dendextend: an R package for visualizing, adjusting and comparing trees of hierarchical clustering. Bioinformatics. 2015;31(22):3718–20. pmid:26209431
View Article
PubMed/NCBI
Google Scholar

[113] View Article

[114] PubMed/NCBI

[115] Google Scholar

[ref39] 39. Costa LDF, Rodrigues FA, Travieso G, Villas Boas PR. Characterization of complex networks: a survey of measurements. Advances in Physics. 2007;56(1):167–242.
View Article
Google Scholar

[117] View Article

[118] Google Scholar

[ref40] 40. Hobday AJ, Smith ADM, Stobutzki IC, Bulman C, Daley R, Dambacher JM, et al. Ecological risk assessment for the effects of fishing. Fish Res. 2011;108(2–3):372–84.
View Article
Google Scholar

[120] View Article

[121] Google Scholar

[ref41] 41. Holsman K, Samhouri J, Cook G, Hazen E, Olsen E, Dillard M, et al. An ecosystem-based approach to marine risk assessment. Ecosyst Health Sustain. 2017;3(1).
View Article
Google Scholar

[123] View Article

[124] Google Scholar

[ref42] 42. Sijing X, Gang L, Biao M. Vulnerability analysis of land ecosystem considering ecological cost and value: a complex network approach. Ecol Indic. 2023;147:109941.
View Article
Google Scholar

[126] View Article

[127] Google Scholar

[ref43] 43. Abson DJ, Dougill AJ, Stringer LC. Using principal component analysis for information-rich socio-ecological vulnerability mapping in Southern Africa. Appl Geogr. 2012;35(1–2):515–24.
View Article
Google Scholar

[129] View Article

[130] Google Scholar

[ref44] 44. Mohan PS. Disasters, disaster preparedness and post disaster recovery: evidence from Caribbean firms. Int J Disas Risk Reduc. 2023;92:103731.
View Article
Google Scholar

[132] View Article

[133] Google Scholar

[ref45] 45. Owolabi TA, Sajjad M. A global outlook on multi-hazard risk analysis: a systematic and scientometric review. Int J Disas Risk Reduc. 2023;92:103727.
View Article
Google Scholar

[135] View Article

[136] Google Scholar

[ref46] 46. Gallina V, Torresan S, Critto A, Sperotto A, Glade T, Marcomini A. A review of multi-risk methodologies for natural hazards: consequences and challenges for a climate change impact assessment. J Environ Manage. 2016;168:123–32. pmid:26704454
View Article
PubMed/NCBI
Google Scholar

[138] View Article

[139] PubMed/NCBI

[140] Google Scholar

[ref47] 47. Yang W, Dun X, Jiang X, Zhou Y, Hou B, Lang R, et al. An integrated risk assessment framework for multiple natural disasters based on multi-dimensional correlation analysis. Nat Hazards. 2023;119(3):1531–50.
View Article
Google Scholar

[142] View Article

[143] Google Scholar

[ref48] 48. Eakin H, Luers AL. Assessing the vulnerability of social-environmental systems. Annu Rev Environ Resour. 2006;31(1):365–94.
View Article
Google Scholar

[145] View Article

[146] Google Scholar

[ref49] 49. Simeoni C, Furlan E, Pham HV, Critto A, de Juan S, Trégarot E, et al. Evaluating the combined effect of climate and anthropogenic stressors on marine coastal ecosystems: insights from a systematic review of cumulative impact assessment approaches. Sci Total Environ. 2023;861:160687. pmid:36473660
View Article
PubMed/NCBI
Google Scholar

[148] View Article

[149] PubMed/NCBI

[150] Google Scholar

[ref50] 50. Lauerburg RAM, Diekmann R, Blanz B, Gee K, Held H, Kannen A, et al. Socio-ecological vulnerability to tipping points: a review of empirical approaches and their use for marine management. Sci Total Environ. 2020;705:135838. pmid:31855803
View Article
PubMed/NCBI
Google Scholar

[152] View Article

[153] PubMed/NCBI

[154] Google Scholar

[ref51] 51. Chernick MR. Resampling methods. WIREs Data Min Knowl Discov. 2012;2(3):255–62.
View Article
Google Scholar

[156] View Article

[157] Google Scholar

[ref52] 52. Godwyn-Paulson P, Jonathan MP, Rodríguez-Espinosa PF, Abdul Rahaman S, Roy PD, Muthusankar G, et al. Multi-hazard risk assessment of coastal municipalities of Oaxaca, Southwestern Mexico: an index based remote sensing and geospatial technique. Int J Disas Risk Reduct. 2022;77:103041.
View Article
Google Scholar

[159] View Article

[160] Google Scholar

Figures

Abstract

Author summary

Introduction

Materials and methods

Study area

Spatial characterization of the Mexican Gulf of Mexico

Selection of the databases

Data standardization

Identification and characterization of variable groups in the Mexican Gulf of Mexico

Influence of variables in the Mexican Gulf of Mexico

Relative importance of variables.

Variables’ system-wide synergistic or antagonistic influence.

Clustering of antagonistic and synergistic variables.

Results

Identification of regions within zones

The relative importance of variables

Oceanic zone.

Coastal zone.

Inland zone.

Insular zone.

System-wide synergistic or antagonistic influence of variables

Inland zone.

Oceanic zone.

Coastal zone.

Insular zone.

Identification of synergistic and antagonistic groups of variables

Discussion

Recommendations

Supporting information

S1 Fig. Zonification of the exclusive economic zone of Mexico in the Gulf of Mexico.

S2 Fig. Cell clustering in the PCA of zones in the exclusive economic zone of Mexico in the Gulf of Mexico.

S3 Fig. Sensitivity analyses of each variable over the others in four zones of the exclusive economic zone of Mexico in the Gulf of Mexico.

S4 Fig. Average node degree (number of connections in nodes) of the synergistic and antagonistic networks within the zones and regions of the exclusive economic zone of Mexico in the Gulf of Mexico.

S1 Table. Databases used in the sensitivity characterization of the exclusive economic zone of Mexico in the Gulf of Mexico.

S2 Table. Main data processes applied to each variable used in Gulf of Mexico environmental analysis.

S3 Table. Explained variance (%) of the first five PCA axes in each zone in the exclusive economic zone of Mexico in the Gulf of Mexico.

S4 Table. Significance test for similitude analysis (ANOSIM) of the four zones in the exclusive economic zone of Mexico in the Gulf of Mexico. a): Permutation significance tests; b): pairwise comparisons of regions within zones.

S5 Table. Number of groups in the synergistic and antagonistic networks in the zones and regions of the exclusive economic zone of Mexico in the Gulf of Mexico.

References