DOI: 10.20937/ATM.53579
Received: February 5, 2026; Accepted: July 30, 2026
Norma Elizabeth Olvera-Fuentes
Instituto de Ingeniería, Universidad Nacional Autónoma de México, Circuito de la Investigación Científica s/n, Ciudad Universitaria, 04510 Ciudad de México, México.
Corresponding author; email: norma.olvera@atmosfera.unam.mx
Fernando Jorge González-Villarreal
Instituto de Ingeniería, Universidad Nacional Autónoma de México, Circuito de la Investigación Científica s/n, Ciudad Universitaria, 04510 Ciudad de México, México.
Carlos Gay-García
Instituto de Ciencias de la Atmósfera y Cambio Climático, Universidad Nacional Autónoma de México, Circuito de la Investigación Científica s/n, Ciudad Universitaria, 04510 Ciudad de México, México.
RESUMEN
El Golfo de California (GC) es un sistema socioecológico expuesto a factores de estrés climáticos y antrópicos interrelacionados. Comprender cómo se propagan estas presiones es fundamental para evaluar la vulnerabilidad y apoyar la gestión ambiental en condiciones de incertidumbre. En este estudio, desarrollamos un mapa cognitivo difuso (MCD) para evaluar la influencia del cambio climático y las actividades humanas, así como para diagnosticar las respuestas del sistema a nivel sistémico. El modelo se construyó a partir de la literatura científica revisada por pares y fue refinado mediante la clasificación y ponderación de las relaciones causales con base en dicha literatura. Se realizaron simulaciones en un laboratorio virtual para explorar el comportamiento del sistema en dos escenarios complementarios: (i) un escenario base que excluye los efectos del cambio climático, utilizado para aislar la influencia de las presiones antropogénicas, y (ii) un escenario con cambio climático. Adicionalmente, se llevaron a cabo simulaciones con pesos mínimos y máximos para evaluar la robustez del modelo. El marco metodológico del MCD integró variables ambientales, socioeconómicas y climáticas dentro de una estructura capaz de representar efectos indirectos, relaciones no lineales e interacciones en cascada. Los resultados indican que el incremento de la temperatura, la presión pesquera y la contaminación costera actúan como factores de estrés dominantes que intensifican la degradación del ecosistema. El deterioro simultáneo de los distintos componentes del sistema refleja una alta interconectividad y una resiliencia limitada. El cambio climático amplifica los mecanismos de retroalimentación y acelera las trayectorias hacia consecuencias ecológicas y económicas más severas. Los análisis de sensibilidad indican que los principales patrones de vulnerabilidad permanecen consistentes a lo largo de un amplio rango de valores de los pesos. Este estudio proporciona un diagnóstico del GC bajo presiones climáticas y humanas, destacando las vías por las que se configura la vulnerabilidad. El enfoque basado en MCD ofrece elementos útiles para identificar factores de estrés clave y posibles puntos de intervención, además de proporcionar un marco metodológico transferible a otros sistemas socioecológicos expuestos a presiones climáticas y antropogénicas interrelacionadas.
ABSTRACT
The Gulf of California (GC) is a socio-ecological system exposed to interrelated climate-driven and anthropogenic stressors. Understanding how these pressures propagate is essential for assessing vulnerability and informing environmental management under uncertainty. In this study, we developed a fuzzy cognitive map (FCM) to evaluate the influence of climate change and human activities and to diagnose system-level responses. The model was derived from peer-reviewed literature and refined through literature-based classification and weighting of causal relationships. Simulations were conducted in a virtual laboratory to explore system behavior under two complementary scenarios: (i) a baseline scenario excluding climate change effects, used to isolate the influence of anthropogenic pressures, and (ii) a climate change scenario. Additional minimum- and maximum-weight simulations were performed to evaluate model robustness. The FCM framework integrated environmental, socio-economic, and climatic variables within a structure capable of representing indirect effects, nonlinear relationships, and cascading interactions. Results indicate that rising temperatures, fishing pressure, and coastal pollution act as dominant stressors that intensify ecosystem degradation. Simultaneous deterioration across system components reflects high interconnectedness and limited resilience. Climate change amplifies feedback mechanisms, accelerating trajectories toward more severe ecological and economic outcomes. Sensitivity analyses indicate that the main vulnerability patterns remain consistent across a broad range of weight values. This study provides a diagnosis of the GC under climate and human pressures, highlighting pathways that shape vulnerability. The FCM approach offers insights for identifying key stressors and intervention points and provides a transferable framework for socio-ecological systems facing interrelated climate-driven and anthropogenic pressures.
Keywords: decision-support tools, ecosystem vulnerability, environmental management, scenario analysis, socio-ecological systems.
1. Introduction
The Gulf of California (GC) is a semi-enclosed sea located between 24º and 32º N along northwestern Mexico, extending 1126 km and reaching depths of up to 3000 m, with a surface area of approximately 177 000 km2 (Nix, 2013). Its geological formation, hydrodynamics, and geographic location have generated exceptional biological richness, resulting in highly productive subtropical marine and coastal ecosystems (Nix, 2013; Páez-Osuna et al., 2017). This high productivity, combined with strong environmental gradients, makes the Gulf of California a highly sensitive socio-ecological system in which small perturbations can generate system-wide consequences. This geographical context is illustrated in Figure 1.

Fig. 1. Geographic location of the Gulf of California and its surrounding coastal region in northwestern Mexico.
The ecological relevance of the GC also implies significant challenges for environmental management. Because of its biological richness, the region has been recognized as a Natural World Heritage Site (CONANP, 2005) and described by Jacques Cousteau as the “aquarium of the world.” More than 6000 marine species have been recorded, including a high proportion of macroinvertebrates, fish, marine mammals, and endemic fauna (Brusca, 2010). Key habitats such as estuaries, wetlands, coastal lagoons, and mangroves support regional productivity and ecosystem stability (González-Villarreal et al., 2024). The coexistence of exceptional biodiversity and intensive human use increases the complexity of conservation and resource management, requiring integrated approaches capable of addressing trade-offs between ecological integrity and socioeconomic activities.
The GC sustains intense human activity and plays a strategic role in the Mexican economy. The region contributes approximately 71.2% of national fishery production, 70% of aquaculture value, 40% of agricultural output, and 27% of mining production (CONAPESCA, 2022; CONAGUA, 2019; SGM, 2021). Tourism further reinforces its socioeconomic importance, attracting nearly five million visitors annually. The concentration of productive activities across sectors generates multiple interacting pressures that operate across spatial and temporal scales, posing substantial challenges for environmental management.
Increasing anthropogenic pressures have led to widespread socio-environmental degradation in the GC and its surrounding ecoregion. Major impacts include groundwater contamination from agrochemical leaching, discharge of pollutants into coastal waters, sedimentation associated with deforestation, and hydrological alterations caused by dam construction (González-Villarreal et al., 2024). Freshwater inflow to the GC has declined by approximately 34% relative to natural levels, increasing salinity and concentrating pollutants (González-Villarreal and Alaniz-Ceja, 2022). Additional stressors include industrial pollution, mining waste, fuel spills, and overfishing driven by inadequate regulatory frameworks (Páez-Osuna et al., 2017; González-Villarreal et al., 2024). These pressures do not act independently; rather, they generate indirect and cascading impacts that accelerate system-level degradation.
A comprehensive review of 150 studies by Páez-Osuna et al. (2017) documented the presence of metals, metalloids, persistent organic pollutants, nutrients, plastics, and harmful algal blooms throughout the region. Despite its ecological and economic relevance, the GC remains comparatively under-studied: only 4408 scientific publications were recorded between 1970 and 2017, far fewer than for other marine systems such as the Mediterranean Sea (38 578), the Gulf of Mexico (24 034), or the Baltic Sea (16 377). This fragmented knowledge base limits the capacity for integrated environmental management and highlights the need for system-level diagnostic tools capable of synthesizing complex interactions.
Climate change further compounds these pressures. According to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC, 2021), global mean surface temperature increased by approximately 1.1 ºC between 2011 and 2020 relative to pre-industrial levels. In Mexico, warming has been more intense and rapid, with an average increase of 1.69 ºC since the early twentieth century and a rate of 2.88 ºC per century, substantially higher than the global average (Estrada et al., 2023). Projections also indicate reduced precipitation and increased climate variability in northern Mexico. Climate change thus acts as an amplifying driver that intensifies existing stressors, reduces system resilience, and increases vulnerability across ecological, economic, and social dimensions of the GC.
Although varied in nature, the pressures affecting the GC are deeply interconnected. Increased nutrient and contaminant discharges promote harmful algal blooms, while hydrological alteration, resource overexploitation, and rising temperatures generate nonlinear feedbacks and cascading effects. Together, these interactions characterize the GC as a complex socio-ecological system whose behavior cannot be understood through isolated analysis of individual components. Understanding such systems requires integrated, system-level assessment approaches capable of capturing indirect effects, feedback mechanisms, and emergent dynamics (Meyers, 2009; Mitchell, 2009; Simon, 2012; Sayama, 2015; Lara-Rosano et al., 2021).
Given this complexity, modeling approaches that integrate heterogeneous variables under conditions of uncertainty and limited data are essential. Fuzzy cognitive maps (FCMs) provide a suitable framework for representing causal relationships among environmental, climatic, and socio-economic factors, capturing indirect effects and simulating system behavior (Groumpos, 2010). Within an environmental management context, FCMs serve as diagnostic and decision-support tools rather than purely technical models. Previous applications include coastal vulnerability assessments (Gray et al., 2014), analyses of stakeholder perceptions of marine risks (Kontogianni et al., 2012), and climate-related impacts on socio-environmental systems (Olvera-Fuentes and Gay-García, 2023).
However, to our knowledge, no previous FCM-based study has provided a system-level diagnosis of the Gulf of California that explicitly integrates persistent anthropogenic pressures with climate change to examine how vulnerability propagates through the system. To support this assessment, a literature search was conducted in November 2025 using the Scopus, Web of Science, Google Scholar, and ScienceDirect databases. The search included combinations of the keywords “Gulf of California” AND “fuzzy cognitive map”, “Sea of Cortez” AND “fuzzy cognitive map”, “Gulf of California” AND “fuzzy systems”, and “fuzzy cognitive mapping” AND “marine management”. No previous studies applying fuzzy cognitive maps to analyze the combined effects of climate and anthropogenic pressures in the Gulf of California were identified. Moreover, few studies have explored how interacting pressures evolve and amplify under contrasting climate conditions. This research addresses this gap by developing an integrated modeling framework to assess how current pressures interact, propagate, and shape system-level outcomes in the GC. Although focused on a specific region, the approach is transferable to other socio-ecological systems facing similar combinations of climate-driven and anthropogenic pressures, offering management-relevant insights beyond the case study.
This paper adopts a multi-factorial and system-level approach to analyze the combined effects of anthropogenic pressures and climate change on socio-ecological vulnerability in the Gulf of California. The remainder of the paper is structured as follows. Section 2 describes the methodological framework, including the construction and dynamic simulation of the fuzzy cognitive map. Section 3 presents the results obtained under contrasting scenarios, highlighting system responses and sensitivity to key drivers. Section 4 discusses the implications of these findings for vulnerability diagnosis and environmental management. Finally, Section 5 summarizes the main conclusions and outlines directions for future research and decision-support applications.
2. Methods
FCMs provide a modeling framework for representing complex systems through networks of causal relationships among interdependent variables (Stylios, 2000; Stylios and Groumpos, 2004; Dodouras and James, 2007; Papageorgiou and Stylios, 2008; Groumpos and Stylios, 2000; Groumpos, 2017; Nair et al., 2019; Rickard et al., 2020). Rather than isolating individual components, FCMs emphasize the collective behavior that emerges from the interaction of multiple factors, allowing system dynamics to be examined from an integrated perspective.
A key feature of FCMs is their graphical representation, which offers a synthetic and intuitive view of cause-and-effect relationships among diverse system components (Dickerson and Kosko, 1994), including environmental processes, socio-economic activities, and policy-related factors. This structure facilitates identifying dominant pathways, feedbacks, and indirect effects that shape overall system behavior, which is essential for analyzing complex socio-environmental systems.
Complex systems are characterized by emergent properties arising from nonlinear interactions among their components, making their behavior difficult to understand through reductionist approaches (Meyers, 2009; Mitchell, 2009; Simon, 2012; Sayama, 2015; Lara-Rosano et al., 2021). FCMs are particularly well suited to address this challenge, as they explicitly represent feedback mechanisms and cascading interactions that drive system-level responses (Koulouriotis et al., 2003).
An additional strength of FCMs lies in their capacity to integrate both quantitative and qualitative information within a single modeling framework (Kosko, 1986, 1996; Ross, 2010). This soft-computing nature enables the incorporation of expert knowledge and evidence derived from the literature, making FCMs especially useful in contexts where data availability is limited or uncertainty is high (Kosko, 1986, 1996; Koulouriotis et al., 2003; Ross, 2010).
Within environmental management research, FCMs have increasingly been applied as diagnostic and decision-support tools, supporting the assessment of vulnerability, the identification of leverage points, and the exploration of system responses under contrasting conditions (Groumpos, 2017; Rickard et al., 2020; Tatarkanov et al., 2022). Given the high interconnectedness of climate-driven and anthropogenic pressures in the Gulf of California, the FCM approach was selected in this study to support a system-level assessment focused on diagnosing how pressures propagate and interact, rather than predicting precise quantitative outcomes.
Conceptually, an FCM represents a system as a network of interconnected concepts. Each concept has an activation level that evolves through time according to the influence exerted by other concepts. These influences are represented by weighted causal relationships, and the system state is updated iteratively until a stable configuration is reached.
Formally, an FCM can be represented as a four-tuple (C, W, A, f) following Félix et al. (2019). The set C = {C1, C2,…, CN} denotes the N concepts (nodes) of the system, each representing a relevant environmental, socio-economic, or climate variable.
Causal relationships among concepts are encoded in the weighted adjacency matrix W: C × C → [–1,1], whose elements wij ∈ [–1,1] quantify the direction and intensity of the influence exerted by concept Ci on concept Cj (Zadeh, 1965). A positive causal relationship (wij > 0), denoted as Ci →+ Cj, indicates a reinforcing interaction, meaning that an increase in Ci leads to an increase in Cj. Conversely, a negative causal relationship (wij < 0), denoted as Ci →– Cj, represents an inhibitory interaction, implying that an increase in Ci results in a decrease in Cj. When wij = 0, no direct causal link is assumed between the two concepts.
The dynamic state of the system is described by the activation vector A(t) = [A1(t), A2(t),…, AN(t)], where Ai(t) ∈ R represents the activation level of concept Ci at discrete time step t. System evolution is governed by an iterative update process that aggregates the combined influence of incoming causal connections on each concept.
In this study, system dynamics were simulated using Kosko’s activation rule (Kosko, 1986; Dickerson and Kosko, 1994), which updates the activation of each concept according to:
(1)
Although alternative inference rules have been proposed in recent years (Schuerkamp and Giabbanelli, 2023), Kosko’s rule remains widely adopted due to its conceptual simplicity, robustness, and compatibility with previous FCM-base studies (Papageorgiou et al., 2006; Papageorgiou and Salmerón, 2013), thereby facilitating comparative analysis.
The transfer function f: R → I constrains activation values to a predefined interval I, commonly I = [0, 1] or I = [–1, 1], depending on the modeling context (Félix et al., 2019). In this application, a sigmoid function was employed:
(2)
with parameters h = 0 and β = 1. This configuration provides a smooth and symmetric transformation that is well suited for exploratory, system-level analysis, ensuring numerical stability while preserving sensitivity to changes in causal influence.
All simulations were implemented in a virtual laboratory developed in Python, enabling reproducible evaluation of system dynamics under contrasting scenarios.
The FCM representing the Gulf of California and its surrounding ecoregion was constructed through an exhaustive review of peer-reviewed literature documenting the impacts of climate change and anthropogenic activities on the region’s environmental, economic, and social dimensions.
Concepts and causal relationships were identified following established FCM construction approaches based on document analysis and expert-informed synthesis (Özesmi and Özesmi, 2004). While causal weights in FCMs are often derived exclusively through expert elicitation, in this study the initial causal structure was grounded in documented evidence reported in the literature, ensuring transparency and traceability of the identified interactions.
To assign relative weights to the causal relationships, the three authors—each with extensive expertise in the Gulf of California and climate change impacts—jointly evaluated the qualitative intensity reported in the reviewed sources. This procedure follows established FCM practice, in which causal influence is commonly expressed using linguistic variables rather than exact numerical values. According to Papageorgiou and Kontogianni (2012), experts may characterize causal relationships through qualitative descriptors that are subsequently transformed into numerical weights for simulation purposes. This approach is particularly suitable for complex socio-environmental systems, where many interactions are supported by qualitative or semi-quantitative evidence rather than by direct measurements.
Accordingly, the causal relationships identified in the literature were classified according to three linguistic categories describing their relative strength: low (L), medium (M), and high (H). These categories were represented by the values 0.3, 0.6, and 0.9, respectively. The selected values preserve equal intervals between categories and maintain the ordinal ranking H > M > L required to represent relative causal strength. This transformation did not aim to estimate exact causal strengths, but to preserve the evidence’s ordinal structure by distinguishing weak, intermediate, and strong influences in a transparent and reproducible manner. Similar linguistic-to-numerical transformations have been widely used in FCM applications to convert qualitative knowledge into computationally tractable models (Papageorgiou and Kontogianni, 2012).
The resulting weights therefore represent relative influence levels rather than direct causal measurements. This distinction is important because the purpose of the model is to explore system behavior, identify dominant pathways of impact propagation, and diagnose patterns of socio-ecological vulnerability, rather than estimate precise quantitative responses. By explicitly documenting the concepts, causal relationships, linguistic classifications, and adjacency matrix, the proposed weighting procedure remains transparent and reproducible for future applications.
The reproducibility of the weighting procedure does not depend on the exact numerical values assigned by the authors, but on the consistent application of the classification scheme. Future studies can replicate the process by identifying causal relationships, assigning them to the linguistic categories low, medium, or high based on available evidence, and subsequently transforming these categories into representative numerical values. Because the classification criteria, adjacency matrix, and modeling assumptions are explicitly documented, the proposed framework remains transferable across applications. Based on the literature synthesis, issues of high relevance and systemic impact were selected and grouped into conceptual categories representing climatic drivers, anthropogenic pressures, and socio-environmental outcomes. Within the model, productive activities are primarily represented through the concepts of pollutant discharges (C6), overfishing (C11), and population growth and unsustainable economic development (C14), which act as the main anthropogenic pressures affecting the Gulf of California socio-environmental system. The resulting FCM comprises 14 concepts (C1-C14), whose definitions and thematic classification are summarized in Table I.
Table I. Concepts included in the fuzzy cognitive map of the Gulf of California.
| Concept ID | Concept description | Category |
| C1 | Increase in temperature | Climate driver |
| C2 | Decrease in precipitation | Climate driver |
| C3 | Increase in evapotranspiration | Climate driver |
| C4 | Increase in extreme climatic events | Climate driver |
| C5 | Degradation and pollution of the Gulf of California | Socio-environmental outcome |
| C6 | Discharge of pollutants | Anthropogenic Pressure |
| C7 | Economic losses | Socio-economic outcome |
| C8 | Deterioration of population health and quality of life | Socio-environmental outcome |
| C9 | Loss of wildlife and marine life | Environmental impact |
| C10 | Loss of terrestrial and marine habitats | Environmental impact |
| C11 | Overfishing | Anthropogenic pressure |
| C12 | Algal proliferation | Environmental impact |
| C13 | Reduction of river flows | Environmental impact |
| C14 | Population growth and unsustainable economic development | Anthropogenic pressure |
In total, 75 positive causal relationships were identified and incorporated into the model, forming the basis for the construction of the weighted adjacency matrix used in subsequent simulations. This causal structure defines the pathways through which pressures propagate across the system and supports the system-level diagnostic analysis presented in the results section.
For modeling purposes, three anthropogenic pressures—discharge of pollutants (C6), overfishing (C11), and population growth and unsustainable economic development (C14)—were treated as persistent external drivers and fixed at their maximum activation level (1) across all simulated scenarios. This configuration represents baseline pressure conditions in the region and allows the assessment of how additional stressors, such as climate change, amplify system-wide impacts. The graphical representation of the FCM was generated using Gephi software (https:////gephi.org) to visualize network structure, link direction, and relative weight magnitude.
To evaluate how climate-driven and anthropogenic pressures propagate through the system, a set of contrasting scenarios was defined and analyzed using the constructed FCM. The objective of the scenario analysis was not to predict precise quantitative outcomes, but to diagnose system-level responses, identify dominant pathways, and assess changes in vulnerability patterns under different forcing conditions. Two main scenarios were considered:
System dynamics were simulated using the iterative update rule defined in (1). At each iteration, the activation level of each concept was updated based on the weighted influence of all incoming causal connections, as defined by the adjacency matrix. Iterations were repeated until the system reached a stable configuration, defined as the point at which changes in activation values between successive iterations became negligible.
The state of the system at each iteration t is represented by a state vector:
(3)
where each element corresponds to the activation level of one concept in the FCM.
Initial activation values were defined consistently with the objectives of each scenario and the diagnostic nature of the analysis. Climate-related concepts were selectively activated or deactivated depending on the scenario, while anthropogenic drivers remained fixed as described above. This approach allowed the isolation of the incremental effect of climate change on the propagation of impacts through the system, while maintaining a realistic representation of ongoing human pressures.
All simulations were implemented in a virtual laboratory developed in Python, ensuring transparency, reproducibility, and consistency across scenarios.
Because activation values in FCMs are expressed on a continuous scale between 0 and 1, relatively small numerical differences may correspond to meaningful changes in system behavior. To support the interpretation and communication of model results, a qualitative severity scale was defined.
This scale was developed to classify activation values into five qualitative levels ranging from very low to very high/critical severity (Table II). The objective is not to modify model dynamics, but to provide an interpretative framework that facilitates comparison across scenarios, identification of dominant stressors, and assessment of the relative severity of impacts across system components.
Table II. Proposed qualitative severity scale for interpreting activation values in the fuzzy cognitive map.
| Severity range | Qualitative level |
| 0.80-1.00 | Very high/critical |
| 0.60-0.79 | High |
| 0.40-0.59 | Medium |
| 0.20-0.39 | Low |
| 0.00-0.19 | Very low/negligible |
By translating numerical activation values into qualitative categories, the severity scale supports a system-level diagnosis that is more accessible for environmental management analysis while preserving the underlying quantitative information generated by the model.
Because FCM dynamics are directly influenced by the magnitude of the causal weights assigned to relationships among concepts, a sensitivity analysis was conducted to explore the range of system responses associated with uncertainty in weight specification. This analysis aims to assess the robustness of model behavior under extreme but plausible configurations of causal influence.
Two limiting configurations of the adjacency matrix were defined based on the causal structure documented in the literature (Table III). In the minimum-weight configuration, all non-zero causal relationships were assigned a uniform value of 0.3, representing a conservative scenario in which interactions among system components are relatively weak. In contrast, the maximum-weight configuration assigns a value of 0.9 to all non-zero relationships, representing a scenario in which causal influences operate at their highest plausible intensity.
Table III. Causal relationships identified, their assigned weights, and supporting references. Destination concepts, weights, and references are listed in the same sequential order within each row; therefore, repeated references indicate that the same source supports multiple causal relationships originating from the same concept.
| Cause (concept origin) | Effect (destination concept) | Weight (influence) | Reference |
| C1 | C2; C3; C4; C8; C9; C10; C12 | 0.9; 0.9; 0.9; 0.9; 0.9; 0.9; 0.6 | Intergovernmental Panel on Climate Change (2021, 2022) |
| C2 | C3; C4; C7; C8; C9; C10; C13 | 0.9; 0.9; 0.9; 0.9; 0.9; 0.9; 0.9 | Intergovernmental Panel on Climate Change (2022) |
| C3 | C8; C9; C10; C13 | 0.9; 0.9; 0.9; 0.9 | Intergovernmental Panel on Climate Change (2022) |
| C4 | C2; C3; C7; C8; C9; C10; C13 | 0.9; 0.9; 0.9; 0.9; 0.9; 0.9; 0.9 | Intergovernmental Panel on Climate Change (2022) |
| C5 | C7; C8; C9; C10; C12 | 0.9; 0.9; 0.9; 0.9; 0.9 | Páez-Osuna et al. (2017, 2019), Cariño et al. (2020), Galindo et al. (2021), Infobae (2021), Alatriste (2024), Araiza (2025), Band-Schmidt et al. (2011) |
| C6 | C5; C8; C9; C10; C12 | 0.9; 0.9; 0.9; 0.9; 0.9 | Páez-Osuna et al. (2017, 2019), Alatriste (2024), Araiza (2025), Band-Schmidt et al. (2011) |
| C7 | C8 | 0.9 | Lustig (2007) |
| C8 | C5; C6; C7; C9; C10; C11; C14 | 0.6; 0.6; 0.9; 0.6; 0.6; 0.9; 0.6 | Blanco, 1996 |
| C9 | C5; C7; C8; C10 | 0.9; 0.9; 0.9; 0.9 | WWF (2020) |
| C10 | C5; C7; C8; C9 | 0.9; 0.9; 0.9; 0.9 | WWF (2020) |
| C11 | C5; C6; C7; C8; C9; C10 | 0.9; 0.6; 0.6; 0.6; 0.9; 0.9 | González-Villarreal et al. (2024), Cisneros-Mata et al. (1995), Blanco (1996), Cariño et al. (2020) |
| C12 | C5; C7; C8; C9; C10 | 0.9; 0.9; 0.9; 0.9; 0.9 | Band-Schmidt, et al. (2011), Arellano-Aguilar, et al. (2016) |
| C13 | C5; C7; C8; C9; C10 | 0.9, 0.9; 0.9; 0.9; 0.9 | González-Villarreal y Alaniz-Ceja (2022), Galindo et al. (2021), Arias (2005) |
| C14 | C5; C6; C7; C8; C9; C10; C11; C13 | 0.9; 0.9; 0.9; 0.9; 0.9; 0.9; 0.9; 0.9 | Blanco, 1996, Sepúlveda (1976), González-Villarreal y Alaniz-Ceja, L. (2022) |
Note: for concept descriptions (C1 to C14), see Table I.
These configurations were applied consistently across scenarios with and without climate change, while maintaining the same initial state vectors and forcing assumptions defined in section 2.4. By comparing system behavior under these extreme configurations, the analysis allows for the identification of the potential activation ranges of each concept and highlights components whose responses are most sensitive to variations in causal strength.
This sensitivity analysis does not aim to identify precise quantitative thresholds, but rather to support a qualitative assessment of uncertainty and to evaluate whether the main patterns of system vulnerability and stressor dominance remain consistent under different weight assumptions. The results of this analysis are presented and discussed in section 3.3.
3. Results
The FCM developed for the Gulf of California provides an integrated representation of the causal structure linking climatic drivers, anthropogenic pressures, and socio-environmental outcomes. The resulting network captures how multiple stressors interact and propagate across the system, revealing interdependence patterns not evident when pressures are considered in isolation.
The final FCM comprises 14 concepts connected through 75 positive causal relationships, reflecting reinforcing interactions among system components. These relationships represent documented pathways through which climate change and human activities jointly influence environmental degradation, economic performance, and human well-being in the region Table III.
Figure 2 presents the graphical representation of the FCM. Nodes correspond to the selected concepts presented in Table I, while directed links—captured in Table IV through the adjacency matrix—indicate the direction and relative strength of influence between them. For visualization purposes, causal links adopt the color of the source concept, and link thickness reflects the relative magnitude of the assigned causal weights. This representation facilitates the identification of dominant influence pathways and highlights the high degree of connectivity within the system. To further support interpretation of the causal structure and improve the model’s transparency and reproducibility, Table IV presents the complete adjacency matrix associated with the FCM. The matrix summarizes the 75 causal relationships represented in Figure 2 and provides the numerical weights assigned to each interaction between concepts. This table complements the graphical visualization by allowing a detailed examination of the direction and relative strength of all causal connections included in the model.

Fig. 2. Fuzzy cognitive map of the Gulf of California. The map includes 14 nodes representing key socio-environmental, climatic, and economic issues in the region, and 75 causal relationships derived from peer-reviewed literature. Links indicate the direction and relative weight of influence, where H denotes high-intensity relationships, and M denotes medium-intensity relationships. Link thickness reflects the relative magnitude of the assigned causal weights.
Table IV. Complete adjacency matrix of the fuzzy cognitive map developed for the Gulf of California. Matrix elements represent the direction and relative strength of causal relationships among the 14 socio-environmental and economic concepts considered in the analysis.
| C1 | C2 | C3 | C4 | C5 | C6 | C7 | C8 | C9 | C10 | C11 | C12 | C13 | C14 | |
| C1 | 0 | 0.9 | 0.9 | 0.9 | 0 | 0 | 0 | 0.9 | 0.9 | 0.9 | 0 | 0.6 | 0 | 0 |
| C2 | 0 | 0 | 0.9 | 0.9 | 0 | 0 | 0.9 | 0.9 | 0.9 | 0.9 | 0 | 0 | 0.9 | 0 |
| C3 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0.9 | 0.9 | 0.9 | 0 | 0 | 0.9 | 0 |
| C4 | 0 | 0.9 | 0.9 | 0 | 0 | 0 | 0.9 | 0.9 | 0.9 | 0.9 | 0 | 0 | 0.9 | 0 |
| C5 | 0 | 0 | 0 | 0 | 0 | 0 | 0.9 | 0.9 | 0.9 | 0.9 | 0 | 0.9 | 0 | 0 |
| C6 | 0 | 0 | 0 | 0 | 0.9 | 0 | 0 | 0.9 | 0.9 | 0.9 | 0 | 0.9 | 0 | 0 |
| C7 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0.9 | 0 | 0 | 0 | 0 | 0 | 0 |
| C8 | 0 | 0 | 0 | 0 | 0.6 | 0.6 | 0.9 | 0 | 0.6 | 0.6 | 0.9 | 0 | 0 | 0.6 |
| C9 | 0 | 0 | 0 | 0 | 0.9 | 0 | 0.9 | 0.9 | 0 | 0.9 | 0 | 0 | 0 | 0 |
| C10 | 0 | 0 | 0 | 0 | 0.9 | 0 | 0.9 | 0.9 | 0.9 | 0 | 0 | 0 | 0 | 0 |
| C11 | 0 | 0 | 0 | 0 | 0.9 | 0.6 | 0.6 | 0.6 | 0.9 | 0.9 | 0 | 0 | 0 | 0 |
| C12 | 0 | 0 | 0 | 0 | 0.9 | 0 | 0.9 | 0.9 | 0.9 | 0.9 | 0 | 0 | 0 | 0 |
| C13 | 0 | 0 | 0 | 0 | 0.9 | 0 | 0.9 | 0.9 | 0.9 | 0.9 | 0 | 0 | 0 | 0 |
| C14 | 0 | 0 | 0 | 0 | 0.9 | 0.9 | 0.9 | 0.9 | 0.9 | 0.9 | 0.9 | 0 | 0.9 | 0 |
Note: for concept descriptions (C1 to C14), see Table I.
The FCM structure shows that the impacts of productive activities and climate change are closely interconnected, establishing multiple cause-and- effect relationships across the network. For example, pollutant discharges (C6) exert a direct and strong influence on the deterioration of population health and quality of life (C8), as documented in previous studies for the region (Páez-Osuna et al., 2017).
Overall, the comprehensive FCM provides a system-level overview of the Gulf of California, illustrating how interacting pressures form a tightly coupled network in which impacts propagate rapidly across components. This structural configuration underpins the dynamic behavior explored in subsequent sections and provides the basis for analyzing how different scenarios modify the severity and distribution of system-level outcomes.
This section presents the results derived from the dynamic evaluation of the FCM of the Gulf of California under contrasting scenarios (Fig. 2). The analysis focuses on how impacts propagate through the system and how the inclusion of climate change modifies the severity and distribution of system-level outcomes.
Two scenarios were examined: (i) a baseline scenario excluding climate change effects, and (ii) a climate change scenario in which temperature increase was explicitly activated as an external driver. In both scenarios, three anthropogenic pressures—pollutant discharges (C6), overfishing (C11), and population growth coupled with unsustainable economic development (C14)—remained active at their maximum activation level (A = 1), reflecting persistent background pressures.
Because climate-related concepts (C1-C4) were used as forcing variables to define the simulation scenarios, they are not included in the graphical representation of system evolution presented in Figures 3 and 4. The analysis focuses instead on the response variables whose activation levels emerge from the propagation of impacts through the network, allowing a clearer assessment of system-level outcomes under contrasting forcing conditions.
Fig. 3. Evolution of concept activation in the scenario without climate change. Concepts C5, C9, and C10 follow overlapping trajectories, reaching values close to the maximum severity level. Concepts C8 and C7 converge toward similarly high values, while C12 and C13 stabilize at lower activation levels. The first displayed points correspond to results from the first iteration of the FCM update process rather than the initial state vector. (C5: degradation and pollution of the Gulf of California; C7: economic losses; C8: deterioration of population health and quality of life; C9: loss of wildlife and marine life; C10: loss of terrestrial and marine habitats; C12: algal proliferation; C13: reduction of river flows.)
Fig. 4. Evolution of concept activation in the climate change scenario. Concepts C5, C9, C10, C8, and C7 follow overlapping trajectories, converging toward severity levels very close to 1. Under this scenario, concepts C12 and C13 show a significant increase in activation. C12 exceeds a value of 0.9, and C13 surpasses 0.95. The first displayed points correspond to results from the first iteration of the FCM update process rather than the initial state vector. (C5: degradation and pollution of the Gulf of California; C7: economic losses; C8: deterioration of population health and quality of life; C9: loss of wildlife and marine life; C10: loss of terrestrial and marine habitats; C12: algal proliferation; C13: reduction of river flows.)
Under baseline conditions, the system evolves solely in response to anthropogenic pressures. The initial state vector was defined as:
A0 = [0,0,0,0,0,1,0,0,0,0,1,0,0,1]
In this configuration, climate-related concepts (C1-C4) were set to zero, representing the absence of climate change forcing. The system dynamics were simulated using Kosko’s iterative rule (1), converging to a stable configuration after nine iterations. The resulting steady-state vector was:
A9 = [0, 0, 0, 0, 0.998512, 1, 0.998510, 0.999393, 0.998512, 0.998512, 1, 0.857986, 0.710950, 1]
The iterative evolution of concept activation until convergence is shown in Figure 3. For visualization purposes, Figures 3 and 4 display the activation levels obtained after the first iteration of the FCM update process rather than the initial state vectors. Because all response variables reached activation levels above approximately 0.70 after the first iteration, the vertical axis was truncated and the horizontal axis adjusted to improve the visualization of the convergence dynamics and facilitate comparison among concepts.
Most concepts reached high or critical activation levels, indicating strong propagation and amplification of impacts across the system even in the absence of climate change. The most affected components—classified as critical severity—were deterioration of population health and quality of life (C8), loss of terrestrial and marine habitats (C10), loss of wildlife and marine life (C9), degradation and pollution of the Gulf of California (C5), economic losses (C7), and algal proliferation (C12).
Reduction in river flows (C13), while not reaching the critical threshold, attained a high activation level, reflecting substantial system deterioration. Overall, these results indicate that anthropogenic pressures alone are sufficient to drive the system toward a critically degraded state, with impacts rapidly spreading across environmental, economic, and social dimensions.
In the climate change scenario, temperature increase (C1) was explicitly activated as an external forcing variable, enabling the propagation of effects toward the associated climate-related concepts (C2-C4). The initial state vector for this scenario was defined as:
A0 = [1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 1]
System dynamics were simulated using Kosko’s iterative update rule (1). As in the baseline scenario, the model converged to a stable configuration after nine iterations, yielding the following final state vector:
A9 = [1, 0.839659, 0.917691, 0.839659, 0.998877, 1, 0.999752, 0.999982, 0.999956, 0.999956, 1, 0.916750, 0.962214, 1]
The temporal evolution of concept activation is shown in Figure 4, which illustrates the rapid convergence of the system toward a highly degraded state under climate change conditions. Compared with the baseline scenario, the activation of climate drivers produces a stronger propagation of impacts from the very first iteration. As a result, several response variables exhibit higher activation levels at the beginning of the simulation process. The most pronounced increases occur in algal proliferation (C12), loss of terrestrial and marine habitats (C10), and deterioration of population health and quality of life (C8), reflecting the rapid amplification of impacts generated by climate forcing. These differences do not arise from changes in the initial state vectors, which were defined independently according to the scenario design, but rather from the propagation dynamics induced by the activated climate drivers.
Under this scenario, all concepts reached activation levels classified as critical according to the qualitative severity scale (Table II). The most affected components were deterioration of population health and quality of life (C8), loss of terrestrial and marine habitats (C10), loss of wildlife and marine life (C9), economic losses (C7), degradation and pollution of the Gulf of California (C5), reduction in river flows (C13), and algal proliferation (C12).
Compared with the baseline scenario, climate change led to a marked increase in the severity of several components. In particular, economic losses (C7) and reduction in river flows (C13) shifted from high to critical severity, indicating a clear amplifying effect of climate change on system-level impacts.
Across both scenarios, the deterioration of population health and quality of life (C8) emerges as the most affected concept. The activation of C8 converges toward values close to 1 under both baseline and climate change conditions. The difference between scenarios is minimal at equilibrium (0.0607% without climate change and 0.0018% with climate change), although the climate change scenario exhibits a higher initial activation during the first iteration. This indicates that climate change accelerates the onset of impacts, while the long-term severity remains critically high in both cases.
A similar convergence is observed for loss of terrestrial and marine habitats (C10), loss of wildlife and marine life (C9), economic losses (C7), and degradation and pollution of the Gulf of California (C5). In these components, the final activation levels are nearly identical across scenarios, suggesting that strong anthropogenic pressures alone are sufficient to drive these outcomes toward critical severity, with climate change exerting a limited influence on their asymptotic states.
In contrast, marked differences between scenarios are observed for river flow reduction (C13) and algal proliferation (C12). Under climate change conditions, algal proliferation increases by 6.85% relative to the baseline scenario, exceeding an activation value of 0.9. The strongest amplification is observed for river flow reduction (C13), which exhibits a 35.34% increase in its final activation level when climate change is included, rising from approximately 0.7 to nearly 0.95.
While several components converge toward critical activation levels regardless of scenario, climate change acts as an amplifying driver for specific hydrological and ecological processes, intensifying existing pressures and exacerbating the overall socio-environmental and economic deterioration of the Gulf of California.
FCM dynamics are inherently influenced by the magnitude of the weights assigned to causal relationships. To assess the sensitivity of the system to uncertainty in weight specification, limit cases were analyzed by examining the minimum and maximum activation levels that system concepts may reach, based on the causal structure reported in Table III.
Two extreme configurations of the adjacency matrix were evaluated under both baseline and climate change scenarios. In the minimum configuration, all non-zero causal relationships were assigned a uniform weight of 0.3, while in the maximum configuration all non-zero weights were set to 0.9. These values correspond to the lower and upper bounds of the linguistic weighting scheme adopted in this study, where causal relationships were classified as low (0.3), medium (0.6), or high (0.9), following established FCM practice based on the transformation of qualitative assessments into representative numerical values (Papageorgiou and Kontogianni, 2012). The value of 0.3 was therefore selected as the representative lower-bound weight for weak causal influences, rather than as an arbitrary numerical threshold. This approach allows the identification of activation ranges and provides insight into the robustness of system-level outcomes under uncertainty.
For both limit cases, the system was initialized using the same state vector as in the baseline scenario:
A0 = [0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 1]
Climate-related concepts (C1-C4) remained inactive, representing the absence of climate change forcing. The system converged to a steady state after nine iterations under both configurations.
Under the minimum-weight configuration, the final activation vector was:
Amin9 = [0, 0, 0, 0, 0.887936, 1, 0.884547, 0.911187, 0.887936, 0.887936, 1, 0.637928, 0.574442, 1]
For the maximum-weight configuration, the final activation vector was:
Amax9 = [0, 0, 0, 0, 0.998898, 1, 0.998897, 0.999551, 0.998898, 0.998898, 1, 0.858028, 0.7109495, 1]
Figure 5 illustrates the evolution of concept activation under both configurations. For visualization purposes, Figures 5 and 6 display activation levels obtained after the first iteration of the FCM update process rather than the initial state vectors. Although the minimum- and maximum-weight simulations share identical initial conditions, differences in causal weight magnitude produce distinct propagation dynamics from the first iteration onward. To improve the visualization of these differences and facilitate comparison among concepts, the vertical axis was truncated and the horizontal axis slightly adjusted.

Fig. 5. Minimum and maximum activation values in the scenario without climate change. The graph compares the evolution of concept activation using minimum (all interaction weights set to L [low-intensity relationships]) and maximum (all weights set to H [high-intensity relationships]) input conditions. In the minimum case, concepts C5, C9, and C10 converge toward activation values approaching 0.89. In the maximum case, concepts C5, C9, and C10 converge toward values very close to 1.0, while C7 and C8 reach similarly high activation levels, resulting in nearly overlapping trajectories. (C5: degradation and pollution of the Gulf of California; C7: economic losses; C8: deterioration of population health and quality of life; C9: loss of wildlife and marine life; C10: loss of terrestrial and marine habitats; C12: algal proliferation; C13: reduction of river flows.)
The comparison reveals that algal proliferation (C12) exhibits the largest relative increase between minimum and maximum cases (34.5%), indicating high sensitivity to interaction strength. This is followed by a reduction in river flows (C13) with an increase of 23.77%. Loss of wildlife and marine life (C9), degradation and pollution of the Gulf of California (C5), and loss of habitats (C10) show increases of approximately 12.5%.
Economic losses (C7) increase by 12.93%, while deterioration of population health and quality of life (C8) shows a smaller but still notable increase of 9.7%. These results highlight which system components are most sensitive to uncertainty in causal influence under baseline conditions.
When climate change was incorporated by activating temperature increase (C1) as a forcing variable, the same initial state vector was adopted:
A0 = [1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 1]
As in the baseline analysis, two extreme configurations of the adjacency matrix were evaluated. In the minimum-weight case, all non-zero causal relationships were assigned a value of 0.3, whereas in the maximum-weight case they were assigned a value of 0.9.
The system again converged to stability after nine iterations. The final activation vectors were:
Amin9 = [1, 0.619100, 0.66183, 0.619100, 0.898881, 1, 0.925943, 0.964867, 0.954272, 0.954272, 1, 0.704676, 0.704748, 1]
Amax9 = [1, 0.839656, 0.917691, 0.839659, 0.999183, 1, 0.999819, 0.999987, 0.999968, 0.999968, 1, 0.936983, 0.962214, 1]
Figure 6 shows the evolution of concept activation under both configurations.

Fig. 6. Minimum and maximum activation values in the climate change scenario. The graph compares the evolution of concept activation using minimum (all interaction weights set to L [low-intensity relationships]) and maximum (all weights set to H [high-intensity relationships]) input conditions. In the minimum case, concepts C9 and C10 reach values around 0.955, resulting in overlapping curves. In the maximum case, concepts C5, C7, C8, C9, and C10 converge toward activation values near 1.0, with C9 and C10 exhibiting overlapping trajectories and the remaining concepts reaching similarly high activation levels. (C5: degradation and pollution of the Gulf of California; C7: economic losses; C8: deterioration of population health and quality of life; C9: loss of wildlife and marine life; C10: loss of terrestrial and marine habitats; C12: algal proliferation; C13: reduction of river flows.)
Under climate change conditions, reduction in river flows (C13) exhibited the highest relative increase between minimum and maximum cases (36.53%), followed by algal proliferation (C12) with 32.97%. Degradation and pollution of the Gulf of California (C5) showed an intermediate increase (11.16%), while economic losses (C7), loss of wildlife and marine life (C9), habitat loss (C10), and deterioration of population health and quality of life (C8) displayed more moderate sensitivity.
Overall, the limiting-case analysis demonstrates that climate change elevates activation levels across the system and increases the sensitivity of certain components, particularly hydrological and biogeochemical processes, to variations in causal weight magnitude. Nevertheless, the comparison between minimum and maximum weight configurations reveals a relatively narrow range of variation for the most critical system components, indicating that the principal vulnerability patterns remain robust despite uncertainty in weight specification. In the scenario without climate change, concepts C8, C10, C9, C5, and C7—representing deterioration of population health and quality of life, loss of terrestrial and marine habitats, loss of wildlife and marine life, degradation and pollution of the Gulf of California, and economic losses, respectively—show percentage differences between their minimum and maximum activation values of only 9.70, 12.50, 12.50, 12.50, and 12.93, respectively. This limited variability indicates a high level of model consistency, even under substantial changes in causal weight assignment.
Under climate change conditions, the range of variation becomes even narrower for these same concepts, with differences of 3.64, 4.79, 4.79, 11.16, and 7.98%, respectively. These results further confirm the robustness of the system response, suggesting that the critical outcomes identified are not sensitive to moderate uncertainty in interaction strength.
In contrast, algal proliferation (C12) and river flow reduction (C13) exhibit substantially larger ranges of variation under both scenarios, indicating higher sensitivity to causal weight specification. Nevertheless, for the core socio-environmental and economic outcomes, the consistently narrow activation ranges imply that alternative plausible configurations of causal relationships would not shift the system away from the critical vulnerability regime identified in this study.
Despite the differences observed between the minimum- and maximum-weight configurations, the overall pattern of system behavior remains consistent across both simulations. The same concepts emerge as the most affected components, and the system converges toward a highly vulnerable state under both assumptions. This result suggests that the principal conclusions of the study do not depend critically on the precise numerical values assigned to individual causal relationships. Rather, they are primarily driven by the underlying causal structure of the system, providing additional support for the robustness and reproducibility of the proposed framework.
4. Discussion
The system-level analysis conducted through fuzzy cognitive maps provides an integrated interpretation of how climate-driven and anthropogenic pressures interact and propagate across the Gulf of California socio-ecological system. Rather than acting independently, these pressures form a tightly coupled network in which impacts amplify through indirect pathways and feedback mechanisms, resulting in high vulnerability across environmental, economic, and social dimensions.
Under the baseline scenario excluding climate change, the model reveals that anthropogenic pressures alone are sufficient to drive the system toward critical states. Key outcomes—deterioration of population health and quality of life (C8), loss of terrestrial and marine habitats (C10), loss of wildlife and marine life (C9), degradation and pollution of the Gulf of California (C5), and economic losses (C7)—rapidly converge toward activation levels close to the maximum severity. The minimal differences among these concepts indicate that degradation across social, ecological, and economic domains occurs almost simultaneously, reflecting strong interdependencies within the system. This finding highlights that current patterns of productive activities already exceed the system capacity to absorb impacts, even in the absence of climate change.
When climate change is explicitly incorporated, the overall structure of system vulnerability remains consistent, but the severity of impacts increases and their convergence becomes even more pronounced. Climate change acts primarily as an amplifying driver rather than an isolated stressor, accelerating existing degradation pathways and reducing the system’s residual buffering capacity. In particular, river flow reduction (C13) and algal proliferation (C12) exhibit substantial increases in activation under climate change conditions, underscoring their sensitivity to climatic forcing and their role as key mediators of cascading impacts.
Across both scenarios, deterioration of population health and quality of life (C8) emerges as the most critical outcome, consistently reaching near-maximum activation. This result is especially relevant from an environmental management perspective, as it demonstrates that the ultimate consequences of environmental degradation extend beyond ecosystems and directly affect human well-being and social stability. Similarly, the convergence of economic losses (C7) toward critical levels highlights the strong coupling between environmental degradation and regional economic vulnerability, reinforcing the need for integrated management approaches that jointly address ecological integrity and socioeconomic resilience.
A novel contribution of this study lies in the exploration of minimum and maximum weight configurations to constrain the range of possible system responses. By testing extreme but plausible values for causal relationships, the analysis demonstrates that the system exhibits relatively narrow activation ranges for the most critical concepts, particularly those related to health, habitat loss, biodiversity loss, environmental degradation, and economic performance. This limited variability indicates that the identified vulnerability patterns are robust to uncertainty in weight assignment and are not artifacts of subjective parameter choices. In contrast, algal proliferation and river flow reduction display wider activation ranges, suggesting that these components are more sensitive to variations in causal strength and may represent leverage points for intervention.
Importantly, the scenario excluding climate change is not intended as a realistic representation of future conditions, but rather as an analytical device to isolate the effects of human activities. The scenario incorporating climate change therefore provides the most relevant insight for environmental management, as it reflects the compound pressures currently shaping the Gulf of California. Under these conditions, even moderate uncertainty in causal relationships does not alter the conclusion that the system operates near critical thresholds across multiple dimensions.
Although the specific causal relationships included in the FCM are tailored to the Gulf of California and were derived from the regional literature, the proposed framework is readily transferable to other socio-ecological systems. The methodology does not depend on a fixed set of concepts or causal links; rather, it provides a structured procedure for integrating scientific evidence, expert knowledge, and system dynamics into a coherent analytical framework. Application to other regions would require the identification of locally relevant concepts and the reconstruction of the corresponding causal structure based on available evidence and expert judgment. Therefore, while the specific network configuration may vary across case studies, the methodological approach remains broadly applicable to complex environmental systems facing interacting anthropogenic and climate-related pressures.
Taken together, these findings indicate that the Gulf of California constitutes a highly vulnerable socio-ecological system in which incremental pressures can generate disproportionate system-level consequences. The FCM-based approach adopted here enables identification of dominant pathways, critical outcomes, and sensitive components, providing a diagnostic foundation for designing targeted management strategies. While this study focuses on system diagnosis rather than intervention assessment, the results underscore the urgency of coordinated actions aimed at reducing key anthropogenic pressures and enhancing adaptive capacity in the face of climate change.
By explicitly representing interactions among environmental, economic, and social factors, this work addresses a critical gap in the Gulf of California analysis and contributes a transferable methodological framework applicable to other complex socio-ecological systems facing similar combinations of climate-driven and anthropogenic stressors.
5. Conclusion
This study provides a system-level diagnosis of the Gulf of California by integrating climate-driven and anthropogenic pressures within a unified fuzzy cognitive mapping (FCM) framework. The results reveal a highly interconnected socio-ecological system in which impacts propagate rapidly across environmental, economic, and social dimensions, even under baseline conditions without climate change.
The analysis demonstrates that persistent anthropogenic pressures alone are sufficient to drive the system toward critical levels of degradation. When climate change is incorporated, these pressures are further amplified, reducing system resilience and intensifying vulnerability across key components. In particular, deterioration of population health and quality of life, loss of habitats and biodiversity, environmental degradation, and economic losses consistently emerge as the most severely affected outcomes across all scenarios.
By exploring limiting cases through minimum and maximum causal weights, the study shows that system responses are relatively robust to variations in weight assignment. The narrow range of variation observed for the most critical concepts indicates that the overall diagnosis is not highly sensitive to subjective differences in expert judgment, reinforcing confidence in the identified vulnerability patterns.
Beyond the regional case, the proposed framework demonstrates the value of FCMs as diagnostic and decision-support tools for environmental management under conditions of uncertainty and limited data availability. Rather than providing precise predictions, the approach supports identifying dominant pathways, leverage points, and priority stressors that shape system-level outcomes.
The results highlight the need for integrated management strategies capable of addressing interacting pressures simultaneously, rather than sectoral or fragmented interventions. Although this study focuses on the Gulf of California, the methodological framework is transferable to other socio-ecological systems facing similar combinations of climate-driven and human pressures.
Future research should build on this diagnostic foundation by evaluating intervention strategies, governance options, and adaptive management pathways. Such extensions would allow the assessment of potential policy responses and their capacity to reduce vulnerability and enhance resilience in complex environmental systems.
Acknowledgments
The authors gratefully acknowledge the Instituto de Ingeniería, Universidad Nacional Autónoma de México (UNAM), for the institutional support that made this research possible. We are especially grateful to Dr. Claudio Bartolini and Dr. Rafael Ibarra Contreras for their valuable comments, insightful suggestions, and continuous support in improving the manuscript’s structure, clarity, and presentation. We also acknowledge the continuous support and encouragement provided by Dr. Rosa María Ramírez Zamora, Director of the Instituto de Ingeniería, UNAM; Dr. Rosa María Flores Serrano, Deputy Director of Hydraulic and Environmental Engineering; and Dr. Norma Patricia López Acosta, Academic Secretary of the Instituto de Ingeniería, UNAM. Their support has been greatly appreciated throughout the development of this research.
References
Alatriste A. 2024. Golfo de California: contaminación deriva en muerte masiva de especies. Available at: https:////emeequis.com/investigaciones/golfo-de-california-contaminacion-deriva-en-muerte-masiva-de-especies/ (accessed on February 21, 2025).
Araiza D. 2025. Alerta ecológica en el Golfo de California: más del 60% de las especies en deterioro. Available at: https:////www.cronica.com.mx/nacional/2025/01/25/alerta-ecologica-en-el-golfo-de-california-mas-del-60-de-las-especies-en-deterioro/ (accessed on January 25, 2025).
Arellano-Aguilar O, García-Mendoza E, Thompson K, Tirado R. 2016. Zonas muertas: los ecosistemas del mundo amenazados por la contaminación con fertilizantes. Greenpeace Mexico. Available at: https:////www.greenpeace.to/greenpeace/wp-content/uploads/2016/05/zonas-muertas.pdf (accessed on March 15, 2025).
Arias Patrón E. 2005. Plan de acción para el uso eficiente del agua en el Golfo de California. World Wildlife Fund-México, San Carlos, Sonora, Mexico. Available at: https:////awsassets.panda.org/downloads/uso_eficiente_agua_golfo_california_290905.pdf (accessed on March 11, 2025).
Band-Schmidt CJ, Bustillos-Guzmán JJ, López-Cortés DJ, Núñez-Vázquez E, Hernández-Sandoval FE. 2011. El estado actual del estudio de florecimientos algales nocivos en México. Hidrobiológica 21: 381-413.
Blanco Orozco ML. 1996. Pobreza y explotación de recursos pesqueros en el alto Golfo de California. M.Sc. thesis. El Colegio de la Frontera Norte, Tijuana, Baja California, Mexico.
Brusca RC, ed. 2010. The Gulf of California: Biodiversity and conservation. University of Arizona Press, Tucson, AZ, USA. https:////doi.org/10.2307/j.ctt181hwrc
Cariño M, Valiente C, Ortega Santos A. 2020. El Golfo de California: un espacio excepcional (1900-2020). In: Nuestro mar. Historia ambiental del Golfo de California (siglos XVI-XXI), vol. III (Cariño M, Ed.). Comares, España.
Cisneros-Mata MÁ, Montemayor-López G, Román-Rodríguez MJ. 1995. Life history and conservation of Totoaba macdonaldi. Conservation Biology 9: 806-814. https:////doi.org/10.1046/j.1523-1739.1995.09040806.x
CONAGUA. 2019. Estadísticas agrícolas de los distritos de riego, año agrícola 2017-2018. Comisión Nacional del Agua, Mexico. Available at: https:////files.conagua.gob.mx/conagua/publicaciones/Publicaciones/EADR_2017-18.pdf (accessed on May 29, 2026).
CONANP. 2005. Serial nomination format for the Islands and Protected Areas of the Gulf of California, Mexico for inscription on the World Heritage List. Comisión Nacional de Áreas Naturales Protegidas-SEMARNAT, Mexico.
CONAPESCA. 2022. Anuario estadístico de acuacultura y pesca 2022. Comisión Nacional de Acuacultura y Pesca, Mexico. Available at: https:////nube.conapesca.gob.mx/sites/cona/dgppe/2022/ANUARIO_ESTADISTICO_DE_ACUACULTURA_Y_PESCA_2022.pdf (accessed on May 29, 2026).
Dickerson JA, Kosko B. 1994. Virtual worlds as fuzzy cognitive maps. Presence: Teleoperators and Virtual Environments 3: 173-189. https:////doi.org/10.1162/pres.1994.3.2.173
Dodouras S, James P. 2007. Fuzzy cognitive mapping to appraise complex situations. Journal of Environmental Planning and Management 50: 823-852. https:////doi.org/10.1080/09640560701610578
Estrada Porrúa F, Zavala Hidalgo J, Martínez Arroyo A, Raga G, Gay García C. 2023. Estado y perspectivas del cambio climático en México: Un punto de partida. Mexico City, Mexico: Programa de Investigación en Cambio Climático e Instituto de Ciencias de la Atmósfera y Cambio Climático, UNAM. Available at: https:////cambioclimatico.unam.mx/wp-content/uploads/2023/11/estado-y-perspectivas-del-cambio-climatico-en-mexico-un-punto-de-partida-unam.pdf (accessed on June 20, 2025).
Félix G, Nápoles G, Falcón R, Froelich W, Vanhoof K, Bello R. 2019. A review on methods and software for fuzzy cognitive maps. Artificial Intelligence Review 52: 1707-1737. https:////doi.org/10.1007/s10462-017-9575-1
Galindo Bect MS, Vázquez León CI, Aguilar Montero D. 2021. El impacto del represamiento del Río Colorado en la pesquería del camarón en el Alto Golfo de California, México. Frontera Norte 33. https:////doi.org/10.33679/rfn.v1i1.2137
González Villarreal FJ, Salas de León DA, Aragón-Hernández JL, Fuentes-Mariles GE, Aguilar-Amilpa E, de la Mora J, Romero-Gil JA, Arriaga Medina JA. 2024. Interacciones entre aguas continentales y marinas en el Golfo de California. 1a ed. Instituto de Ingeniería, Universidad Nacional Autónoma de México. Available at: https:////amigosgolfocalifornia.org/interaccion-entre-aguas-continentales-y-marinas-en-el-golfo-de-california/ (accessed on May 29, 2026).
González-Villarreal FJ, Alaniz-Ceja L. 2022. El Golfo de California y sus interacciones entre aguas continentales y marinas bajo el enfoque de la fuente al mar. Instituto de Ingeniería, UNAM. Available at: https:////www.iingen.unam.mx/es-mx/AlmacenDigital/Gaceta/Gaceta_Julio_Agosto_2022/Paginas/El-golfo-de-California-y-sus-interacciones-entre-aguas-continentales-y-marinas.aspx (accessed on March 15, 2025).
Gray SR, Gagnon AS, Gray SA, O’Dwyer B, O’Mahony C, Muir D, Gault J. 2014. Are coastal managers detecting the problem? Assessing stakeholder perception of climate vulnerability using fuzzy cognitive mapping. Ocean & Coastal Management 94: 74-89. https:////doi.org/10.1016/j.ocecoaman.2013.11.008
Groumpos PP, Stylios CD. 2000. Modeling supervisory control systems using fuzzy cognitive maps. Chaos, Solitons & Fractals 11: 329-336. https:////doi.org/10.1016/S0960-0779(98)00303-8
Groumpos PP. 2010. Fuzzy cognitive maps: Basic theories and their application to complex systems. In: Fuzzy cognitive maps (Glykas M, Ed.). Springer, Berlin, Heidelberg, 1-22. https:////doi.org/10.1007/978-3-642-03220-2_1
Groumpos PP. 2017. Why modeling complex dynamic systems using fuzzy cognitive maps? Robotics & Automation Engineering Journal 1: 555-563. https:////doi.org/10.19080/RAEJ.2017.01.555563
Infobae. 2021. Del océano al cuerpo humano: comprobaron los graves efectos en la salud de la contaminación con plásticos. Available at: https:////www.infobae.com/america/carbononews/2021/04/03/del-oceano-al-cuerpo-humano-comprobaron-los-graves-efectos-en-la-salud-de-la-contaminacion-con-plasticos/ (accessed on March 20, 2025).
IPCC. 2021. Climate change 2021: The physical science basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (Masson-Delmotte V, Zhai P, Pirani A, Connors SL, Péan C, Berger S, Caud N, Chen Y, Goldfarb L, Gomis MI, Huang M, Leitzell K, Lonnoy E, Matthews JBR, Maycock TK, Waterfield T, Yelekçi O, Yu R, Zhou B, Eds.). Cambridge University Press, Cambridge and New York. https:////doi.org/10.1017/9781009157896
IPCC. 2022. Climate change 2022: Impacts, Adaptation and vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (Pörtner H-O, Roberts DC, Tignor M, Poloczanska ES, Mintenbeck K, Alegría A, Craig M, Langsdorf S, Löschke S, Möller V, Okem A, Rama B, Eds.). Cambridge University Press, Cambridge and New York. https:////doi.org/10.1017/9781009325844
Kontogianni A, Papageorgiou E, Salomatina L, Skourtos M, Zanou B. 2012. Risks for the Black Sea marine environment as perceived by Ukrainian stakeholders: A fuzzy cognitive mapping application. Ocean & Coastal Management 62: 34-42. https:////doi.org/10.1016/j.ocecoaman.2012.03.006
Kosko B. 1986. Fuzzy cognitive maps. International Journal of Man-Machine Studies 24: 65-75. https:////doi.org/10.1016/S0020-7373(86)80040-2
Kosko B. 1996. Fuzzy engineering. Prentice-Hall, Englewood Cliffs, NJ, USA.
Koulouriotis DE, Diakoulakis IE, Emiris DM, Antonidakis EN, Kaliakatsos IA. 2003. Efficiently modeling and controlling complex dynamic systems using evolutionary fuzzy cognitive maps. International Journal of Computational Cognition 1: 41-65.
Lara-Rosano FJ, Gallardo Cano A, Almanza Márquez S. 2021. Teorías, métodos y modelos para la complejidad social: un enfoque de sistemas complejos adaptativos. Comunicación Científica-UNAM, Mexico.
Lustig N. 2007. Salud y desarrollo económico. El caso de México. El Trimestre Económico 74: 793-822. Available at: https:////www.eltrimestreeconomico.com.mx/index.php/te/article/view/383/580 (accessed on March 15, 2025).
Meyers RA, ed. 2009. Encyclopedia of complexity and systems science. Springer, New York, NY, USA. https:////doi.org/10.1007/978-0-387-30440-3
Mitchell M. 2009. Complexity: a guided tour. Oxford University Press, New York, NY, USA. https:////doi.org/10.1093/oso/9780195124415.001.0001
Nair A, Reckien D, van Maarseveen MFAM. 2019. A generalised fuzzy cognitive mapping approach for modeling complex systems. Applied Soft Computing 84: 105754. https:////doi.org/10.1016/j.asoc.2019.105754
Nix RK. 2013. The Gulf of California: A physical, geological & biological study. University of Texas, Dallas, TX, USA. Available at: https:////archive.org/details/thegulfofcaliforniaaphysicalgeologicalbiologicalstudy (accessed on June 20, 2025).
Olvera-Fuentes NE, Gay-García C. 2023. Fuzzy cognitive maps to explore the repercussions of less precipitation on the water supply service of the Mexico City Metropolitan Area. Atmósfera 36: 299-316. https:////doi.org/10.20937/ATM.53037
Özesmi U, Özesmi SL. 2004. Ecological models based on people’s knowledge: A multi-step fuzzy cognitive mapping approach. Ecological Modeling 176: 43-64. https:////doi.org/10.1016/j.ecolmodel.2003.10.027
Páez Osuna F, Álvarez Borrego S, Ruiz Fernández AC, García Hernández J, Jara Marini ME, Bergés Tiznado ME, Piñón Gimate A, Alonso Rodríguez R, Soto Jiménez MF, Frías Espericueta MG, Ruelas Inzunza JR, Green Ruiz CR, Osuna Martínez CC, Sánchez Cabeza JA. 2017. Environmental status of the Gulf of California: A pollution review. Earth-Science Reviews 166: 181-205. https:////doi.org/10.1016/j.earscirev.2017.01.014
Páez-Osuna F, Álvarez-Borrego S, Ruiz-Fernández AC, García-Hernández J, Jara-Marini ME, Bergés-Tiznado ME, Piñón-Gimate A, Alonso-Rodríguez R, Soto-Jiménez MF, Frías-Espericueta MG, Ruelas-Inzunza J, Green-Ruiz C, Osuna-Martínez CC, Sánchez-Cabeza JA. 2019. Estatus ambiental de la contaminación en el Golfo de California: una síntesis actualizada. In: Costas y mares mexicanos: contaminación, impactos, vulnerabilidad y cambio climático (Botello AV, Villanueva FS, Gutiérrez J, Eds.). Universidad Nacional Autónoma de México-Universidad Autónoma de Campeche, 71-94. https:////doi.org/10.26359/epomex.0419
Papageorgiou EI, Stylios CD, Groumpos PP. 2006. Unsupervised learning techniques for fine tuning fuzzy cognitive map causal links. International Journal of Human-Computer Studies 64: 727-743. https:////doi.org/10.1016/j.ijhcs.2006.02.009
Papageorgiou EI, Stylios CD. 2008. Fuzzy cognitive maps. In: Handbook of granular computing, chap. 34 (Pedrycz W, Skowron A, Kreinovich V, Eds.). John Wiley & Sons, Chichester, UK, 755-775. https:////doi.org/10.1002/9780470724163.ch34
Papageorgiou EI, Kontogianni A. 2012. Using fuzzy cognitive mapping in environmental decision making and management: A methodological primer and an application. In: International perspectives on global environmental change (Young SS, Silvern SE, Eds.). InTech, 427-450. https:////doi.org/10.5772/29375
Papageorgiou EI, Salmerón JL. 2013. A review of fuzzy cognitive maps research during the last decade. IEEE Transactions on Fuzzy Systems 21: 66-79. https:////doi.org/10.1109/TFUZZ.2012.2201727
Rickard JT, Aisbett J, Morgenthaler DG, Yager RR. 2020. Modeling of complex system phenomena via computing with words in fuzzy cognitive maps. IEEE Transactions on Fuzzy Systems 28: 3122-3132. https:////doi.org/10.1109/TFUZZ.2019.2953615
Ross TJ. 2010. Fuzzy logic with engineering applications. John Wiley & Sons. https:////doi.org/10.1002/9781119994374
Sayama H. 2015. Introduction to the modeling and analysis of complex systems. Geneseo: Open SUNY Textbooks, State University of New York. Available at: https:////anvari.net/5_SA/Complex%20Systems/Modeling%20Complex%20Systems.pdf (accessed on June 8, 2025).
Schuerkamp R, Giabbanelli PJ. 2023. Extensions of fuzzy cognitive maps: A systematic review. ACM Computing Surveys 56: 1-36. https:////doi.org/10.1145/3610771
Sepúlveda AF. 1976. La conservación ecológica del Golfo de California. Natural Resources Journal 16: 465-473.
SGM. 2021. Anuario estadístico de la minería mexicana, 2020. Servicio Geológico Mexicano, Hidalgo, Mexico. Available at: https:////www.sgm.gob.mx/productos/pdf/Anuario_2020_Edicion_2021.pdf (accessed on May 29, 2026).
Simon HA. 2012. The architecture of complexity. In: The roots of logistics (Klaus P, Müller S, Eds.). Springer, Berlin, Heidelberg, 335-361. https:////doi.org/10.1007/978-3-642-27922-5_23
Stylios CD, Groumpos PP. 2004. Modeling complex systems using fuzzy cognitive maps. IEEE Transactions on Systems, Man, and Cybernetics, Part A: Systems and Humans 34: 155-162. https:////doi.org/10.1109/TSMCA.2003.818878
Tatarkanov AA, Alexandrov IA, Chervjakov LM, Karlova TV. 2022. A fuzzy approach to the synthesis of cognitive maps for modeling decision making in complex systems. Emergent Science Journal 6: 368-381. https:////doi.org/10.28991/ESJ-2022-06-02-012
WWF. 2020. Living planet report 2020: Bending the curve of biodiversity loss (Almond REA, Grooten M, Petersen T, Eds.). World Wildlife Fund, Gland, Switzerland. Available at: https:////www.wwf.org.uk/sites/default/files/2020-09/LPR20_Full_report.pdf (accessed on June 20, 2025).
Zadeh LA. 1965. Fuzzy sets. Information and Control 8: 338-353. https:////doi.org/10.1016/S0019-9958(65)90241-X