Framing social systems for ecosystem-based management: The Guadalquivir estuary-Gulf of Cadiz coupled SES as case study

Environmental Development 55 (2025) 101206 Available online 13 March 2025 2211-4645/© 2025 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Framing social systems for ecosystem-based management: The Guadalquivir estuary-Gulf of Cadiz coupled SES as case study Alfredo García-de-Vinuesa a,* , David Florido b , Cesar Vilas c , María ´ Angeles Torres a , Marina Delgado a , Isabel Mu˜ noz a , Remedios Cabrera-Castro d,e , Fernando Ramos a , Marcos Llope a,f a Centro Oceanogr´ afico de C´ adiz, Instituto Espa˜ nol de Oceanografía (IEO-CSIC), Puerto Pesquero, Muelle de Levante, 11006, C´ adiz, Spain b Grupo para el Estudio de las Identidades Socioculturales en Andalucia (Geisa), Departamento de Antropología Social, Universidad de Sevilla (US), Seville, Spain c Instituto Andaluz de Investigaci´ on y Formaci´ on Agraria, Pescara, Alimentaria y de la Producci´ on Ecol´ ogica (IFAPA), Puerto Santa María, C´ adiz, Spain d Departamento de Biología, Facultad de Ciencias del Mar y Ambientales, Universidad de C´ adiz, Avda. República Saharaui, s/n, 11510, Puerto Real, C´ adiz, Campus de Excelencia Internacional del Mar (CEIMAR), Spain e Instituto Universitario de Investigaci´ on Marina (INMAR), Campus de Excelencia Internacional del Mar (CEIMAR), Avda. República Saharaui, s/n, 11510, Puerto Real, C´ adiz, Spain f Centro Oceanogr´ afico de Gij´ on/Xix´ on, Instituto Espa˜ nol de Oceanografía (IEO-CSIC), Camín de L’Arbeyal, s/n, 33212, Xix´ on, Asturies, Spain ARTICLE INFO Keywords: Social network analysis (SNA) Social–ecological systems (SES) Integrated ecosystem assessment (IEA) Human dimension Ecosystem services Nature’s contributions to people (NCP) ABSTRACT Conserving and using the oceans, seas, and marine resources sustainably is a high-level management goal encouraged by the United Nations Decade of Ocean Science for Sustainable Development and endorsed by most national policies. Estuaries are complex Social-Ecological Systems (SES) impacted by pressures from multi-sectoral activities. In these contexts, a holistic management approach, such as Ecosystem-Based Management (EBM), is essential to prevent the loss of ecosystem services. The Guadalquivir estuary-Gulf of Cadiz (Ge-GoC) is an intricate SES that faces pressures from numerous sectoral activities, including fishing, agriculture, shipping, aquaculture, and mining. The cumulative effects of these pressures (such as juvenile exploitation, eutrophication, pollution, riverbank erosion, and the introduction of alien species) could potentially drive the SES toward an ecological regime shift and deplete current ecosystem services such as its nursery role. Although there is a good understanding of the Ge-GoC ecosystem dynamics, no efforts have been made to consider and incorporate the human dimension, which is essential for successful EBM implementation. Social Network Analysis (SNA) is a tool from social sciences that characterizes the relationships among stakeholders within a given social setting. In an effort to frame the Ge-GoC social system, the first SNA was conducted, involving interviews with 55 stakeholders representing 11 sectors. The SNA identified key stakeholders from the government, fishing, shipping, surveillance, local city councils energy and NGO sectors due to their high centrality. While the shipping and energy sectors displayed significant influence in estuary management, they demonstrated limited interest and, in some cases, disagreement with the overall SES objectives, in contrast to the NGO and surveillance sectors. The primary management goals identified by stakeholders include reducing water pollution, controlling invasive species, combating drug trafficking, and addressing illegal fishing. However, the majority of stakeholders * Corresponding author. E-mail address: [email protected] (A. García-de-Vinuesa). Contents lists available at ScienceDirect Environmental Development journal homepage: www.elsevier.com/locate/envdev https://doi.org/10.1016/j.envdev.2025.101206 Received 7 January 2024; Received in revised form 4 March 2025; Accepted 9 March 2025 Environmental Development 55 (2025) 101206 2 expressed reluctance regarding the goal of shipping optimization. The information extracted through SNA provides a valuable knowledge base for creating participatory processes that can guide complex SES toward EBM. 1. Introduction Historically, estuaries have been prime locations for human settlements due to their abundant natural resources (Small and Nicholls, 2003; Boerema and Meire, 2017). These environments are complex Social-Ecological Systems (SES) that face significant anthropogenic impacts from multiple sectoral activities, including pollution, resource overexploitation and dredging, among other challenges. These pressures often lead to cascading effects (Wang et al., 2011; Biggs et al., 2022). Impacts caused by these pressures can result in the loss of vital ecosystem services, which are crucial for human well-being (Ostrom, 2019). In these contexts, a holistic approach such as Ecosystem-Based Management (EBM) is well-suited to addressing these impacts by determining the optimal balance of ecosystem goods and services and assessing cross-sector risks, with the ultimate goal of preserving SES functionality (Levin et al., 2009; Dolan et al., 2015; Link and Browman, 2017). Rigid top-down institutional regimes are common in complex SES but often lack a holistic understanding of social-ecological interactions, leading to reduced management efficiency (Ding et al., 2023). As a result, they are ill-suited for advancing EBM. In recent years, we have witnessed the collapse of several SES due to the lack of EBM considerations within such regimes, with notable examples including the Black Sea in the 1980s and the Mar Menor more recently. In these cases, high mortality events among fauna occurred due to eutrophication caused by nitrates and phosphates from agricultural activities, which also economically impacted human populations through the loss of cultural (aesthetic), regulating, and provisioning ecosystem services (IPBES, 2013; Lloret et al., 2008; Sandonnini et al., 2021; Cort´ es-Melendreras et al., 2022; Lazar et al., 2024). The Guadalquivir estuary and the Gulf of Cadiz (Ge-GoC) constitute a complex coupled SES (Llope, 2017). The estuary and its adjacent waters are recognized as nursery grounds for several commercial species, such as anchovy and prawn (Gonz´ alez-Orteg´ on et al., 2010a), and they also provide habitat for other non-commercial but ecologically important species, such as mysids (Vilas-Fern´ andez et al., 2008). This nursery role offers a regulating service to the entire Gulf of Cadiz, connecting the estuary with the neighbouring marine ecosystem and playing a crucial role in supporting fisheries (De Carvalho-Souza et al., 2018; Mir´ o et al., 2020). The positive synergies between EBM and Marine Protected Areas (MPAs) are well-established, with both strategies complementing each other to enhance ocean health (Halpern et al., 2010). In 2004, an MPA was established at the mouth of the estuary and its neighbouring waters to protect its nursery services (BOJA, nº 123, June 24, 2004; IEO, 2005). However, this protection mechanism only regulates fishing activities within the MPA (Florido del Corral and Maya-Jariego, 2018), leaving other human activities such as agriculture, aquaculture, shipping, tourism, and mining unregulated. This lack of regulation directly or indirectly impacts the MPA, affecting its nursery role (Llope, 2017; De Carvalho-Souza et al., 2018). Consequently, the MPA faces numerous unregulated pressures, including: (i) invasions of non-indigenous species primarily driven by increased maritime traffic (Gonz´ alez-Orteg´ on et al., 2010b; Reyes-Martinez and Gonzalez-Gordillo, 2019; Ruiz-Delgado et al., 2019; Gonz´ alez-Orteg´ on and Moreno-Andr´ es, 2021), (ii) eutrophication due to fertilizer runoff and outdated urban water purification systems (Ruiz et al., 2017), (iii) pollution from heavy metals resulting from mining sewage discharge (Tornero et al., 2014; Sanz-Ramos et al., 2022), and (iv) juvenile stock exploitation due to illegal fishing (Garrido et al., 2020). Utilizing the EBM approach in this SES could significantly benefit both the role of the MPA and the sustainability of fishery resources. As in all complex SES, the Ge-GoC exhibits competing management objectives among different sectors, leading to recurrent conflicts. Some of these conflicts include: (i) Water uses: In recent years, competition for water resources has intensified due to climate change, resulting in more frequent drought episodes. This situation has particularly affected the agricultural sector, where stakeholders compete fiercely for freshwater (Bea Martínez et al., 2021). Water dynamics are also crucial for other sectors, such as fishing, which could be negatively impacted by changes in ecological flow. This may lead to episodes of eutrophication, resulting in decreased abundance and biodiversity of marine species, as observed in other regions (Bunn and Arthington, 2002; Lloret et al., 2008). (ii) Dredging: This activity is conducted on the estuarine channel that provides access to the Port of Seville. There are differing scientific opinions regarding its impact on the estuarine ecosystem (Donadei, 2020). While some experts believe that dredging significantly affects the abundance and diversity of flora and fauna in the estuary, as well as its physical and biological processes (Gonz´ alez-Orteg´ on et al., 2010a; Caballero et al., 2018), others argue that the effect is negligible from a global perspective, primarily impacting local conditions (Don´ azar-Aramendía et al., 2018; Mir´ o et al., 2022). The purpose of this dredging is to maintain a minimum navigation depth of 6.5 m (Ruiz et al., 2015). The Port of Seville aims to optimize shipping operations, including the ongoing dredging outlined in the "Basic Project for Navigation Optimization in the Eurovia E.60.02 Guadalquivir" (Pascual et al., 2023). This project intends to facilitate maritime traffic, particularly in the cruise shipping sector, thereby boosting tourism in Seville. However, it raises concerns among sectors like agriculture due to the potential for increased turbidity and salinity levels (Díez-Minguito et al., 2012; 2013). On the other hand, the governance of the Guadalquivir estuary is based on a siloed top-down institutional regime that lacks the adaptability needed to address current problems and tackle future challenges (M´ endez et al., 2012, 2019). Additionally, the current A. García-de-Vinuesa et al. Environmental Development 55 (2025) 101206 3 governance structure consists of multiple layers of policymakers (EU, national, and regional) with diverse competencies (fisheries, environment, coasts, infrastructure, etc.), which do not facilitate a comprehensive approach to resource management and, therefore, obstruct the implementation of EBM. In summary, natural and anthropogenic pressures on the Ge-GoC, combined with the current rigid governance system, are creating an unstable situation for resource conservation and challenging stakeholders. Therefore, it is necessary to employ concrete tools to reverse this situation. Collaboration within networks is essential for solving complex SES problems that cannot be addressed by individual stakeholders or sectors alone, and it can be beneficial in progressing toward EBM (Ressurreiç˜ ao et al., 2012; Smythe et al., 2014; Richter et al., 2015; Bodin et al., 2017; Hedlund et al., 2021; Franco-Melendez et al., 2021). But, the successful implementation of EBM largely depends on the ability to understand how each social system works (Arkema et al., 2006; Dolan et al., 2015). While the natural system and its functioning are relatively well understood (Vilas-Fern´ andez et al., 2008; Torres et al., 2013; Gonz´ alez-Orteg´ on et al., 2018; Ca˜ navate et al., 2021), no studies have been conducted so far to describe the social system in the Ge-GoC. Social Network Analysis (SNA) is a commonly used tool in sociology and social anthropology to describe social systems, particularly in natural resource management settings (Bodin and Crona, 2009). SNA entails a participatory process that usually involves interviewing stakeholders using a pre-established questionnaire (Maya-Jariego et al., 2016a). The information extracted by SNA is useful for identifying key stakeholders based on their representation within the system, uncovering trade-offs, and describing communication patterns. It can also shed light on trust levels, management goals, and power dynamics (Bodin et al., 2006; Blacketer et al., 2022). All these properties within a social system could be highly valuable for informing and shaping a balanced and equitable participatory process (Jentoft, 2000; Crona and Bodin, 2010) in the Ge-GoC, which would help advance toward EBM and serve as a model for other complex SES. The aim of this work is to provide the first description of the Ge-GoC social system, regarding stakeholders’ relationships, their sectoral management objectives and role in terms of trust and power dynamics using SNA. We hope that this knowledge will lay the foundations of a science-based participatory process suited to support and facilitate a legitimate and successful EBM. 2. Material and methods 2.1. Study area The Guadalquivir estuary is the longest and only navigable estuary in Spain, stretching from Sanlúcar de Barrameda at its mouth to a dam located in Alcal´ a del Río, just above the city of Seville, covering a distance of 110 km (Ruiz et al., 2015). Our study focuses on the Ge-GoC coupled SES in southwestern Europe (Fig. 1), which include three provinces: Huelva, Cadiz, and Seville, along with their corresponding Spanish territorial waters. However, our efforts were concentrated more on the Guadalquivir estuary, as the majority of Fig. 1. Guadalquivir estuary-Gulf of Cadiz (SW Spain). Main land uses within the Ge-GoC SES are represented. A. García-de-Vinuesa et al. Environmental Development 55 (2025) 101206 4 stakeholders involved in the study conduct their activities there. Additionally, we included three external universities (Granada, C´ ordoba, and M´ alaga) that conduct research related to the Ge-GoC. 2.2. Stakeholders mapping Firstly, experts (members of the research team) compiled a list of stakeholders representing key sectors in the study area. These stakeholders were classified into the categories of resource exploitation, conservation, research, and management. Secondly, a preliminary interview was conducted with one stakeholder from each sector to validate the list and, if necessary, identify any additional relevant organizations that might have been overlooked using a snowball sampling approach. During these preliminary interviews, we also asked stakeholders about their sectoral management goals (Table 1), which were later used in the SNA to identify the main management goals in the Ge-GoC. The final list comprised 55 stakeholders from 11 different sectors (Table 2). All interviewees held relevant positions within their respective organizations, such as director, president, or secretary general. This ensured that they not only had a deep understanding of their own sectors but also a comprehensive overview of the broader stakeholder landscape. 2.3. Interview structure Information for the SNA was collected through semi-structured, face-to-face interviews, each lasting an average of 1 h. A questionnaire was designed for this purpose (Table 3), based on the instrument developed by Maya-Jariego in 2016 (Maya-Jariego et al., 2016a,b). The questionnaire inquired about the interviewed stakeholder’s relationships with others in terms of recognition (question #1a), interaction frequency (question #1b), and specific management-related interactions (question #1c). Additionally, it included questions about trust, power dynamics, and management goals (questions #2, #3, and #4, respectively). Only question #1a generated a yes/no response, producing a dichotomous square matrix (Matrix #1a). Questions #1b, #1c, #2, and #3 were assigned weights, resulting in four weighted square matrices (Matrices #1b, 1c, 2, and 3, respectively). Question #1c was divided into two subquestions, with the second subquestion used only when the response to the first was positive. In question #4, interviewees were specifically asked about 12 management goals (identified during preliminary interviews), which generated an asymmetric matrix (Matrix #4). Due to its internal policy (don’t interested in assess relationships details), 1 of the 55 organizations interviewed did not answer questions 2, 3 and 4. 2.4. Data analysis 2.4.1. Matrices characterization Firstly, matrices generated from questions #1a (recognition), #1b (frequency), and #1c (management) were analysed. To characterize these matrices, we calculated the number of ties, density, and reciprocity for each (Table 5). The number of ties refers to the total number of links in each matrix, density is the number of ties divided by the total possible ties, and reciprocity was calculated as the number of reciprocal ties in a network divided by the total number of existing ties (Borgatti et al., 2002). Correlations between these matrices were analysed using the Quadratic Assignment Procedure (QAP) at a 95 % confidence level (pvalue <0.05) (Table 6). QAP is a non-parametric statistical test that assesses the correlation between two matrices using quadratic assignment procedures. Specifically, our QAP analysis provided Pearson’s correlation coefficient (r) and significance levels. The Pearson correlation coefficient indicates the strength and direction of the correlation; the farther "r" is from zero, the stronger the association or similarity. The sign of "r" indicates the direction: for example, a negative sign means that when the values of one network increase, the values of the other network decrease (inverse correlation). Lastly, to examine the association between frequency (matrix #1b) and management (matrix #1c) in more detail, we created dichotomous matrices for each frequency level (rarely, occasionally, and habitually) as well as for each management type (any management and estuarine management). We then compared these matrices using a QAP correlation analysis (Table 7). Table 1 Management goals. List of main SES management goals proposed by stakeholders during preliminary interviews. Management goals 1-Marshlands restoration 2-Reducing riverbank erosion 3-Decreasing turbidity 4-Increasing control over salinity changes (e.g., control over the dam) 5-Increasing control over invasive species (especially those introduced through maritime traffic) 6-Decreasing pollution (e.g., improvement of water treatment systems) 7-Shipping optimization (e.g., entry of larger capacity ships) 8-Increasing surveillance over activities in the fishing reserve 9-Removing illegal fishing 10-Removing drug trafficking 11-Updating activity census (e.g., number of fishing vessels) 12-Development of sustainable economic activities (e.g., ecotourism) A. García-de-Vinuesa et al. Environmental Development 55 (2025) 101206 5 2.4.2. Centrality Centrality refers to the extent to which a node (in our case, a stakeholder) is connected to other nodes within a network. We analysed centrality based on degree, betweenness, and closeness (see Freeman, 1979). Degree centrality represents the number of direct connections a node has, where in-degree refers to incoming links and out-degree refers to outgoing links. Closeness centrality represents the inverse of the sum of the distances between a node and all other nodes in the network; a stakeholder is considered highly "close" if it is connected to many others through short paths. Betweenness centrality quantifies the number of shortest paths that pass-through a given node, highlighting its role as a bridge between other nodes. To identify central (key) stakeholders, we first calculated the "in" and "out" degree means for each stakeholder using matrices #1a, #1b, and #1c. Next, two multidimensional scaling (MDS) analyses were conducted using centrality measures (degree, closeness, and betweenness), calculated for each stakeholder in matrices #1a, #1b, and #1c, based on the Bray-Curtis similarity matrix, after applying a square root transformation (Fig. 3): −1. To analyse the stakeholder’s spatial distribution by sector, the first MDS analysis was conducted including sector as factor (Fig. 3a). −2. In the second MDS analysis, we examined when a stakeholder is considered key in the social system based on its centrality measures. This was done by calculating the mean of each centrality measure (degree, betweenness, and closeness) for each stakeholder using matrices #1a, #1b, and #1c. Stakeholders were then categorized into two groups: key stakeholders, whose mean centrality measures were greater than the median of all stakeholders, and non-key stakeholders. Finally, to represent this in the MDS, we used these categories as a factor (Fig. 3b). Table 2 Number of stakeholders by sector. The table displays the number of stakeholders in each of the eleven sectors considered in the study, along with the type of activity they engage in. Sector Stakeholders number Activity Fishing 14 Exploitation: marine living resources Aquaculture 3 Exploitation: marine living resources Agriculture 3 Exploitation: rice fields Shipping 2 Exploitation: commercial shipping Mining 3 Exploitation: metalliferous mines NGOs 9 Conservation: social and nature resources Research 9 Research: social and nature Local city councils 6 Management: local management Government 4 Management: regional goverment Energy 1 Exploitation: hydropower Surveillance 1 Management: SES surveillance Total 55 . Table 3 Questionnaire. Questions asked during the interview. The associated matrix type, possible answers and given scoring options are indicated. Questions Matrix type Answers and score 1a – Recognition: do you recognize the following organizations? Dichotomous Yes (1) or No (0) 1b – Frequency: how often does your organization interact (talks to, participates in events or meetings) with the following organizations? Weighted Never (0), Rarely (1), Occasionally (2), or Habitually (3) 1c – Management: have your organization participated with any of the following organizations in management tasks? And, more specifically, in relation to the Guadalquivir estuary? Weighted 1# question No (0) or Yes (1) 2# subquestion No (0) or Yes (1) 2 – Trust: do you consider that the organizations on the list agree with the objectives of your organization regarding the management of the Guadalquivir estuary or not? Weighted Agree (1), Neutral (0) or Disagree (−1) 3 – Power dynamic: how much influence do you think the following organizations have on the decision-making in the management of the Guadalquivir estuary? Weighted None (0) ….A lot (5) 4 – Management goals: are the following management objectives positive or negative to you organization? Weighted From very positive (4) to very negative (−4) A. García-de-Vinuesa et al. Environmental Development 55 (2025) 101206 6 To identify key sectors, composite boxplot was generated to display centrality measures by sector for each matrix (#1a, #1b, and #1c). A Shapiro-Wilk test was then conducted to assess data normality, followed by a Levene test to check for homogeneity of variance. Since the data did not meet the assumptions of a parametric ANOVA, a pairwise Wilcoxon rank-sum non-parametric test was used to identify significant differences (p-value <0.05) between sectors in terms of centrality measures for each matrix. 2.4.3. Cohesion Social cohesion refers to the proximity, coordination, and stability of relationships between members of a group, which benefit the group as a whole (Taylor and Davis, 2018). To study cohesion by sectors, we used density and the E-I index. Sector density was calculated as the number of existing links within a sector divided by the maximum possible number of links. Density ranges from 0 to 1, with higher values indicating a denser network and lower values indicating a sparser one (Zedan and Miller, 2017). Density was calculated for matrices #1a, #1b, and #1c, resulting in density matrices #1a, #1b, and #1c. The mean values of these matrices were then calculated to produce a general density matrix. The E-I index, on the other hand, is defined as the number of ties external to the group minus the number of ties internal to the group, divided by the total number of ties (Krackhardt and Robert, 1988). It is used to study homophily, which in sociology refers to the tendency of individuals to associate and bond with others who are similar to them. Negative E-I index values indicate a tendency toward homophily. The E-I index was calculated for each pair of sectors in matrices #1a, #1b, and #1c. 2.4.4. Trust, power dynamics and management goals In our study, trust refers to the level of confidence in management goals among stakeholders, while power dynamics represent the influence or control each stakeholder exerts over SES resources, specifically considering the goods and services provided by the Guadalquivir estuary. To analyse trust and power dynamics, we first employed a Pairwise Wilcoxon Rank-Sum non-parametric test to assess significant differences (p-value <0.05) between sectors individually. Then, we compared them with each other. To do this, we first added +1 to all scores in matrix #2 to transform negative scores into positive values. We then calculated the average scores by sector and standardized them as percentages for both matrix #2 and matrix #3. To compare trends, we created a bar chart. Finally, to investigate the correlation between trust and power dynamics, we performed a QAP analysis on matrices #2 and #3 at a 95 % confidence level (p-value <0.05). For the analysis of management goals (Table 1), we first calculated the mean and standard deviation of the scores obtained from question #4 of the questionnaire (Table 2). A circular bar chart was then generated to visualize the results (Fig. 6). Since shipping optimization was the only goal perceived negatively by the social system, we compared this goal with the others to identify significant differences using a Pairwise Wilcoxon Rank-Sum non-parametric test (p-value <0.05) (see Fig. 7). The software UCINET was used to calculate matrix characteristics, including QAPs, centrality measures, density, and the E-I index (Borgatti et al., 2002). Pairwise Wilcoxon Rank-Sum non-parametric tests, boxplots, and bar charts were performed using R 3.3.1 (R Core Team, 2016). The MDS analysis was conducted using PRIMER 6.1.2 (Clarke and Gorley, 2006) (see Fig. 2). In order to provide a comprehensive overview of the data analysis section, a summary is presented in Table 4: Fig. 2. Analytical framework. Conceptual diagramme showing the methodological design, logic and workflow and is linked to data sources. A. García-de-Vinuesa et al. Environmental Development 55 (2025) 101206 7 3. Results 3.1. Matrices characterization As expected, there was a gradual decline in the values of all network characteristics (number of ties, density, and reciprocity) from the general recognition matrix (matrix #1a) to the more specific management matrix (matrix #1c). Management relationships were the least common among stakeholders, with the number of ties and density in the management matrix being less than half of those in the recognition matrix (see Table 5). The QAP analysis revealed a significant positive relationship across all matrices (see Table 6), indicating that a management relationship is more likely to exist when stakeholders have participated together in seminars, events, or meetings (r =0.757), compared to when they simply recognize each other (r =0.419). Lastly, we found significant associations between all frequencies of interaction (rarely, occasionally, and habitually) and the two management scenarios (any management and estuarine management) (see Table 7). The greater the frequency of interaction, the higher the likelihood of joint management, and an even greater probability of managing the estuary together. 3.2. Centrality MDS analyses reveal that within certain sectors, stakeholders exhibit similar centrality measures, indicating a tendency toward homogeneity. This pattern is evident in the government sector (Fig. 3a), where all stakeholders are key stakeholders (central) (Fig. 3b). A similar trend is observed in the mining and agriculture sectors, though in these cases, all stakeholders are non-key (non-central). The Table 4 Data Analysis Summary: Table 4 displays information regarding the topics and their associate matrices, specific measures, statistical analyses, and outputs. Data analysis Topic 2.4.1 Matrices characterization 2.4.2 Centrality 2.4.3 Cohesi´ on 2.4.4 Trust, power dynamics and goals Matrices 1a,1b, and 1c 1a,1b, and 1c 1a,1b, and 1c 2,3 and 4 Specific measures Number of ties, density and reciprocity Degree, betweenness and closeness Density and E-I index – Statistical analysis QAP MDS and Wilcoxon –QAP and Wilcoxon Output Tables 5–7 Fig. 3 (MDSs) and Fig. 4 (box plots) Table 8 Fig. 5 (bar chart), and Fig. 6 (circular bar chart) Table 5 Recognition (matrix #1a), frequency (matrix #1b) and management (matrix #1c). Characteristics are showed as number of ties, density, and reciprocity. Characteristics/Matrix Recognition Frequency Management Number of ties 2008 1171 863 Density 0.676 0.394 0.291 Reciprocity 0.720 0.714 0.626 Table 6 QAP correlations among recognition (matrix #1a), frequency (matrix #1b) and management (matrix #1c). Values indicate the Pearson Correlation Coefficient (r). Matrix Recognition Frequency Management Recognition 1 0.505 0.419 Frequency 0.505 1 0.757 Management 0.419 0.757 1 Table 7 QAP Correlations between frequency and management: Pearson Correlation Coefficient (r) indicates strength and direction in the association between the different frequencies of stakeholder engagement (rarely, occasionally, and habitually) and the type of management. Frequency Any management Estuary management Rarely 0.123 0.109 Occasionally 0.184 0.286 Habitually 0.273 0.445 A. García-de-Vinuesa et al. Environmental Development 55 (2025) 101206 8 aquaculture sector shows a mix of both key and non-key stakeholders. In contrast, the fishing, shipping, research, local city councils, and NGO sectors show a tendency toward heterogeneity in their centrality measures, as these sectors include both key and non-key stakeholders. Notably, the two shipping stakeholders are widely separated in the MDS space. Finally, the energy and surveillance sectors are each represented by a single key stakeholder. Fig. 4 displays boxplots of centrality measures by sector for the recognition, frequency, and management matrices. A decrease in degree and closeness is observed across almost all sectors from recognition to management. However, an inverse trend is noted for betweenness, indicating that certain sectors, such as government, gain importance by acting as a bridge between sectors in management situations, where there is a lower level of connection. In fact, the government generally showed higher scores in centrality measures than all other sectors. These differences were significant when compared with the fishing sector, as well as with local city councils (in terms of degree and closeness in management) and NGOs (in terms of degree in recognition). These results suggest that the government is a key sector due to its high centrality measures and could play a crucial role in establishing contact between sectors during management tasks. It is worth noting that the fishing sector is positioned in the middle-low range of the boxplot for nearly all centrality measures across the three matrices. These measures are significantly higher when compared to sectors such as agriculture and mining, which exhibit limited connections with the social system, and significantly lower compared to research and local city councils. Furthermore, it is remarkable that the surveillance sector demonstrates some of the highest centrality scores in both recognition and frequency matrices. However, in the management matrix, its centrality values are notably low, indicating that this sector is practically not involved in management activities. 3.3. Cohesion The analysis of density by sectors revealed that, in general, they are poorly connected, exhibiting greater intrasectoral than intersectoral density (Table 8), except in specific cases. The mining, NGO, local city council, and government sectors demonstrate Fig. 3. MDS analysis: Fig. 3a shows the MDS analysis on stakeholders’ centrality measures using sectors as factor, while in Fig. 3b, the same analysis was conducted using in this case centrality level (key and non-key stakeholders) as factor. A. García-de-Vinuesa et al. Environmental Development 55 (2025) 101206 9 Fig. 4. Key sectors boxplot. The boxplot show degree, betweenness and closeness by sectors for recognition, frequency and management matrices. Table 8 General density matrix representing the density by sectors in percentage. The colour gradient represents a spectrum of relationships ranging from low density (white) to high density (green). The intrasectoral density is highlighted in bold font. A. García-de-Vinuesa et al.