Hybrid AHP-DEMATEL Model for Prioritizing Key Resilience and Sustainability Drivers and Controllers in Agri-Food Supply Chains

For the development of this research, a mixed methodological approach was adopted, combining a SLR with a hierarchical analysis of the determining factors in AFSC. In the first phase, an SLR was conducted following guidelines established in the literature [24, 25], with the aim of identifying the main controllers and drivers that influence the efficiency and sustainability of AFSCs.

Subsequently, in the second phase of the study, the AHP and DEMATEL methods were employed to analyze the hierarchical structure and causal interdependencies between the identified factors. This methodological combination not only enabled a structured classification of the key elements but also facilitated the evaluation of their relationships of influence, thus providing a robust analytical framework for the formulation of strategies to optimize the management of AFSCs.

2.1 Systematic literature review

In agricultural systems (AS), the interaction among the actors involved enhances resilience capacity. An AS is the result of understanding the roles that each individual plays within it, enabling greater development of skills, knowledge acquisition, and increased trust among the participants. This leads to improvements in both the economic and social aspects, which in turn affect the evaluation of diverse perspectives among farmers concerning environmental sustainability and SC performance [2].

AFSCs face constant challenges due to product shortages and disruptions in their flow, causing significant environmental impacts. In this context, CE proposes strategies such as recycling, extending product lifecycles, reusing, and maximizing the value of resources [26], facilitating more efficient and sustainable management of AFSCs. Implementing these principles requires that each link in the chain recognizes the value of the products they deliver and optimizes the use of raw materials, including the waste generated in previous stages, thereby promoting process redesign, enhancing social responsibility, and increasing environmental sustainability. As a result, adopting a strategy based on CE improves operational efficiency and the resilience of AFSCs, optimizing their performance and reducing environmental impacts.

Various events have disrupted SCs, prompting a strong motivation for innovation [27]. This is why flexibility in AFSCs suggests that each link in the chain (farmers, producers, food industry, market, and consumers) must develop and implement technology at different stages of the SC, adding value to AFSCs and leading to an eco-innovative perspective where competitive advantages become evident [28].

The Internet of Things (IoT) enhances productivity and sustainability in the agro-food system [29, 30]. With IoT, numerous aspects could be better managed, such as real-time traceability in the SC, considering more convenient factors when making decisions about product care and harvesting, and even using technology in the field to monitor climate conditions more frequently [31].

AFSCs also face challenges in product quality due to perishability [32]. Therefore, quality management throughout the AFSC plays a significant role. For this reason, the design of the SC is considered, where the use of blockchain (BC) is promoted as it fosters transparency within the regulatory system, building trust between producers and consumers. It also allows consumers to make decisions based on accurate information regarding SC traceability, thereby adding value to the system [33].

The research was based on a SLR to identify the key drivers and controllers of sustainability in AFSCs. The literature search was conducted across three high-impact academic databases: Scopus, ScienceDirect, and Web of Science (WoS), chosen for their relevance in the field of sustainability and AFSC management. The search period was limited to articles published between 2020 and 2024. Additionally, the search was restricted to articles written exclusively in English, as this is the most common language in international scientific literature. Various combinations of keywords were used to optimize the relevance and comprehensiveness of the search in the SLR. The main keywords and their combinations covered the key topics of the research, including "agricultural supply chain," "sustainability," "drivers and controllers," "agro-industrialization," "supply chain sustainability," "environmental impact," and "water footprint" to identify relevant drivers and controllers. About sustainability in the agricultural supply chain, terms such as "agricultural systems," "supply chain," "agroecosystems," "biodiversity conservation," "organic farming," "water usage," and "life cycle assessment" were used. Furthermore, combinations related to analytical methods such as "AHP", "DEMATEL", "analytical hierarchy process," "decision-making in supply chains," and "multicriteria decision analysis" were included. Specific search phrases included "sustainability and agricultural supply chain," "drivers of sustainability in supply chains," and "controllers of agro-industrial systems," aiming to comprehensively cover key sustainability aspects and methods applied in decision-making within AFSCs.

Initially, 850 potentially relevant articles were identified, distributed across 410 from Scopus, 270 from ScienceDirect, and 170 from Web of Science. After removing duplicates and conducting a preliminary screening of titles and abstracts, predefined inclusion and exclusion criteria were applied. Inclusion criteria required studies to be published between 2020 and 2024, written in English, explicitly address sustainability, resilience, or risk management within AFSCs, and employ analytical frameworks or empirical models. Exclusion criteria included studies not peer-reviewed, those belonging to grey literature (e.g., theses or institutional reports), conceptual papers lacking a robust methodological approach, and those unrelated to the agri-food context. Additionally, to ensure transparency and quality in the selection process, a quality assessment was conducted using a structured rubric considering (i) methodological rigor, (ii) thematic relevance to AFSCs, and (iii) analytical depth. Only studies scoring above 70% in this evaluation were included. As a result of this rigorous process, the final sample was reduced to 320 studies, of which 181 directly addressed the research objectives. After a second evaluation, 34 high-quality studies were selected for in-depth analysis, ensuring alignment with the topic and methodology of the SLR. The findings from this selection provided the empirical and conceptual foundation for identifying 18 key drivers and controllers of sustainability in AFSCs, which are presented in Table 1.

Table 1. Drivers and controllers

Subsequently, the grouping and structured coding of the drivers and controllers were carried out in three levels, allowing for an accurate classification of the key factors. Based on this structure, a systematic methodology was developed aimed at applying the hierarchical process, ensuring a rigorous and consistent analysis of the relationships between the identified elements (Figure 1).

11.jpg

Figure 1. Framework for grouping drivers and controllers

2.2 Analytical Hierarchy Process (AHP)

AHP is a widely used multicriteria decision-making (MCDM) method for evaluating and prioritizing criteria in complex decision-making scenarios [52]. Its application allows a problem to be broken down into a hierarchical structure and assigns relative weights to the considered factors, facilitating a structured and objective analysis [51, 53].

For the implementation of the AHP method in this study, the following steps were followed:

Step 1. Definition of inclusion and exclusion criteria: The elements to be prioritized were established through a SLR, selecting the most relevant drivers.

Step 2. Construction of the comparison matrix: A pairwise comparison matrix was developed using Saaty’s nine-point scale (Table 2), which allows for quantifying the relative importance of each driver [54].

Table 2. Pairwise comparison scale

Step 3. Normalization of the matrix: The matrix was normalized by dividing each driver’s weight by the sum of its corresponding column.

Step 4. Determination of the weight of each driver: The relative weight of each driver was calculated as the average of the values obtained in the normalized matrix.

Step 5. Evaluation of consistency: The validity of the experts’ judgments was assessed by calculating the Consistency Ratio (CR), defined as CR = CI/RI, where CI denotes the Consistency Index and RI the Random Index. A pairwise comparison matrix is considered acceptably consistent when CR < 0.1. Twelve international experts from Mexico, Spain, Ecuador, and Venezuela participated in the analysis. Half of the panelists held doctoral degrees, and the remaining 50% held master’s degrees. All experts had a minimum of five years of experience in applying MCDM methods. Their academic and professional backgrounds span key areas such as industrial engineering, SCM, and agri-food sustainability, including specialists in risk management and sustainable supply chains. The interdisciplinary and international composition of the panel contributed to the robustness, credibility, and objectivity of the weighting process.

2.3 Decision-Making Trial and Evaluation Laboratory (DEMATEL) method

The DEMATEL method has become an effective tool for analyzing complex systems, allowing for the management of large volumes of variables and quantifying their causal interactions [55]. Its application facilitates the identification of influence relationships between the analyzed factors, providing an analytical framework to evaluate their impact on the system's structure.

One of the main reasons for adopting the DEMATEL method in this study is its ability to model bidirectional relationships between factors, enabling a deeper understanding of the interaction dynamics within a hierarchical system [56]. Furthermore, its matrix-based approach allows for the classification of factors into causal and effect groups, aiding in problem planning and visualization through graphical representations that highlight causal relationships between elements [57].

Table 3. Linguistic influence scale

For the application of the DEMATEL method, the following methodological steps were followed [58, 59]:

Step 1. Construction of the initial matrix: The evaluation matrix was established using a linguistic influence scale (Table 3), where experts assigned values based on the degree of relationship between the factors [60].

Step 2. Calculation of the direct influence matrix (X): This is obtained by multiplying the minimum value of the initial matrix by each element in the matrix.

Step 3. Construction of the identity matrix (I): A matrix is generated in which the main diagonal contains values of 1, and the rest of the elements are 0.

Step 4. Determination of the total relationship matrix (T): This is calculated using the equation $T=X-(I-X)^{-1}$, where all interactions between the factors are represented.

Step 5. Calculation of the R and C vectors: R is obtained as the sum of the values in each row of the total matrix, representing the influence exerted by each factor. Likewise, C is the sum of each column, reflecting the influence received by each factor.

Step 6. Determination of the importance and influence of the factors: The value of R+C is calculated, indicating the relevance of each factor in the system, while R−C helps distinguish causal factors from effect ones.

The results obtained in this study highlight the importance of each driver and factor within the sustainable agri-food system, as well as how they interrelate and influence decision-making in intervention strategies. By analyzing the driver weights obtained through the AHP method, it is observed that the driver "AS" holds the highest weight (0.314), emphasizing its central relevance within the model. This finding aligns with previous studies that stress the need to adapt and evolve agricultural systems to ensure long-term sustainability, highlighting the importance of innovation in agricultural practices to address challenges such as climate change and food security [1, 2]. On the other hand, "WF" ranks second with a weight of 0.231, underscoring the significance of efficient water management in the agri-food sector, in line with growing global concerns about water scarcity and its impact on agricultural sustainability [61].

The "RA" driver ranks third with a weight of 0.115. This result reflects the growing interest in the ability of the agri-food system to adapt and recover from negative impacts, such as climate change or economic crises, which is consistent with proposals from various studies emphasizing the importance of resilience for sustainability [62]. Fourth and fifth places are occupied by "BC" and "ES" with weights of 0.093 and 0.075, respectively. These drivers are crucial for promoting integration and coordination within AFSCs, which, according to authors like Fusto et al. [63], facilitates process optimization and improves efficiency in agri-food production systems. Environmental sustainability also stands out as a key factor for the sustainability of the chain, particularly regarding the preservation of natural resources and the reduction of negative environmental impacts [64].

Regarding the "IOT" driver, which occupies the last position with a weight of 0.034, this result reflects the still emerging role of digital technologies in the agri-food sector. Although the Internet of Things holds great potential to enhance agricultural management efficiency and product traceability [65], its impact is smaller compared to other more traditional drivers, such as water management and system resilience.

The consistency of the evaluation was validated by calculating the CR, with a value of 0.094, which is below the threshold of 0.1, ensuring that the results obtained are reliable and do not present significant contradictions in the driver weights.

On the other hand, the importance and influence matrix constructed using the DEMATEL method reinforces the hierarchy of the drivers within the model. The AS2 driver, with a value of 10.646, emerges as the most important factor, indicating that improving operations and coordination within the agri-food chain is crucial to optimizing system performance. This finding is supported by studies that emphasize the need for a more integrated and coordinated approach to managing AFSCs to improve efficiency and sustainability [63]. In second place is RA1, highlighting the relevance of having solid regulatory frameworks and inter-institutional agreements to support the transition to more sustainable agri-food systems. This point aligns with literature that underscores the need for public policies that incentivize sustainable and resilient practices within the agri-food sector [1, 66].

As for the drivers with lesser relevance, such as WF1, which ranks last with a value of 7.559, its low impact may reflect the growing importance of other factors more directly related to operational efficiency and sustainability in terms of production and distribution, while water management, although essential, does not appear to be a decisive factor in the overall model analyzed.

The analysis of the influence of the factors on the system, represented in the causal diagram, shows how certain drivers act as causal controllers, while others are more susceptible to external influences. This differentiation is essential for understanding the dynamics of interaction between the various elements of the agri-food system and identifying those factors with the greatest capacity to bring about structural transformations. In particular, the most relevant drivers, such as AS2, have a significant impact on other key components of the model, emphasizing their strategic role in the sustainability and resilience of AFSCs.

This study rigorously answers the research question posed by identifying and prioritizing the key drivers and factors in the sustainability of AFSCs. Through a SLR and the use of robust analytical methodologies such as AHP and DEMATEL, an integrated methodological framework is established to evaluate the relevance and impact of these factors. The findings indicate that drivers like AS and WF are determinants, underscoring the need to incorporate environmental criteria and climate change adaptation into strategic decision-making. Moreover, the prioritization of drivers such as AS2 and RA1 highlights the importance of strengthening organizational capacities and ensuring effective policy alignment to promote a systemic approach to sustainability within AFSCs.

The results of this study provide a clear view of the most influential factors in agri-food sustainability and offer a solid foundation for designing intervention strategies that prioritize the key drivers. The findings also highlight the importance of consistent evaluation and the need to apply robust methodological approaches like AHP and DEMATEL to understand the complexity of agri-food systems and their transition to more sustainable and resilient practices.

Furthermore, the main contribution of this study lies in the application of advanced analytical tools to prioritize these drivers and factors, providing a quantitative basis for strategic decision-making in the context of agri-food sustainability. This approach not only answers the research question posed but also contributes to the advancement of sustainability science in AFSCs, offering a replicable model for other supply chains.

Regarding future lines of research, it would be relevant to explore the practical implementation of the proposed models in specific contexts, analyzing their applicability and effectiveness in different regions and types of AFSCs. It is also recommended to delve into the analysis of the interaction between drivers through longitudinal studies that allow observing their evolution and the effects of policies and strategic decisions over the long term. Additionally, it would be valuable to incorporate socio-economic and cultural variables into the analysis of drivers, as these factors could significantly influence the sustainability and resilience of AFSCs in diverse contexts.

This study presents some limitations that should be considered in future research. Firstly, the SLR conducted was limited to English-language articles published between 2020 and 2024, which may have excluded important previous studies or significant contributions from non-English literature. This restriction may have reduced the breadth of the theoretical framework used, so the inclusion of more diverse sources and older literature could enrich the analysis and provide a more comprehensive view of the drivers and controllers of sustainability and resilience in AFSCs.

Another limitation of the study is its primarily theoretical approach, based on the hybrid AHP-DEMATEL model for prioritizing drivers. While this model provides a robust analytical framework, its applicability in real-world scenarios has not yet been empirically validated. Therefore, it is essential for future research to validate these models in practical contexts, assessing their effectiveness in different regions and types of AFSCs. This validation will provide evidence of the model's utility in specific situations, enhancing its value in strategic decision-making within AFSCs.

Despite these limitations, the significance of the research topic lies in the growing importance of sustainability and resilience in AFSCs, particularly in the current context of climate change, economic crises, and demand fluctuations. The proposed model not only contributes to prioritizing key drivers of sustainability but also offers a quantitative tool for strategic decision-making, making it a replicable model for other supply chains. Therefore, this research represents a significant advancement in the field of agri-food sustainability, providing a solid foundation for future research and practical applications.