Evaluation of Adaptive Management Strategies for Agricultural Production to Climate Change Using Interval Type-2 Fuzzy Sets with Symmetric Fuzzy Numbers

Abstract

Climate change affects all sectors, with a particularly significant impact on agricultural production. Therefore, agricultural production must adapt to these changes, and adaptive strategies for managing agricultural production should be applied. This research evaluates which adaptive strategies yield the best results in agricultural production in Bosnia and Herzegovina and Serbia through expert decision-making. In doing so, the interval type-2 fuzzy set (IF2S) is applied, using symmetric fuzzy numbers through the membership function. The results obtained by applying IF2S M-SiWeC (Modified Simple Weight Calculation) show that the criteria of the greatest importance are yield stability and climate risk reduction. The ranking of the six selected adaptive strategies is carried out using the IF2S MABAC (Multi-Attributive Border Approximation area Comparison) method, which indicates that agricultural production diversification and adaptive water management strategies provide the best results according to expert assessments. These results are confirmed by additional analyses, including comparative analysis and sensitivity analysis. The contribution of this research is reflected in proposing guidelines on the adaptive strategies that should be applied in practice in agricultural production in order to reduce the negative effects of climate change.

1. Introduction

Climate change has become a major challenge in the 21st century [1], affecting agricultural production, which is crucial for food supply. It is necessary to change the way agricultural production is carried out and adapt to climate factors. These factors are reflected in increasing average air temperatures, changes in precipitation intensity, more frequent and prolonged droughts, as well as the occurrence of extreme weather events [2]. All this affects the productivity and stability of agricultural systems. Agriculture is highly sensitive to climate change, and, on the other hand, the implementation of intensive agricultural production activities contributes to climate change [3]. Therefore, it is necessary to achieve a balance by applying sustainable and regenerative agricultural production in order to manage soil and agricultural resources [4].

The application of the concept of sustainable agricultural production aims to reduce the negative impact on the environment and climate change. The application of this concept tries to preserve agricultural resources for future generations [5]. Recently, increasing attention has been given to the concept of regenerative agricultural production, which seeks to overcome the concept of sustainable agricultural production by trying to restore biodiversity, stimulate biological activities and strengthen the ecological resilience of agricultural production. These concepts aim to adapt agricultural production to climate change and, at the same time, to have a positive effect on climate change, trying to eliminate negative impacts.

The applications of this concept of agricultural production are limited to the area of Southeastern Europe. The specificity of this region is a temperate continental and Mediterranean climate, which is characterized by high exposure to droughts and heat waves. In addition, agricultural production in this region is reflected in a fragmented structure of land holdings, limited investment capacities and varying degrees of technological development and the application of digital tools in agriculture. Therefore, the countries of this region must expand their capacities to influence the structural factors that have led to this state of agriculture. Moreover, there is an increasing need for scientific and systematic approaches that include different strategies. The goal of these strategies is to enable the agricultural systems of particular countries to better respond to climate change.

The implementation of the strategies involves certain challenges regarding maintaining agricultural production at a certain level while simultaneously reducing the negative effects of climate change. For this reason, strategies need to be evaluated from different aspects and include environmental, economic, technological and social dimensions. In this way, the selection of an appropriate strategy must be carried out by applying multiple criteria, while it is also necessary to incorporate uncertainty into decision-making due to the specificity of the observed area, the lack of full information and the application of subjective decision-making. In order to evaluate these strategies, the countries Serbia and Bosnia and Herzegovina (BiH) were selected. These countries were chosen due to their similar climate conditions, and the level of development of these two countries is approximately at the same level. Both of these countries are European Union pre-accession countries. In addition, these countries are primarily connected by geopolitical and other ties, while they both also have certain similarities and differences.

Due to the complexity present in the evaluation of strategies, it is necessary to additionally adapt the methodological concept that will encompass imprecision and uncertainty in decision-making. The standard approach of using fuzzy sets is not sufficient when considering more complex decisions, as it is important to define input data and preferences. Based on this, it is essential to upgrade this approach, utilizing the opportunity of using linguistic assessments obtained by a subjective approach based on the use of expert knowledge. For this reason, an approach based on the type-2 fuzzy set is used, and it allows for the inclusion of uncertainty in decision-making as well as the possibility of the membership function fluctuating [6]. In addition, the approach is further upgraded through the use of interval membership functions since, in the standard approach, this function is not clear during transformation. Therefore, it is necessary to define the membership functions as an interval. Moreover, the use of an interval membership function allows for the inclusion of a wider range of values, which provides additional security in decision-making, particularly when the decision is made with imprecise data. The application of the IF2S enables modeling under conditions of uncertainty and risk, additionally including insecurity in the decision. This contributes to greater robustness and security in decision-making.

The motivation for this research stems from the need to harmonize the productivity and efficiency of agricultural production in relation to climate change, reduce the impact of climate change, make decisions in conditions of insecurity and uncertainty, and include the countries of Serbia and BiH. Accordingly, the aim of the research is to evaluate the strategies for adapting agricultural production to climate change in the conditions of Serbia and Bosnia and Herzegovina using expert opinions and IF2S. In order to achieve this aim, the following research questions are posed:

-

How to evaluate strategies for adapting agricultural production to climate conditions while incorporating uncertainty and insecurity into decision-making?

-

Which strategies show the best potential for use in Serbia and BiH?

-

How do the selected criteria affect certain strategies in terms of their ranking?

This research provides multiple contributions. First, by examining strategies for agricultural production management, theoretical knowledge is obtained on how to adjust this production and on the transition toward sustainable and regenerative agricultural production. Then, the method for evaluation and assessment of these strategies is improved by applying IF2S and appropriate multi-criteria decision-making (MCDM) methods in order to make a decision on the selection of these strategies. The upgrade of this approach is carried out through the formation of symmetric membership functions of fuzzy numbers. In this way, equal importance is assigned to individual evaluations. By developing this approach, the theoretical foundations for the improvement of IF2S are also provided.

2. Literature Review

The literature review will first address studies on the application of MCDM in agricultural production management under climate change, followed by studies on the use of selected strategies in agricultural production management. In this way, the studies that are directly related to the subject of this research will be selected.

2.1. Application of MCDM Methods in the Management of Agricultural Production Due to Climate Change

In systematizing the research, the starting point was the parameter that MCDM methods are applied in the context of adapting agricultural production to climate conditions. In addition, recent studies were selected and then grouped into certain thematic units in relation to specific research focuses. The studies focused on the use of MCDM methods were aimed at evaluating adaptive strategies and were directed toward the implementation of investments at the regional or national level. Zamani et al. [7] developed a decision support system for the selection of adaptive strategies in response to climate trends using the fuzzy TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) and PROMETHEE (Preference Ranking Organization Method for Enrichment Evaluation) II methods. The following studies also took into account the social aspect of decision-making and examined the views of different experts. Stričević et al. [8] used the AHP (Analytic Hierarchy Process) and TOPSIS methods to rank potential adaptive measures in Serbia. The authors also compared experts’ priorities and determined which criteria played a greater role in ranking the measures. Karimi et al. [9] applied group AHP in their study, involving experts and farmers in Iran, to investigate the need for changes in agriculture caused by climate change. Additionally, in Iran, Zobeidi et al. [10] used the PROMETHEE-GAIA method to evaluate adaptive methods and further improve the understanding of the specificities of managing these methods. Shayanmehr et al. [11] used the AHP and TOPSIS methods to assess sustainable crop production, including sustainability factors. The same methods were also used by Mahmoudi et al. [12], who compared sustainable agricultural production systems taking into account climate resilience and greenhouse gas emissions as key research indicators.

In addition to including sustainability in agricultural production, previous studies considered the use of smart agriculture in the context of climate change. Singh et al. [13] developed a climate smartness index using the AHP method to determine the importance of indicators, establishing an important link between the application of smart technologies and crop yields. Rajbonshi et al. [14] used a combination of MCDM methods that included TOPSIS, COPRAS (Complex Proportional Assessment) and SAW (Simple Additive Weighting) methods to rank smart technologies in the context of climate factors in rice cultivation and worked on mitigating greenhouse gas emissions. Mohapatra et al. [15] developed a climate smartness index for agricultural systems based on rice production, proposing diversification with legumes as an alternative to climate change. In addition to sustainable management and the application of smart technologies, Yazdani et al. [16] focused their approach on risk management, particularly in relation to flood risks, applying the SWARA (Stepwise Weight Assessment Ratio Analysis) method.

MCDM methods have also been used in various other analytical tools for addressing specific agricultural production; thus, these studies are grouped. Crop selections were carried out in the research by Saqlain et al. [17], who used the SPOTIS (Stable Preference Ordering Towards Ideal Solution), Random Forest and MULTIMOORA (Multi-Objective Optimization by Ratio Analysis plus the Full Multiplicative Form) methods for selecting the most suitable crop. Nandi et al. [18] used the CRITIC (Criteria Importance Through Intercriteria Correlation) and WASPAS (Weighted Aggregated Sum Product Assessment) methods for crop selection, while Ali and Khan [19] used the VIKOR (ser. VIšeKriterijumska Optimizacija I Kompromisno Rešenje) method to assess the impact of climate conditions on different crops. The assessment of agricultural land suitability was carried out by Hossen et al. [20], using the AHP and GIS (Geographic Information System) methods, to evaluate land suitability in Bangladesh. MCDM methods were used to consider water and land resource management, where Kavand et al. [21] assessed water conservation policies and Zagaria et al. [22] evaluated adaptation potential at the farm level. In addition, MCDM methods were used to improve decision-making frameworks and, in this context, Hakim et al. [23] used PIPRECIA (PIvot Pairwise RElative Criteria Importance Assessment) and MACROS (Measurement of Alternatives and Ranking according to Compromise Solution) methods to select locations for sustainable agricultural production.

In addition, some authors have considered future directions for using MCDM methods in adapting agricultural production to climate change. Streimikis [24] considered the application of MCDM methods to preserve agricultural resources for future generations, where the concepts of uncertainty, circular economy and social equity should be used in decision-making in order to enable the transition of agricultural systems for better adaptation to climate change. Morkunas and Volkov [25] applied the SAW, TOPSIS and VIKOR methods to indicate inequality in EU countries in terms of adapting agricultural production. Padma et al. [26] used the AHP method to rank strategies for promoting precision agricultural production, which is one of the responses to climate change.

2.2. Strategies for Adapting Agricultural Production to Climate Conditions

In practice, there are numerous strategies that can be applied. However, this research selected six strategies, and the following section provides an overview of studies in which they have been applied.

The diversification of agricultural production (Strategy 1) represents one of the strategies used by farmers to adapt to climate change that will be used in this study. Whether it is spatial and temporal crop diversification, the introduction of new crops, crop rotation or combination with livestock farming, diversification reduces the risk of complete loss of income due to extreme weather conditions. Numerous factors influence the use of diversification, and various studies confirm that climatic conditions have a direct impact on diversification. Asfaw et al. [27] examined the countries of Malawi, Niger and Zambia and showed that diversification is directly related to extreme climate conditions. Ochieng et al. [28] confirmed that diversification is widespread and used as a risk management mechanism by examining 31-year climate data from Kenya. Mulwa and Visser [29], based on the research conducted in Namibia, showed that information on climatic conditions is a key factor in determining the level of diversification. Wang et al. [30] specified that climatic conditions and rainfall deficits during the planting season in Zambia encourage diversification, while studies by Matsuura-Kannari et al. [31] in Bangladesh and Antonelli et al. [32] in Uganda confirmed that climate shocks are drivers of crop diversification. Mohammed et al. [33] demonstrated that households that use diversification have stronger resilience to climate change.

The adaptive water management strategy (Strategy 2) is the second strategy that will be used in this research. The objective of this strategy is to respond to climate change caused by drought or heavy rainfall. It is due to the fact that water in agricultural production is imperative in the fight against climate change, which causes increasing uncertainty in terms of water availability. Therefore, it is necessary to implement certain practices that include technological innovations and institutional reforms. In their research, El-Nashar and Elyamany [34] selected the most efficient irrigation strategy and claimed that the Value Engineering methodology provides the best results. On the other hand, Deligios et al. [35], in their research, demonstrated an innovative canopy cooling system that can increase yields by 30% with 34% water savings. In this way, this agricultural production requires less water. Garg et al. [36] demonstrated that the use of rainwater increased the groundwater level, thus providing more water to roots. Babaeian et al. [37] used the Adaptation Pathways approach with socio-hydrological modeling to test actions such as changes in sowing dates and irrigation deficit to increase crop yields. Institutional frameworks are key to success. Ahmadi et al. [38] showed that institutional frameworks are also key to improving agricultural production, identifying 40 barriers and proposing 12 strategies, which include water resources and law reform.

The regenerative soil management strategy (Strategy 3) represents the third strategy that will be used in this research. This strategy in agricultural production represents a holistic approach to soil management where the focus is on restoring soil health. This is achieved by increasing the amount of organic matter and carbon sequestration, which directly contributes to mitigating climate change and strengthening the resilience of agricultural systems to climate change. The concept of this strategy was defined by Lal [39], who stated that regenerative agriculture is a systemic approach to agricultural production that simultaneously positively affects the environment and addresses climate change caused by agricultural production. Jayasinghe et al. [40], through a review of previous studies, emphasized that regenerative agriculture exceeds sustainable practices, as sustainability focuses on preserving soil for future generations, whereas regenerative agricultural production goes further and seeks to restore soil health. Vejendla et al. [41] emphasized that organic and regenerative agriculture are sustainable strategies for reducing carbon dioxide. Ghosh et al. [42] presented a regenerative agriculture production strategy as a soil management method that improves organic matter and mitigates climate change while preserving food production. Mishra et al. [43] advocated a shift from conventional to regenerative agriculture production that increases carbon sequestration.

The digital predictive management strategy (Strategy 4) for agricultural production represents the fourth strategy that involves the integration of advanced technologies to predict climate impacts and optimize decisions in real time. Dhanke et al. [44], in their research, proposed the application of an intelligent irrigation system based on neural networks that predicts soil moisture, irrigation time and water distribution. Bayar et al. [45] investigated the application of AIoT (Artificial Intelligence of Things) in precision agriculture for dynamic irrigation planning, optimized fertilization strategies, and early detection of abiotic stress in crops in agricultural production. Dhanaraj et al. [46] introduced an intelligent system for crop yield prediction through optimization of water use based on climatic factors. Chaiyana et al. [47] conducted their research to predict crop yield in Thailand, integrating climate data and monitoring conditions via sensors. Yunianta et al. [48] applied multiple machine learning models to predict rice production in Indonesia using climatic factors such as rainfall, humidity and temperature.

The institutional risk management strategy (Strategy 5) represents the fifth strategy that will be applied in this research. This strategy involves formal mechanisms of advisory services, financial instruments and insurance, which can help farmers prepare for climate conditions. Khan et al. [49] analyzed the perspectives and capacities of public institutions in Pakistan using an institutional capacity index. Khan et al. [50] developed an institutional support index to assess the effectiveness of services and found that farmers perceive low to medium levels of institutional support. Madaki et al. [51] examined agricultural insurance as a modern risk management strategy, highlighting the need to support this practice. Amarnath et al. [52] investigated packaged climate solutions that combine insurance with hybrid seeds and mobile advice. Birthal et al. [53] examined how smallholder farmers rely on traditional measures to mitigate climate risks due to a lack of access to institutional measures. Saqib et al. [54] highlighted the need to invest in climate-smart agriculture and provide financial assistance to farmers for building sustainable agricultural systems. Singh [55] analyzed key policies at the European Union, national and regional levels to address the impacts of climate change on agriculture. Zong et al. [56] introduced a comprehensive framework for assessing climate-resilient agriculture that combines four dimensions: agricultural productivity, farmer income, climate resilience and the level of green development.

The transformational adaptation strategy (Strategy 6) represents the sixth strategy that will be examined in this research. This adaptation strategy involves deep, systemic changes in agricultural systems that alter the fundamental aspects of agricultural production in response to climate change. Zagaria et al. [57] used an agent-based model to explore transformational strategies by quantifying how climate change, farmer behavior, and water resource management affect decisions to change production, reallocate water resources and change farm sizes. Gil-Clavel et al. [58] defined transformational adaptation in agriculture as migration, crop relocation or change in farming systems, with access to information and technology being key factors toward that transformation. Käyhkö et al. [59] applied an integrated analytical framework to assess transformative adaptation processes in agro-food systems, exploring why transformative adaptation is not widely adopted. Fedele et al. [60] reviewed 60 empirical case studies in tropical and subtropical countries and found that a quarter of reported adaptation actions involved transformative adaptation. Deubelli and Mechler [61] noted in their research that the concept of transformative change has heterogeneous applications and little practical insight, indicating a lack of studies that apply this strategy. Hellin et al. [62] explained that transformative orientations require a deeper approach to social, institutional, technological and cultural changes. Castella [63] detailed the application of transformative adaptation in Laos, where agroecological transformation was carried out in hill communities through collective planning, experimentation and social learning. Gosnell [64] conceptualized the significant process of change from conventional to regenerative agriculture as a type of transformative adaptation that requires a transformation in thinking. Cradock-Henry et al. [65] applied the approach of adaptation pathways in New Zealand, emphasizing long-term strategies toward agricultural transformation as actions to adapt to the impacts of climate change accelerate.

In practice, there are other strategies that can be used to adapt agricultural production to climate conditions, but these six strategies were examined in this research. In addition, previous studies have not considered all these strategies; thus, this research fills that gap. Furthermore, these strategies have not been analyzed in this form in the selected Southeast European countries, which represents another gap that this research fills and thereby provides a certain contribution. In addition, the application of IF2S in the analysis of the strategies in this way provides a certain contribution and fills a gap present in previous studies. Thus, this research fills certain gaps present in previous studies and provides certain contributions in this regard.

3. Research Methodology

In order to evaluate adaptive strategies for the management of agricultural production in the context of climate change, it is first necessary to determine the criteria by which these alternatives will be observed and to select experts who will be respondents in this research. For this reason, it is required to form a research methodology (Figure 1) and explain its individual parts.

3.1. Basic Definitions of Fuzzy Sets

A fuzzy set is an extension of classical logic where the value of the basic set can be nuanced so that the elements of the set can take any value from the interval [0, 1]. Unlike classical logic, where the value of the basic set is binary (0 or 1), a fuzzy set allows the nuance of these values using a defined fuzzy number. This number is a special form of a fuzzy set defined on the set of real numbers, with a membership function that is convex and normalized. In practice, triangular fuzzy numbers of the form (a, b, c) are most often used, where b is the vertex and “a” and “c” are the left and right boundaries. Unlike a classical fuzzy set, an interval fuzzy set of type 2 (IF2S) is its extension where each value has two membership functions: upper and lower. This models additional uncertainty in determining the degree of membership and defining the membership function. IF2N is defined as a pair of triangular fuzzy numbers: upper ($a_{i1}^U$, $a_{i2}^U$, $a_{i3}^U$) and lower ($a_{i1}^L$, $a_{i2}^L$, $a_{i3}^L$), with the lower interval within the upper interval (Figure 2).

3.2. Research Preparation Phase

In order to evaluate the selected strategies, it is first necessary to define how these strategies will be evaluated and the objectives they should fulfill. Defining these objectives is completed through the selection of criteria. The selection of criteria represents a key segment of any research where MCDM methods are used since the selection of criteria defines the perspectives from which the alternatives, in this case strategies, will be considered [66]. Therefore, as this research examines the impact of climate change on agricultural production, it is required to include criteria that involve a broader range of perspectives for considering these strategies. For this reason, four groups of criteria were used: climate resilience, environmental sustainability, economic justification and operational feasibility. Since this evaluation is based on climate change strategies, it is necessary to first identify criteria that examine the climate resilience of the selected strategies. Then, it is essential that the implementation of these strategies does not negatively affect the environment and that the principles of sustainability and regenerative agriculture are respected. It is important to preserve the resources required for agricultural production by implementing these strategies, while also contributing to their improvement. In order to implement these strategies, they must be economically justified in order to be implemented in practice without requiring large investments and high costs. Finally, these strategies need to be feasible in practice in order to be implemented. Based on these main criteria, auxiliary criteria were defined to explain these criteria in more detail (

Table 1. Evaluation criteria.

Criteria		Description	References
C1   Climate resilience			[33,57]
C11	Climate risk reduction	Reducing the risk of negative climate effects.	[28,30,51,52]
C12	Adaptive system flexibility	Flexibility of the agricultural system and adaptability to climate conditions.	[31,32,45,58]
C13	Yield stability	Stability of yields despite climate conditions.	[29,47,48]
C2   Environmental sustainability			[35,36,39,40]
C21	Impact on soil quality and ecosystem	Impact of the strategy on land quality and biodiversity.	[42,43]
C22	Efficient water resource management	Water use efficiency.	[37,44]
C23	Long-term sustainability	Long-term sustainability without resource degradation.	[41,60,62,64]
C3   Economic justification			[35,50,54]
C31	Initial investment cost	Initial investment required for implementation.	[7,8,9,16,17,18]
C32	Operating cost	Operational on an annual basis.	[7,10,11,12,21]
C4   Operational feasibility			[38,50]
C41	Implementation complexity	Complexity of implementation by farmers.	[53,59,61]
C42	Dependence on institutional support	Dependence on institutional support.	[38,49,55,56]

). These criteria were defined based on a literature review, and the references where these criteria were used are also provided. Based on this, the criteria that have already been individually used in previous studies are defined.

After defining strategies and the criteria by which the strategies will be evaluated, it is necessary to determine the research sample, i.e., the countries to be included in this research. The sample involves two countries from Southeast Europe, namely Serbia and BiH. Serbia and BiH are located in the continental part of Southeast Europe and have a temperate-continental climate with pronounced seasonal variations in temperature and precipitation. Both countries have similar climate patterns: warm summers, moderate winters and irregular distribution of precipitation throughout the year. This climatic homogeneity allows the effects of climate change to manifest in a comparable way. Agriculture in both countries has significant economic and social importance, particularly in rural areas. Small and medium-sized farms with similar production patterns are dominant. Both countries face similar challenges in terms of the increasing frequency of droughts, unpredictable precipitation and changes in crop growth stages affected by climate. In addition, Serbia and BiH share a similar institutional legacy framework dating back to the former Yugoslavia in the field of agriculture and risk management, and there is now an effort to align these processes with the Common Agricultural Policy of the European Union (EU), as both countries are candidates for EU membership. Therefore, the selection of Serbia and BiH allows the research results to be representative of other Western Balkan countries that are characterized by similar climate and agricultural conditions. These two countries form a natural geographical and agroecological unit, and the results obtained based on these countries can be generalized to the wider area of Southeastern Europe.

When selecting respondents for this research, care was taken to ensure that they had a certain level of expertise in the research area. Therefore, professors from agricultural faculties in Serbia and BiH and researchers from agricultural institutes were selected as respondents in this research. The selected respondents possess specialized and scientifically grounded knowledge in the field of agriculture and the impact of climate change on agriculture. In addition, the respondents have long-term academic and research experience, which enables them to recognize the effects of climate change on agricultural production. These selected respondents also possess knowledge of crop yields, soil health, irrigation, as well as specific agricultural crops, and are engaged in studying the impact of climate on overall agricultural production. This is essential in order to better evaluate adaptive strategies in agricultural production. Furthermore, these respondents have become familiar with linguistic scales and their application through their research experience, so no additional explanation regarding how to complete the questionnaire is required. Through their research, they collaborate with other colleagues, forming interdisciplinary teams that enable these strategies to be viewed holistically, encompassing various aspects. Therefore, these respondents can provide assessments that are relevant, objective and balanced, which is a prerequisite for the validity of the entire research process.

3.3. Research Implementation Phase

After the selection of experts, the first phase of this research is completed, and the second phase, which is the research implementation, begins. In order to conduct the research, it is first necessary to collect evaluations from the selected respondents. For the purpose of collecting these evaluations, it is necessary to first design a survey questionnaire. Due to the specificity of the research, it is necessary to determine the importance of the selected criteria and then evaluate the alternatives in accordance with these criteria. This will be carried out in this research by applying a seven-level linguistic value scale. Consequently, this value scale is used to determine the importance of criterion weights in the first part of the questionnaire, while in the second part of the questionnaire, it is used to evaluate the strategies.

After the questionnaire has been designed, it is sent to the respondents for completion, after which the completed questionnaires are collected and processed in order to prepare the data for analysis. In order to apply IF2S, it is necessary to determine which linguistic values will be used and then define the membership function of these linguistic values (

Table 2. Defined linguistic value scales.

Linguistic Term	IF2Ns Values
Very low (VL)	(1, 2, 3; 1), (1.5, 2, 2.5; 0.9)
Low (LO)	(2, 3, 4; 1), (2.5, 3, 3.5; 0.9)
Moderately low (ML)	(3, 4, 5; 1), (3.5, 4, 4.5; 0.9)
Neutral (NE)	(4, 5, 6; 1), (4.5, 5, 5.5; 0.9)
Moderately high (MH)	(5, 6, 7; 1), (5.5, 6, 6.5; 0.9)
High (HI)	(6, 7, 8; 1), (6.5, 7, 7.5; 0.9)
Very high (VH)	(7, 8, 9; 1), (7.5, 8, 8.5; 0.9)

). In this way, the linguistic values are transformed into the appropriate interval type-2 fuzzy numbers (IF2N), and the data is prepared for the implementation of the IF2S method.

The specificity of determining the membership function (Table 2) is that symmetry was applied in determining IF2N. First, the same mean values were set for both intervals, thus applying complete symmetry. When determining the first interval, the values were defined more broadly, such that the second interval had boundary values lower by 0.5 than the first interval. The value of 0.5 was selected to ensure a clear distinction between the boundaries of the integral, thereby modeling uncertainty while simultaneously achieving complete symmetry among the fuzzy numbers. This is particularly important to maintain the proper arrangement of the applied linguistic values. In this way, another symmetry was achieved among these fuzzy numbers. Since the first interval was wider, it was assigned a greater height compared to the second interval (Figure 2). Based on the defined intervals and fuzzy numbers, the collected data were processed.

3.4. Research Results Phase

The next phase of this research is the phase of research results. In this phase, the steps of the selected fuzzy methods are carried out, and additional analyses are performed. In order to obtain research results, it is first necessary to select MCDM methods that will easily harmonize the opinions of multiple experts, particularly when determining the weights of the criteria. In this regard, the M-SiWeC (Modified Simple Weight Calculation) and MABAC methods in IF2S form were selected in this research.

The IF2S M-SiWeC method was selected since it is simple to apply and allows the use of evaluations provided by a large number of respondents, while facilitating the determination of criteria importance for respondents [67]. Namely, the respondents are not required to rank the criteria, identify the most or least important criteria or compare criteria with one another. They are only required to determine how important a certain criterion is to them. In this research, the M-SiWeC method was used instead of the standard approach. The specificity of this modified method is that deviations in respondents’ evaluations are not determined using the standard deviation, nor is any weighting procedure applied [68]. Thus, this method consists of two preparatory steps and three specific steps for this method. Those steps are as follows.

Step 1. Assessment of criteria importance using linguistic evaluations. In this way, a decision matrix is formed where there is an “m” set of experts who evaluate a set of “n” criteria.

Step 2. Application of the membership function and transformation of linguistic evaluations into IF2N.

Step 3. Normalization of IF2N.

$${\widetilde{\stackrel{~}{n}}}_{ij}=\left({\displaystyle \frac{a_{ij}^{U1}}{\mathit{max}{x}_{ij}^U}},{\displaystyle \frac{a_{ij}^{U2}}{\mathit{max}{x}_{ij}^U}},{\displaystyle \frac{a_{ij}^{U3}}{\mathit{max}{x}_{ij}^U}}; 1\right),\left({\displaystyle \frac{a_{ij}^{L1}}{\mathit{max}{x}_{ij}^U}},{\displaystyle \frac{a_{ij}^{L2}}{\mathit{max}{x}_{ij}^U}},{\displaystyle \frac{a_{ij}^{L3}}{\mathit{max}{x}_{ij}^U}}; 0.9\right)$$

(1)

where $\mathit{max}{x}_{ij}^U$ is the maximum value of IF2N.

Step 4. Determination of the aggregate weights of the criteria

$${\widetilde{\stackrel{~}{s}}}_{ij}={\displaystyle {\sum }_{j=1}^m}{\widetilde{\stackrel{~}{n}}}_j$$

(2)

Step 5. Determination of the weight of the criteria

$${\widetilde{\stackrel{~}{w}}}_{ij}=\left({\displaystyle \frac{s_{ij}^{U1}}{{\displaystyle {\sum }_{j=1}^n}{s}_{ij}^{U3}}},{\displaystyle \frac{s_{ij}^{U2}}{{\displaystyle {\sum }_{j=1}^n}{s}_{ij}^{U2}}},{\displaystyle \frac{s_{ij}^{U3}}{{\displaystyle {\sum }_{j=1}^n}{s}_{ij}^{U1}}}; 1\right), \left({\displaystyle \frac{s_{ij}^{L1}}{{\displaystyle {\sum }_{j=1}^n}{s}_{ij}^{L3}}},{\displaystyle \frac{s_{ij}^{L2}}{{\displaystyle {\sum }_{j=1}^n}{s}_{ij}^{L2}}},{\displaystyle \frac{s_{ij}^{L3}}{{\displaystyle {\sum }_{j=1}^n}{s}_{ij}^{L1}}}; 0.9\right)$$

(3)

The MABAC method was originally created by Pamučar and Ćirović [69]. This method was selected for several reasons in this research, and it allows consistent results to be obtained regardless of the form of criteria used and measurement units applied [70]. Furthermore, this method is efficient when there is a large number of criteria and alternatives and is suitable when there is a large number of respondents as well [71,72]. It can be combined with fuzzy sets and different forms, making it suitable for solving problems when there is insecurity and uncertainty in the assessment, particularly when subjective human preferences are present [73]. In order to use this method in the research, the following steps will be used:

Step 1. Evaluation of strategies via linguistic evaluations. In this evaluation, “m” alternatives are evaluated, which are evaluated using “n” criteria, and in this way, an m × n matrix is formed.

Step 2. Application of the membership function and transformation of linguistic evaluations into IF2N.

Step 3. Normalization of IF2N.

$$\begin{gathered} {\widetilde{\stackrel{~}{n}}}_{ij}=\left({\displaystyle \frac{a_{ij}^{U1}-min{ a}_{ij}^{U1}}{\mathit{max}{x}_{ij}^{U3}-min{ a}_{ij}^{U1}}},{\displaystyle \frac{a_{ij}^{U2}-min{ a}_{ij}^{U1}}{\mathit{max}{x}_{ij}^{U3}-min{ a}_{ij}^{U1}}},{\displaystyle \frac{a_{ij}^{U3}-min{ a}_{ij}^{U1}}{\mathit{max}{x}_{ij}^{U3}-min{ a}_{ij}^{U1}}}; 1\right), \\ \left({\displaystyle \frac{a_{i1}^L-min{ a}_{ij}^{L1}}{\mathit{max}{x}_{ij}^{L3}-min{ a}_{ij}^{L1}}},{\displaystyle \frac{a_{i2}^L-min{ a}_{ij}^{L1}}{\mathit{max}{x}_{ij}^{L3}-min{ a}_{ij}^{L1}}},{\displaystyle \frac{a_{i3}^L-min{ a}_{ij}^{L1}}{\mathit{max}{x}_{ij}^{L3}-min{ a}_{ij}^{L1}}}; 0.9\right) \end{gathered}$$

(4)

Step 4. Development of weighted decision matrices.

$${\widetilde{\stackrel{~}{v}}}_{ij}={\widetilde{\stackrel{~}{n}}}_{ij}\times {\widetilde{\stackrel{~}{w}}}_j+{\widetilde{\stackrel{~}{w}}}_j$$

(5)

Step 5. Determination of the border approximation area matrix (G). This value is actually the geometric mean of the weighted values for individual IF2Ns.

$${\widetilde{\stackrel{~}{g}}}_j={\left({\displaystyle {\prod }_{i=1}^n}\left({\widetilde{\stackrel{~}{v}}}_{ij}\right)\right)}^{1/n}$$

(6)

Step 6. Calculation of the deviation from the border approximation area (Q).

$${\widetilde{\stackrel{~}{g}}}_j=\left(v_{ij}^{U1}-g_j^{U3},v_{ij}^{U2}-g_j^{U2}, v_{ij}^{U3}-g_j^{U1}; 1\right),\left(v_{ij}^{L1}-g_j^{L3},v_{ij}^{L2}-g_j^{L2}, v_{ij}^{L3}-g_j^{L1}; 0.9\right)$$

(7)

Step 7. Calculation of cumulative deviations for all criteria.

$${\widetilde{\stackrel{~}{S}}}_i=\left({\displaystyle {\sum }_{i=1}^n}\left({\widetilde{\stackrel{~}{q}}}_{ij}\right)\right)$$

(8)

Step 8. Defuzzification using the centroid method.

Step 9. Calculation of the MABAC method values, where the mean value of the defuzzified values is calculated.

$$S_j={\displaystyle \frac{S_{defj}^U+S_{defj}^L}{2}}$$

(9)

After the selected methods are applied, additional analyses are also used in this research. Since the respondents in this research come from two countries, the evaluations will be compared separately for each country. First, the weights of the criteria will be compared, followed by the ranking of the strategies and the identification of whether there are differences in the respondents’ assessments. In this way, it will be determined whether the same criteria are considered important by the respondents from Serbia and Bosnia and Herzegovina and whether the ranking of strategies is the same according to respondents’ opinions. Furthermore, any difference will be identified, and its possible cause will be analyzed.

In addition to comparing the results across these countries, a comparative analysis will also be conducted to compare the rankings obtained by using different methods and determine whether there is a difference in these rankings when the same weights of criteria are applied but different steps of various methods are used. Finally, a sensitivity analysis will be performed to change the weights of the criteria and determine whether a change in the weights of the criteria affects the ranking of strategies. In this way, it will be determined whether individual criteria contribute to different rankings of strategies.

4. Results

When selecting experts for the purposes of this research, it was ensured that there were an equal number of respondents. Due to the complexity of the topic being studied, it was decided to include ten experts from Bosnia and Herzegovina and ten from Serbia, resulting in a total of twenty experts. The specificity of these experts is that they work at faculties of agriculture and agricultural institutes, which is important in order to ensure a high level of academic and professional expertise. In addition, it was ensured that the experts were from different areas of agriculture in order to obtain general results that could be applied to any branch of agriculture. In this way, interdisciplinary collaboration among the experts was enabled in order to achieve a holistic approach to the research problem. The experts had two tasks: first, to assess the importance of the observed criteria, and then to evaluate the selected strategies according to these criteria. They were asked to provide their evaluations independently and to rely on their scientific and practical experience in agricultural production. By respecting this principle, the possibility of obtaining different evaluations was ensured, with consensus being reached through their reconciliation. It should be noted that it was intended that only one expert per institution was included in order to prevent the experts from consulting with each other regarding their evaluations.

First, the importance of the research criteria will be determined, while only ten auxiliary criteria will be considered, excluding the main criteria, in order to assign equal importance to each auxiliary criterion. If a larger number of criteria is included within a main criterion, this may give greater importance to that criterion, which would then require adjustment of the evaluations to neutralize this importance. Therefore, only the auxiliary criteria are considered. In order to determine the importance of these criteria, the selected experts provide linguistic evaluations (Table 2), which are later processed using the steps of the M-SiWeC method. Therefore, data matrices with linguistic evaluations must first be formed (

Table 3. Experts’ evaluations regarding the importance of the research criteria.

Experts from Bosnia and Herzegovina
	C11	C12	C13	C21	C22	C23	C31	C32	C41	C42
E1	HI	MH	VH	VH	HI	VH	NE	NE	NE	HI
E2	MH	NE	HI	NE	MH	ML	VH	VH	MH	VH
E3	VH	HI	HI	MH	VH	NE	MH	MH	ML	MH
E4	HI	VH	VH	MH	HI	NE	HI	HI	NE	NE
E5	HI	VH	VH	MH	MH	MH	NE	NE	MH	MH
E6	VH	MH	HI	MH	HI	NE	MH	MH	ML	HI
E7	HI	MH	HI	VH	VH	MH	NE	MH	NE	MH
E8	MH	HI	MH	NE	HI	ML	HI	HI	HI	MH
E9	VH	HI	HI	MH	HI	NE	NE	NE	NE	HI
E10	HI	MH	MH	MH	MH	NE	VH	HI	MH	VH
Experts from Serbia
	C11	C12	C13	C21	C22	C23	C31	C32	C41	C42
E11	HI	MH	HI	VH	HI	HI	NE	MH	NE	MH
E12	NE	NE	HI	ML	NE	ML	VH	HI	MH	VH
E13	VH	MH	HI	MH	VH	NE	MH	MH	ML	MH
E14	MH	VH	VH	MH	HI	NE	VH	HI	MH	NE
E15	MH	VH	VH	NE	MH	MH	NE	MH	MH	MH
E16	VH	MH	HI	NE	MH	ML	MH	MH	ML	MH
E17	MH	HI	HI	VH	HI	MH	MH	MH	MH	MH
E18	NE	HI	MH	ML	HI	ML	VH	HI	VH	MH
E19	VH	MH	HI	NE	MH	NE	ML	NE	NE	MH
E20	MH	NE	MH	NE	NE	ML	VH	VH	MH	VH

).

After the data has been collected and the linguistic decision matrix has been formed, these values are transformed into IF2N using the defined membership function (Table 2). By applying this membership function, the IF2S decision matrix is formed, and the steps of the M-SiWeC method are carried out. The first step is the application of normalization. Since the highest IF2N value is 9, all values are divided by this number. Let us consider Expert 1 and the evaluations for criterion C11 as an example. This expert assigned the evaluation “High”, which is then transformed into IF2N (6, 7, 8; 1), (6.5, 7, 7.5; 0.9), after which it is divided by the value 9, resulting in the normalized IF2N decision matrix.

$${\widetilde{\stackrel{~}{n}}}_{11}=\left({\displaystyle \frac{6}{9}}=0.67, {\displaystyle \frac{7}{9}}=0.78,{\displaystyle \frac{8}{9}}=0.89; 1\right),\left({\displaystyle \frac{6.5}{9}}=0.72, {\displaystyle \frac{7}{9}}=0.78,{\displaystyle \frac{7.5}{9}}=0.83; 0.9\right)$$

The next step is to calculate the sum of the weights for the individual criteria (Equation (2)) and then calculate the weights of the criteria (Equation (3)). The results of applying the IF2S M-SiWeC method (

Table 4. Results of the importance of criteria for selecting adaptive strategies.

	${\widetilde{\stackrel{~}{\mathit{s}}}}_{\mathit{j}}$	${\widetilde{\stackrel{~}{\mathit{w}}}}_{\mathit{j}}$	Rank
C11	(12.9, 15.1, 17.3; 1), (14.0, 15.1, 16.2; 0.9)	(0.08, 0.11, 0.15; 1), (0.09, 0.11, 0.13, 0.9)	2
C12	(12.2, 14.4, 16.7; 1), (13.3, 14.4, 15.6; 0.9)	(0.08, 0.10, 0.14; 1), (0.09, 0.10, 0.12; 0.9)	4
C13	(13.4, 15.7, 17.9; 1), (14.6, 15.7, 16.8; 0.9)	(0.08, 0.11, 0.15; 1), (0.10, 0.11, 0.13; 0.9)	1
C21	(10.9, 13.1, 15.3; 1), (12.0, 13.1, 14.2; 0.9)	(0.07, 0.09, 0.13; 1), (0.08, 0.09, 0.11; 0.9)	8
C22	(12.6, 14.8, 17.0; 1), (13.7, 14.8, 15.9; 0.9)	(0.08, 0.11, 0.15; 1), (0.09, 0.11, 0.12; 0.9)	3
C23	(9.2, 11.4, 13.7; 1), (10.3, 11.4, 12.6; 0.9)	(0.06, 0.08, 0.12; 1), (0.07, 0.08, 0.10; 0.9)	10
C31	(11.8, 14.0, 16.2; 1), (12.9, 14.0, 15.1; 0.9)	(0.07, 0.10, 0.14; 1), (0.09, 0.10, 0.12; 0.9)	7
C32	(11.8, 14.0, 16.2; 1), (12.9, 14.0, 15.1; 0.9)	(0.07, 0.10, 0.14; 1), (0.09, 0.10, 0.12; 0.9)	6
C41	(9.9, 12.1, 14.3; 1), (11.0, 12.1, 13.2; 0.9)	(0.06, 0.09, 0.12; 1), (0.07, 0.09, 0.10; 0.9)	9
C42	(12.1, 14.3, 16.6; 1), (13.2, 14.3, 15.4; 0.9)	(0.08, 0.10, 0.14; 1), (0.09, 0.10, 0.12; 0.9)	5
Sum	(116.8, 139.0, 161.2; 1), (127.9, 139.0, 150.1; 0.9)

) show that the most important criterion according to the experts’ assessments is criterion C13, Yield stability, followed by criterion C11, Climate risk reduction. The least important criterion according to the experts’ assessments is criterion C23, Long-term sustainability. It should be noted that there are no significant differences in the weights of the criteria, and therefore all of them will influence the final selection of strategies.

In order to identify how certain experts have determined the importance of the weights of the criteria, the weights of the criteria are calculated individually for experts from BiH and Serbia. Applying the same steps as in the previous case, the results obtained show that the most important criteria, according to the opinions of experts from BiH and Serbia, are criteria C13, Yield stability, and C11, Climate risk reduction (

Table 5. Ranking of criterion importance according to the country of origin of the experts.

	${\widetilde{\stackrel{~}{\mathit{w}}}}_{\mathit{j}}$—Bosnia and Herzegovina	Rank	${\widetilde{\stackrel{~}{\mathit{w}}}}_{\mathit{j}}$—Serbia	Rank
C11	(0.08, 0.11, 0.15; 1), (0.10, 0.11, 0.13; 0.9)	1	(0.08, 0.11, 0.15; 1), (0.09, 0.11, 0.12; 0.9)	2
C12	(0.08, 0.10, 0.14; 1), (0.09, 0.10, 0.12; 0.9)	4	(0.08, 0.10, 0.14; 1), (0.09, 0.10, 0.12; 0.9)	3
C13	(0.08, 0.11, 0.15; 1), (0.10, 0.11, 0.13; 0.9)	1	(0.08, 0.11, 0.16; 1), (0.10, 0.11, 0.13; 0.9)	1
C21	(0.07, 0.10, 0.13; 1), (0.08, 0.10, 0.11; 0.9)	6	(0.06, 0.09, 0.13; 1), (0.08, 0.09, 0.11; 0.9)	8
C22	(0.08, 0.11, 0.15; 1), (0.09, 0.11, 0.13; 0.9)	3	(0.08, 0.10, 0.14; 1), (0.09, 0.10, 0.12; 0.9)	3
C23	(0.06, 0.08, 0.12; 1), (0.07, 0.08, 0.10; 0.9)	9	(0.06, 0.08, 0.12; 1), (0.07, 0.08, 0.10; 0.9)	10
C31	(0.07, 0.10, 0.13; 1), (0.08, 0.10, 0.11; 0.9)	6	(0.08, 0.10, 0.14; 1), (0.09, 0.10, 0.12; 0.9)	3
C32	(0.07, 0.10, 0.13; 1), (0.08, 0.10, 0.11; 0.9)	6	(0.08, 0.10, 0.14; 1), (0.09, 0.10, 0.12; 0.9)	3
C41	(0.06, 0.08, 0.12; 1), (0.07, 0.08, 0.10; 0.9)	10	(0.06, 0.09, 0.13; 1), (0.08, 0.09, 0.11; 0.9)	8
C42	(0.08, 0.10, 0.14; 1), (0.09, 0.10, 0.12; 0.9)	4	(0.07, 0.10, 0.14; 1), (0.09, 0.10, 0.12; 0.9)	7

). However, among experts from Serbia, the same ranking is maintained for the first two criteria, while the ranking for other criteria differs significantly, with four criteria having the same weights and being ranked third. For the criteria among experts from BiH, two criteria share first place, and three criteria share sixth place.

After the weights of the criteria are obtained, the evaluation of the adaptive strategies for the management of agricultural production is carried out. The selected experts evaluate the strategies in terms of how much they satisfy the defined research criteria. This evaluation is conducted in the same way as when determining the weights of the criteria using linguistic values (

Table 6. Evaluation of the adaptive strategies for agricultural production.

E1	C11	C12	C13	C21	C22	C23	C31	C32	C41	C42
Str. 1	HI	VH	MH	HI	NE	HI	HI	MH	HI	NE
Str. 2	HI	MH	VH	MH	VH	MH	NE	NE	NE	MH
Str. 3	HI	MH	HI	VH	HI	VH	MH	MH	NE	MH
Str. 4	MH	HI	MH	NE	MH	NE	ML	NE	NE	MH
Str. 5	MH	NE	NE	NE	NE	MH	MH	MH	NE	HI
Str. 6	HI	HI	NE	HI	MH	VH	NE	NE	ML	MH
⁝	⁝	⁝	⁝	⁝	⁝	⁝	⁝	⁝	⁝	⁝
E20	C11	C12	C13	C21	C22	C23	C31	C32	C41	C42
Str. 1	HI	HI	MH	MH	NE	MH	VH	HI	HI	MH
Str. 2	HI	MH	HI	NE	HI	NE	MH	MH	MH	MH
Str. 3	MH	NE	MH	HI	MH	MH	HI	MH	MH	MH
Str. 4	MH	HI	MH	NE	MH	NE	NE	NE	NE	MH
Str. 5	VH	MH	MH	NE	NE	MH	HI	HI	MH	VH
Str. 6	HI	HI	NE	MH	NE	HI	NE	NE	NE	HI

). After that, these values are transformed into IF2N in the same way as when applying the steps of the IF2S M-SiWeC method. When the IF2N is formed, the steps of the IF2S MABAC method are performed.

Aggregation of expert opinions was performed using the arithmetic average of the IF2N values for each criterion and each strategy. The reason why the same weighting was used for the experts is that all experts were selected based on the same criteria, and all are experts. In addition, the aim of the paper is to reach a consensus assessment that represents the position of the academic and research community in both countries, without favoring individuals. It is common in the literature that when using the M-SiWeC and MABAC methods with multiple experts, equal weighting is applied unless there are clear criteria for differentiation [67]. On this basis, an aggregated IF2S decision matrix was formed.

Since the IF2Ns have been formed, an aggregated matrix has been created, representing the average evaluations of all experts. In this way, all experts equally influence the final ranking of the strategies. The first standard step of the IF2S MABAC method is the normalization of the IF2N. The normalization procedure is carried out using Equation (4), and, for example, for Strategy 1 and criterion C11, it is calculated as follows:

$$\begin{gathered} {\widetilde{\stackrel{~}{n}}}_{11}=\left({\displaystyle \frac{5.7-5.2}{8.2-5.2}}=0.17,{\displaystyle \frac{6.7-5.2}{8.2-5.2}}=0.51,{\displaystyle \frac{7.7-5.2}{8.2-5.2}}=0.85; 1\right), \\ \left({\displaystyle \frac{6.2-5.7}{7.7-5.7}}=0.26,{\displaystyle \frac{6.7-5.7}{7.7-5.7}}=0.51,{\displaystyle \frac{7.2-5.7}{7.7-5.7}}=0.77; 0.9\right) \end{gathered}$$

The next step is the weighting of the normalized IF2Ns (Equation (5)). Unlike other MCDM methods, at this step, in addition to multiplying the normalized data by the criterion weight, addition with this weight is also performed in order to prevent the weighted value from being equal to zero, thereby enabling the calculation of the geometric mean, which represents the border approximation area matrix. Using the same example, the weighting is performed as follows:

$$\begin{gathered} {\widetilde{\stackrel{~}{v}}}_{11}=\left(0.17·0.08+0.08=0.09, 0.51·0.11+0.11=0.16, 0.85·0.15+0.15=0.27; 1\right), \\ \left(0.26·0.09+0.09=0.12, 0.51·0.16+0.16=0.16, 0.77·0.22+0.22= 0.22; 0.9\right) \end{gathered}$$

After that, the average is calculated using the geometric mean (Equation (6)), followed by the calculation of the deviation of the weighted values from this average, ensuring that the first fuzzy number is smaller than the second, and that the second is smaller than the third. This is achieved using Equation (7). Finally, the sum of this deviation is calculated, after which the centroid method is used to determine the defuzzified values of the upper and lower bounds, and the final value of the IF2S MABAC method is obtained (

Table 7. Ranking results of adaptive strategies using the IF2S MABAC method.

	${\widetilde{\stackrel{~}{\mathit{S}}}}_{\mathit{j}}$	${\mathit{S}}_{\mathit{d}\mathit{e}\mathit{f}\mathit{j}}^{\mathit{U}}$	${\mathit{S}}_{\mathit{d}\mathit{e}\mathit{f}\mathit{j}}^{\mathit{L}}$	${\mathit{S}}_{\mathit{j}}$	Rank
Str. 1	(−1.44, 0.08, 1.62; 1), (−0.72, 0.11, 0.97; 0.9)	0.11	0.08	0.10	1
Str. 2	(−1.45, 0.07, 1.60; 1), (−0.74, 0.09, 0.94; 0.9)	0.09	0.07	0.08	2
Str. 3	(−1.49, 0.02, 1.54; 1), (−0.80, 0.03, 0.87; 0.9)	0.03	0.02	0.03	3
Str. 4	(−1.55, −0.07, 1.41; 1), (−0.90, −0.10, 0.72; 0.9)	−0.07	−0.09	−0.08	6
Str. 5	(−1.54, −0.05, 1.44; 1), (−0.87, −0.05, 0.77; 0.9)	−0.05	−0.05	−0.05	4
Str. 6	(−1.55, −0.06, 1.43; 1), (−0.89, −0.08, 0.74; 0.9)	−0.06	−0.07	−0.06	5

). Based on these results, it can be concluded that, according to the experts’ assessments, the best results would be provided by the diversification of the agricultural production strategy, followed by the adaptive water management strategy. The difference between these two strategies is moderate, whereas the difference between these strategies and the others is significantly greater. The weakest results based on these assessments were achieved by Strategy 4, the Digital predictive management strategy.

In order to determine whether experts’ opinions differ depending on the country they come from, the ranking of these strategies will be performed based on the expert evaluations from Bosnia and Herzegovina and Serbia. In this analysis, the expert evaluations of criteria and strategies from these countries are considered. The results of such an approach (

Table 8. Ranking of the adaptive strategies by the observed countries.

	BiH		Serbia
	${\mathit{S}}_{\mathit{j}}$	Rank	${\mathit{S}}_{\mathit{j}}$	Rank
Str. 1	0.096	1	0.102	1
Str. 2	0.084	2	0.074	2
Str. 3	0.029	3	0.022	3
Str. 4	−0.080	6	−0.086	6
Str. 5	−0.050	4	−0.033	4
Str. 6	−0.062	5	−0.066	5

) show that experts from these countries agree on the ranking of the strategies, and the same results are obtained. In this way, the results show that these two countries do not differ significantly in terms of the development level of agricultural production and that the same measures included in these strategies should be implemented in both countries.

After comparing the evaluations from these countries, an additional analysis is carried out in the form of a comparative analysis. The comparative analysis ranks the strategies using the aggregated decision matrix and the weights obtained by applying the IF2S M-SiWeC method. In this way, the ranking of strategies is influenced only by the specific steps of the other methods used in this analysis [74]. The comparison of the results obtained using the IF2S MABAC method will be conducted by applying four additional methods: SAW, ARAS (Additive Ratio Assessment), RAWEC (Ranking of Alternatives with Weights of Criterion) and CRADIS (Compromise Ranking of Alternatives from Distance to Ideal Solution). These methods will be used in the IF2S form. The specificity of the SAW and ARAS methods lies in the fact that these are earlier-developed MCDM methods and relatively simple to use, so they are usually applied for comparative analysis. In the SAW method, the ranking is formed based on the sum of weighted values [75], while in the ARAS method, the ranking is formed based on the distance from the optimal function [76]. The specificity of both methods is that they use different normalization procedures compared to the MABAC method. The RAWEC and CRADIS methods are more recent MCDM methods, and they have certain specific characteristics compared to other MCDM methods. The RAWEC method does not apply the standard procedure of data weighting, but immediately computes distance in relation to the weights of criteria. In addition, this method uses two normalization procedures, which are identical but perform data maximization and data minimization. The CRADIS method is a bit more complicated than other methods [77], as it first calculates the deviations from the highest and lowest normalized IF2N values, then determines the deviations from the ideal values of the alternatives, and finally performs a compromise of these deviations.

The results of applying these methods (Figure 3) show that only the IF2S MABAC and IF2S SAW methods provide identical rankings of the strategies. In addition, all these methods produce the same rankings for the first two ranked strategies. The greatest difference in ranking is with the IF2S ARAS method, followed by the IF2S RAWEC method. These differences arise from the specific characteristics of these methods. In the ARAS method, the difference is due to the applied data normalization procedure, while in the RAWEC method, the difference is based on the use of two normalization procedures and the fact that the standard weighting of normalized data is not applied. In the IF2S method, compared to the MABAC and SAW methods, the ranking differs in the ordering of Strategies 6 and 5, where these two strategies exchange their ranking. Based on this analysis, it can be concluded that, according to expert opinions, Strategies 1 and 2 would provide the best results. Therefore, these strategies should be applied to a greater extent compared to others in order to support the adaptation of agricultural production in the observed countries so that climate change has fewer negative impacts on agricultural production.

After the comparative analysis, a sensitivity analysis will also be performed. The sensitivity analysis aims to determine the extent to which changes in the weights of the criteria influence the ranking of the strategies [78,79]. Due to its specific implementation, this analysis can be carried out in different ways [80]. However, this research will use existing weights that will be changed. Changing these weights is carried out in such a way that each individual criterion is reduced by 30%, 60% and 90%, respectively, and the other criteria are increased proportionally. By applying the sensitivity analysis, the extent to which an individual criterion influences the ranking of strategies is examined [81]. If a reduction in the weight of an individual criterion leads to a deterioration in the ranking of a certain strategy, this indicates that the strategy has better indicators for that specific criterion compared to another strategy, and vice versa. By applying the sensitivity analysis in this way, a total of 30 scenarios are formed since each individual criterion is reduced by three times, and there is a total of ten individual criteria.

The results of this analysis (Figure 4) show that only Strategy 3 did not change its ranking position, whereas all other strategies experienced changes in their rank order. Furthermore, what indicates that Strategies 1 and 2 are the best adaptive strategies for agricultural production management is also the fact that these two strategies exchanged their ranking positions in eight scenarios. In this way, it was determined that the application of these two strategies together would provide the best results in addressing changes caused by climate change. These results show that Strategy 1 has better indicators in the criteria C12, C23, C31, C32 and C41. By developing Strategy 2 and improving these criteria, it would yield better results than Strategy 1. Furthermore, Strategy 5 was weaker than Strategy 6 and Strategy 4 in two scenarios, while Strategy 6 was weaker than Strategy 4 in six scenarios. The sensitivity analysis shows that individual criteria have influences on the ranking and that the selection of one strategy for implementation would not provide adequate results compared to the implementation of multiple strategies. Certainly, this raises the question of the feasibility of implementing all such strategies, so it is recommended to use Strategies 1 and 2, as these strategies have demonstrated the best results.

5. Discussion

The conducted research provided insight into determining priorities when selecting adaptive strategies for the management of agricultural production in the context of the impacts of climate change, using BiH and Serbia as examples. This research is based on the evaluation conducted by experts from these two countries. Ten experts from each country were selected, including professors from faculties of agriculture and researchers from agricultural institutes. These experts were selected based on their theoretical and practical knowledge of agriculture, acquired through long-term academic and research experience. To enable the selection of strategies and the determination of the importance of criteria, the experts used linguistic values as evaluations. In addition, the IF2S approach was applied, enabling imprecision and uncertainty to be included in decision-making, as qualitative indicators are used to evaluate strategies. The specificity of this research is the defined membership function based on the application of symmetric fuzzy numbers. Since intervals were used, it was ensured that the first range of fuzzy numbers included the second range in such a way that the first interval was larger and broader, while the second interval was smaller and more precise. The differences in fuzzy numbers are constant and symmetric. Intervals defined in this way allow for a realistic representation of the ambiguity that may exist among experts when providing evaluations. In addition, symmetry is preserved, and confidence is introduced into the defined hierarchy of evaluations by maintaining the constancy of the width of the defined fuzzy numbers. Furthermore, to facilitate the experts’ task, the same linguistic scale of values was used to assess the importance of both criteria and strategies.

When conducting the research, the importance of criterion weights was determined first since these weights are required to form a ranking order of the strategies. For this purpose, the IF2S M-SiWeC method was used. It represents a simplified SiWeC method in order to determine the weights of criteria in the easiest way possible. The difference from the classical SiWeC method is that the weights of expert responses were not calculated using the standard deviation indicator. The results of this method, based on expert evaluations, show that criteria C13—Yield stability and C11—Climate risk reduction have the greatest importance, while the least important criterion is C23—Long-term sustainability. Such results are obtained since these criteria are directly affected by climate change, which introduces uncertainty in agricultural production and shows the economic vulnerability of farmers in these countries. The results obtained in this way are in line with previous studies conducted by Mulwa and Visser [29] and Asfaw et al. [27], which indicate that these factors are key for the application of adaptive strategies.

This research also compared the evaluations provided by experts from BiH and Serbia, and the results showed that there was agreement that criteria C13 and C11 were the most significant in both countries. However, differences were identified in the weights of the other criteria. Experts from Serbia evaluated the criteria in such a way that as many as four criteria had the same weight and shared third place in the ranking, whereas experts from BiH showed greater differentiation among the criteria and ranked some of them lower. Such results were obtained due to differences in the institutional and economic context in these countries. Although both of these countries are developing countries and are at a similar level of development, the situation regarding agriculture is slightly better in Serbia, which has had a higher inflow of investments in this sector and received greater support from certain funds financed by the governments of other countries. Based on this, the experts ranked criterion C42—Dependence on institutional support higher than their colleagues from Serbia, due to the more complex administrative structure in BiH. This structure makes it more challenging to coordinate activities and implement institutional support for agricultural production. This is in line with research conducted by Khan et al. [50], who found that farmers perceive low to medium levels of institutional support, which significantly influences the selection of adaptive strategies for agricultural management.

In order to determine which strategy would show the best results according to the experts’ evaluations, the IF2S MABAC method was used. This method first ranked the strategies for all expert evaluations and then individually for the selected countries. In both cases, the same results were obtained, indicating that the agricultural production diversification strategy (Strategy 1) and the adaptive water management strategy (Strategy 2) represented the most promising courses of action for these countries. In this way, it was shown that the different weights of the criteria did not affect the final ranking of the strategies. These results were obtained based on the characteristics of the strategies and the specificities of the observed countries. Based on all the analyses conducted in this research, it is demonstrated that the agricultural production diversification strategy has a dominant position, and the experts believe that this strategy represents an effective mechanism for climate risk management. The results obtained correspond to the results of the research by Ochieng et al. [28], who confirmed, using the example of Kenya, that the diversification of agricultural production is used as the primary risk management mechanism. Similar results were obtained by Asfaw et al. [27], who showed that diversification is directly related to reducing the impact of climate conditions on agricultural production. The reason why this strategy was identified as the best should be sought in its inherent ability to reduce the risks that arise due to the effects of negative climate change. The advantage of this strategy was also confirmed by Wang et al. [30], who determined that climate conditions encourage the application of this strategy. The results of the research by Matsuura-Kannari et al. [31] and Antonelli et al. [32] confirmed that this strategy is flexible and therefore helps reduce risks caused by climate shocks. This proved to be a key advantage of this strategy compared to other strategies.

The adaptive water management strategy was ranked second, and in certain scenarios within the sensitivity analysis, it was also identified as the first-ranked strategy. This strategy emphasizes the key importance of water resources in the context of climate change. In addition, this strategy would achieve good results in these countries, as there are periods of drought requiring water management, which is why this strategy was highly ranked. The importance of this strategy is emphasized by Deligios et al. [35], who stated that its implementation can lead to significant water savings through increased crop yields. In addition, El-Nashar and Elyamany [34] emphasized that this strategy is imperative for the efficient management of water resources, particularly under the impact of climate change. The regenerative soil management strategy is in third place, indicating the importance of soil in agricultural production. Therefore, regenerative agriculture must be applied in order to maintain soil health, as emphasized by Lal [39] and Jayasinghe et al. [40] in their research. The lower ranking of this strategy can be explained by the fact that this strategy requires significant changes in agricultural practices and is therefore more complex to implement, as emphasized by Mishra et al. [43]. A specificity of the sensitivity analysis was that only this strategy did not change its ranking position. This indicates that this strategy receives better evaluations for all criteria compared to the three lower-ranked strategies, so the ranking of this strategy did not change.

The research results showed that the digital predictive management strategy was rated the lowest compared to the other strategies. At first glance, this result may seem counterintuitive considering the global trends in the digitalization of agriculture. However, this result can be explained by the specific characteristics of the observed countries, including limited investment capacities, a highly fragmented structure of agricultural production, and a low level of technological development and the application of digital tools in agriculture. This is expected to change over time as investments in digital technologies in agriculture grow in these countries. To achieve this, it is necessary to co-finance these investments by the state or use certain funds that will be available to agricultural producers in the near future. In addition, credit lines could be provided to agricultural producers to implement this strategy [82]. This strategy received such results since it requires significant investment costs, and the implementation of these technologies is complex, as agricultural producers must be trained to use them. This is also indicated by studies conducted by Bayar et al. [45] and Dhanaraj et al. [46], who pointed out the complexity of introducing these technologies as well as the complexity of these systems in practice. The sensitivity analysis showed that this strategy can be improved, and thus it would be more competitive with other strategies.

The institutional risk management and transformational adaptation strategies ranked fourth and fifth, respectively, although their results varied in the comparative analysis depending on the MCDM method applied. In the case of Strategy 5, the results were obtained in accordance with the research conducted by Khan et al. [50], who found that farmers perceive weak institutional support. Regarding Strategy 6, its relatively low ranking can be explained on the basis of the research carried out by Käyhkö et al. [59], who demonstrated that transformative adaptation is not widely represented.

The results of the sensitivity analysis provide an additional level of stability in the ranking of the strategies. This analysis showed that Strategies 1 and 2 changed their ranking in eight out of thirty scenarios, confirming that these two strategies are very close in performance and that a combination of these methods would most likely provide optimal results. Therefore, greater attention must be paid to these strategies if an adequate response to the negative impacts of climate change is to be achieved. On the other hand, the results of the comparative analysis showed that the use of IF2S with symmetric membership functions allows the evaluation of strategies under conditions of uncertainty and imprecision, which is characteristic of expert decision-making. This analysis indicated a high degree of consistency in terms of ranking the best strategies, which confirmed the validity of the results. Variations in the ranking of lower-performing strategies can be attributed to the specific features of the methods used, particularly in the application of data normalization and aggregation calculations.

If this model is compared with existing models, it can be shown that the proposed IF2S approach with symmetric fuzzy numbers offers several advantages over previously used models and methods. Unlike the classic fuzzy AHP and TOPSIS models [8,9], the presented approach allows modeling a higher level of uncertainty through interval membership functions, which is crucial when using subjective expert judgment. Compared to standard type-2 fuzzy approaches [6], the application of symmetric fuzzy numbers maintains the consistency of linguistic values. Also, unlike other methods that require comparison of criteria such as SWARA or AHP, the M-SiWeC method does not burden experts with additional ranking or comparison, which makes it more practical for application in this research. The results of ranking strategies (Table 7) are in line with the findings of Ochieng et al. [28] and Deligios et al. [35], thus confirming the validity of the research model used.

5.1. Research Implications

The results of this research have significant implications for the management of agricultural production in BiH and Serbia, as well as more broadly in the region of Southeastern Europe. Using the results of this research, agricultural policymakers should base the development of agricultural production on a diversification strategy. This would involve cooperation with farmers through providing certain incentives for the introduction of new crops, crop rotation or combining crop production with livestock farming. This strategy has advantages in terms of flexibility and relatively lower investment costs. Therefore, it is necessary to reduce barriers to entry for agricultural producers into diversified agricultural systems and provide them with certain subsidies, support from advisory services and the development of market channels for these products.

Along with the development of diversification, it is necessary to significantly improve water resource management systems. This is required since, due to the specific characteristics of these countries, the likelihood of more frequent droughts is increasing. Therefore, it is essential to work on accelerating investments in irrigation infrastructure and improving these systems to utilize available water resources. In addition to investing in irrigation systems, it is important to develop institutional frameworks for irrigation systems that would provide support to farmers. Based on the results of this research, it is suggested that managers of agricultural enterprises or direct agricultural producers integrate the principles of diversification and efficient management of water resources. This can be achieved by encouraging individual producers to associate, either through cooperatives or other forms of organizations, in order to secure a joint appearance on the market. They can jointly diversify agricultural production, which may include both horizontal and vertical integration. In addition, it is necessary for agricultural producers to begin using digital technologies more regularly, which will enable them to monitor soil moisture and crop conditions. In this way, they could apply a more precise strategy for managing water resources and optimize crop irrigation.

In addition to introducing digital technologies, agricultural producers should also consider implementing regenerative soil management practices, even though this strategy is not at the top of the ranking. This is due to the fact that the implementation of standard agricultural production depletes the nutritional properties of the soil and leads to lower crop yields. In this way, climate change would be responded to in the long term, which would improve the performance of agricultural production in these areas. Thus, multiple strategies would be applied in practice rather than only one, as the effects achieved through these strategies would be improved since a synergistic effect occurs.

In order to implement these strategies in practice, it is also essential to carry out regional cooperation among agricultural producers. This is due to the high degree of agreement in the evaluations provided by experts from the observed countries, suggesting that similar conditions prevail in agricultural production. Therefore, through cooperation, it is possible to first implement cross-border projects of water resource management, then exchange knowledge and experience in terms of implementing certain adaptive strategies for managing agricultural production, and jointly present projects to international funds for financing adaptation to climate change. Given that these countries are in the process of accession to the European Union (EU), they can jointly access the Union’s pre-accession funds, through which agricultural production would be aligned with EU policies.

5.2. Research Limitations

In addition to the contributions offered by this research in terms of considering adaptive strategies for managing agricultural production in the context of climate change, this research also has certain limitations that need to be addressed in future research. When using MCDM methods, it is necessary to determine which criteria and alternatives will be considered in the analysis and how the necessary data will be collected. All of these can be characterized as limitations of this research. First, the selection of ten criteria, although conducted in accordance with previous studies and expert opinions, may represent a limitation. This is because it can always be questioned why these criteria were selected rather than others, as the selection of criteria also affects the selection of alternatives in the decision-making process. However, this paper provided the basis for evaluating adaptive strategies for managing agricultural production. On the other hand, the selection of strategies can also be a limitation of this research, since six strategies were used, whereas in practice, other strategies may also be applied in responding to climate change in agricultural production. In addition, these strategies were viewed as separate units, and combinations of these strategies were not considered. Certainly, the combination of strategies could result in synergistic effects that are not included in this research.

The selection of experts can be considered a limitation of this research since other participants in agricultural production were not included, and they might have provided different assessments compared to professors and researchers, as shown by Stričević et al. [10] in their research. The number of experts may also be considered a limitation of the research since ten experts from each of the two countries participated. However, with an increase in the number of experts, the complexity of the research would also increase, and it is unclear how this would affect the ranking of strategies, as it could lead to inconsistent assessments of criteria and strategies. However, the inclusion of experts such as agronomists employed in large agricultural enterprises could offer a new perspective and provide more balanced assessments. Since the research is geographically limited to two Western Balkan countries, it raises the question of whether these results are representative of other countries in this region. This is due to the fact that each country has specific climate conditions, so perhaps other strategies might provide better results in managing agricultural production in conditions of climate change. In addition, each country has a specific institutional framework, as well as a different level of agricultural production development and the availability of existing agricultural resources in these countries.

Furthermore, the use of the IF2S approach is more complex compared to the usual fuzzy approach and can be considered a research limitation. However, the inclusion of IF2S allows for the inclusion of insecurity and uncertainty in decision-making, which is characteristic of the use of expert opinion. In addition, the use of symmetric membership functions is not in accordance with climate risk, as this risk is often asymmetric. Nevertheless, the use of this membership function simplifies the true nature of uncertainty in expert decision-making. The use of the selected MCDM methods may also be considered a limitation of this research, since the advantage of the M-SiWeC method, that it is not necessary to compare criteria in the assessment process, can also be viewed as a limitation or an advantage, depending on the perceptions of decision-makers. The use of the MABAC method also represents a research limitation, as perhaps other methods might be better in ranking these adaptive strategies.

5.3. Future Research Guidelines

Based on the limitations of this research and the results obtained, guidelines can be proposed for future studies regarding the evaluation of adaptive strategies for managing agricultural production in the context of climate change. First, future research should include not only experts such as professors and researchers, but also farmers of various profiles and other individuals involved in agricultural production. In addition, it is necessary to include a larger number of experts in the assessment of these strategies. When assessing these strategies, future research should also include those not involved in this research and incorporate a larger number of criteria in order to obtain a comprehensive understanding of the characteristics of these adaptive strategies. It is also required to take into account specific segments of the country being observed, as the configuration of the terrain in which agricultural production is carried out may affect the selection of strategies. When selecting strategies, it is necessary to consider their combinations in order to determine which strategies yield the best synergistic effect. Similar research should be extended to other countries in Southeast Europe in order to examine the patterns of acceptance of these strategies. In addition, significant information on how specific characteristics of the countries affect the selection of the strategies could be obtained.

When observing the methodological basis of this research, other approaches can be used in future research, allowing for comparisons among different approaches and the analysis of whether the application of different approaches, while using the same methods, affects the final decision on the selection of the strategy. In addition, it is necessary to use other methods to determine the significance of criteria since, in this research, a comparative analysis was not conducted on the methods for determining the weights of criteria. The comparative analysis carried out in this research showed that the application of different MCDM methods affects the final ranking of strategies, so it is necessary to determine the extent to which the application of a particular MCDM method for determining the weights of criteria influences their weight. It is also important to examine whether some of these methods produce a greater difference and assign greater priority to certain criteria, thus affecting the final ranking of adaptive strategies. In addition, it is required to use other MCDM methods in future research to establish the ranking of strategies. Since the comparative analysis has determined that the normalization used affects the final ranking, it is necessary to examine how each method influences the ranking.

6. Conclusions

This research sought to identify key priorities and provide recommendations for improving agricultural production by applying adaptive strategies. A comprehensive evaluation of six adaptive management strategies for agricultural production in the face of climate change was conducted. The research covered the territory of Bosnia and Herzegovina and Serbia using expert decision-making and the IF2S methodology. A total of ten experts from each of the two countries participated, representing the academic and research sectors. The results of the research, when applying the IF2S M-SiWeC and MABAC methods, show that the diversification and water resource management strategies represent the best courses of action for solving problems in agriculture caused by climate change. The results demonstrate consistency regarding the observed countries, as the ranking of the strategies does not change when considering experts from a particular country. In addition, a comparative analysis indicates that these two strategies provide the best results according to expert assessments. The sensitivity analysis has further confirmed this.

The contributions of this research are as follows: improving theoretical knowledge on the application of regenerative and sustainable adaptive strategies in agriculture in the context of climate change, developing a methodological framework that uses IF2S with symmetric membership functions of fuzzy numbers, which increases accuracy in conditions of uncertainty and uncertainty, providing concrete guidelines for agricultural policy makers on how to solve the problem of the negative impact of climate change, improving knowledge on the application of adaptive strategies and how to respond to climate change with them. The adaptation of agricultural production is not a matter of choice but a reality that all countries face. Therefore, this research represents a step in understanding how agricultural production adapts to climate change.