Spatial and temporal zoning of watershed resilience using a multidimensional composition approach

The present initiative study has been planned to develop a conceptual model for watershed resilience for which rare documents have been reported yet, particularly in developing countries where such studies are necessary. In this vein, different ecological, social, economic, and infrastructural and cultural key domains were applied for the modeling processes of watershed resilience for the Shazand Watershed in Iran. To this end, watershed resilience was firstly conceptualized according to the prevailing conditions for three periods 1986–1998, 1999–2008, and 2009–2016. Accordingly, the watershed health index was used for the ecological dimension, and 13, 8, and 13 criteria were consequently considered for social, economic, and infrastructural and cultural dimensions, respectively. The whole data required for the last three dimensions were collected through the distribution of questionnaires among the stockholders of the Shazand Watershed. The principal component analysis (PCA) method was used to weighting and prioritize the criteria and dimensions. The overall resilience index was ultimately calculated by aggregating all four dimensions using a geometric mean. The effect of each dimension on resilience was also assessed by applying multivariable regression. According to the resilience map of the period 2009–2016, 40%, 9%, 34%, and 17% of the Shazand Watershed has been classified as very low and low resilience, moderate, high, and very high resilience. This study showed that the resilience of the Shazand Watershed has improved over time. The necessity of resilience modeling for practical and integrated management of watersheds was also confirmed during the present research.


| INTRODUCTION
The natural and anthropogenic hazards can change the structure and function of social and ecological processes in watersheds depending on their extent, durations, and frequencies.
Hazards may have a significant impact on watershed systems in the form of disruption of water supply, increase in natural disasters, increase in water pollution, water quality issues, species extinction, biodiversity loss, economic losses, and exploitation (Almutairi et al., 2020;Nemec et al., 2014). Therefore, studying the quality and magnitude of hazards and resilience potential at the watershed scale is of particular importance for the better management of the system. On the other hand, the effects vary depending on the geographical location and condition of the watershed system. For example, due to climate change, some watersheds in South Africa's arid regions face extreme drought and desertification, while others in high-humidity areas like northeastern India face more severe rainfall and floods. These impacts trigger transient or permanent changes in the watershed's social and ecological processes and temporary or permanent changes in the watershed's systems (Duffy et al., 2018;Randhir, 2014).
Since a watershed is a diverse and complicated environment, disruption to its structure and resources causes ecosystems to fail and even collapse. As a result, one of the goals and significant challenges of earth systems science is a proper understanding of the diverse interactions of an ecosystem and associated livening and nonliving components. Watershed management's most valuable feature is the ability to improve the lives of watershed stockholders (Randhir, 2014). Watershed management, on the other hand, necessitates a versatile solution at all times. In this respect, one of the latest approaches to watershed management and environmental risks is resilience. Since threats have impacts on different aspects of human life, and in certain instances due to the depth of disasters, the likelihood of reversibility is entirely weak, the resilience solution focuses on the reliability, rehabilitation, and survival of human communities, as well as reducing instability in sensitive circumstances (Adger, 2006;Djalante, 2012;Wang et al., 2021). As a result, the resilience evaluation method focuses on increasing the potential for accelerated reversibility and self-organization. According to these interpretations, it is a resilient watershed that can withstand the consequences of natural and human hazards at the same time, prevent a certain amount of damage, and also through interaction with the social and human dimensions, adapt to stresses and the capacity to recover after these stresses to maintain itself in the current domains of attraction or to become a new sustainable watershed in ecological and social dimensions.
The resilience approach has recently been integrated as a vital element of the United Nations International Disaster Reduction Strategy. In its vision for disaster prevention, the UN Secretary-General stressed the importance and necessity of reducing inequalities, promoting resilience, and preventing social collapse, increasing the risk of conflict (Saja et al., 2020;United Nations Office for Disaster Risk Reduction, 2015). The international community is not able to tame the natural forces, but it can predict the intensity of the forces to protect the people and make their assets and infrastructure as well as their livelihoods more resilient to hazard adjustment activities (Burton et al., 1978;Kates, 1976;Lindell & Perry, 2000;Ryan & King, 2020). In addition, the identified criteria and dimensions according to the conditions of each watershed can act as a starting point for the participation of stakeholders and experts in the planning and preparation processes, both inside and outside the watershed community. This, in turn, makes it possible to address the various economic, social, and environmental challenges that societies face more effectively. This can also be ensured by creating constructive and quantitative frameworks for assessing resilience so that the resilience approach becomes a "viable strategy" in watershed management (Papp & Alcorn, 2013). Watersheds are complex systems that include many reciprocal subsystems. Meanwhile, resilience as a dynamic process requires a focus on the inherent strength and capacities of the community. Therefore, assessing of resilience approach at the watershed scale by better identifying the dynamic features of this system and their interaction with humans and emphasizing its resources and capacities is necessary for better performance.
In resilience research, to move from a conceptual framework to a practical evaluation, a comprehensive conceptual approach with the use of quantitative data (e.g., rainfall data and drought; Moghadas et al., 2019), appropriate existing data based on previous studies (Cutter & Derakhshan, 2020) and collecting unavailable essential data using questionnaires or formal and informal interviews (Gelcich et al., 2006;Komugabe-Dixson et al., 2019) are used to demonstrate measurable dimensions and criteria. After analyzing the quality, quantity, relevance, and consistency of the available data, the primary criteria for evaluating the resilience index in different dimensions are determined. Designating and developing the questionnaire are made according to the objectives of the research and the subject under study (Ebrahimi Gatgash & Sadeghi, 2022). So the questions are designed in such a way that the relationship between the questions and the desired variables and criteria is adequately obtained (Cutter et al., 2008;Gelcich et al., 2006).
A multidimensional state space that describes all potential combinations of variables in a structure is landscape stability (Bitterman & Bennett, 2016). Walker et al. (2004) used the definition of stability perspective to describe resilience metrics of resistance, latitude, precariousness, and panarchy. The adaptive cycle describes the principle of resilience's evolution essence and suggests that transformation is a cyclical process, but the mechanism never returns to its original state (Bec, 2016). As a result, there are four ecological and social systems: exploitation, conservation, release, and reorganization . To measure community resilience, the community-based model stresses the critical role of local governments and their participation in the natural disaster management process. It is typically used in conjunction with the place-based model (Disaster Resilience of Place [DROP]). The problem with this paradigm is that it focuses mainly on the position of social resilience, which cannot be differentiated from social processes (Gunderson, 2010). The DROP model depicts the relationship between intrinsic resilience (e.g., pre-existing capabilities and properties that enable a society to survive during noncrisis times) and adaptive resilience (capabilities and versatility that allow populations to change and create new responses to postevent problems) in locations (Cutter et al., 2008;Rose, 2009;Tierney & Bruneau, 2007). The BRIC assessment is based on the theory of a place-based model (DROP) for assessing group resilience to natural disasters (Cutter & Derakhshan, 2020;Cutter et al., 2008). All of these types of resilience affect a community's capacity to recover from a disaster. BRIC, on the other hand, only assesses populations' natural stability (pre-event). It does not assess the mechanisms or methods used by populations to cope with, respond to, or deal with accelerated transition in the face of adversity in both short and long-term situations. The Obstacle Degree Model was used to analyze the critical obstacle variables, which provides management reference for the risk aversion approach based on the spatial assessment of social-ecological resilience. In reality, using the Obstacle Degree Model, the dynamics of driving factors influencing social-ecological resilience are explained (Zhang et al., 2020).
Some models have been used to assess community resilience in recent years. However, most models proposed depend on similar factors (such as economic resources, capital, expertise, information, awareness, support and support networks, access to facilities, and shared community values) that can help communities become more resilient to threats, like, natural disasters. In addition, the operationalization of the models discussed emphasizes the conceptual rather than the measurement aspects of resilience, and there is no conceptual framework that makes it possible to evaluate the effectiveness of a nature-based solution option (Mahmood et al., 2021). However, implementing a systematic resilience research approach in diverse and complex watershed systems has yet to be conceptualized and reported, especially in Iran, by taking into account the main dimensions.
Scrutinizing reviewed literature verified that there was no consensus on the indicators explaining resilience, and each study based on a specific approach pointed to separate criteria in different dimensions of resilience. Considering these cases, in examining the resilience index at the watershed scale, a basic need has been proved to design comprehensive dimensions and criteria according to the conceptual framework and theoretical foundations of resilience against natural and human disasters (Duffy et al., 2018;Krievins et al., 2015). In general, international research and strategic approaches involving innovative worldviews can be used to integrate efforts toward the resilience and sustainability of watershed systems. Furthermore, disseminating knowledge and decision-making processes at the watershed scale may further improve resilience to natural disasters. As a result, this research aims to use the main ecological, social, economic, and infrastructure aspects to model the idea of resilience at the watershed scale. The outcome of the present research may provide an appropriate background for future insights into integrated watershed management.

| RESEARCH DESIGN AND METHODOLOGY
In this study, the resilience assessment scheme focused on the inherent characteristics and capacities of the Shazand Watershed with 24 subwatersheds (Sharifi Moghadam et al., 2023). Therefore, an index-based approach was utilized for the current study at the watershed scale. The conceptualization of watershed resilience was achieved for this reason by defining an initial collection of parameters based on watershed circumstances, intrinsic characteristics, and factors influencing the watershed's overall resilience. The required data and information were then collected following the cause-and-effect concept. Then, to estimate the resilience index of the Shazand Watershed and identify the main affecting associated factors, the necessary data and information were systematically collected, augmented, and mined for the appropriate criteria selected according to the prevailing conditions of the studied watershed (Ebrahimi Gatgash & Sadeghi, 2022;Sharifi Moghadam et al., 2023). Due to the multifaceted nature of watershed resilience, four dimensions, ecological, social, economic, and infrastructural and cultural, were identified, prepared, extracted, and data-mined. In the data section required for resilience dimensions, the watershed health (WH) index was used to assess the ecological dimension based on the pressure-state-response (PSR) model. In addition, some 220 questionnaires were filled through interviews with people living in the watershed during the autumn of 2019, and associated data and information for three dimensions, the social, economic, and infrastructural and cultural, were extracted.
Completing the questionnaire at the watershed level was done randomly among the people over 40 years old living in the watershed due to the necessity of information for the last 30 years considered for the study. Consequently, the corresponding criteria were standardized, weighed via PCA, and combined for each different study dimension through geometric mean. The weights obtained in the previous step for the selected criteria in each dimension were then multiplied by the standardized values of the area and population of 24 subwatersheds of the Shazand Watershed to influence spatial and social changes (population) of each subwatershed to evaluate the resilience of each dimension. The resilience of each dimension is then calculated using geometric averaging. The geometric mean was employed to make an average from study variables because it is more sensitive to variations in an individual variable than other averages (Hazbavi et al., 2020;Wiegand et al., 2013). To develop the multivariable linear regression, the first four ecological, social, economic, and infrastructural and cultural criteria were calculated and then combined to obtain the resilience index of the watershed. A multivariable linear relationship was consequently established between four ecological, social, economic and infrastructural and cultural dimensions as independent variables and watershed resilience index (WRI) as a dependent variable whose standardized regression coefficient (β) were employed to find out the priority of determinant criterion affecting resilience index of the Shazand Watershed. WRI was ultimately calculated at the subwatershed level and the entire watershed. The temporal variation of watershed resilience was assessed during three-time spans of 1986-1998, 1999-2008, and 2009-2016. Virtually, the entire study period was subdivided into three snaps to assess the intravariability of watershed resilience as a result of changes in dynamic affecting factors in all study dimensions. Most of the study variables were deemed to change from one period to the another over last 30 years resulting in changes in watershed resilience. The entire procedure and detailed methodology of the assessment of WRI have been shown in Figure 1.

| Application of the framework to the study watershed
The present methodology on the resilience assessment was applied to the Shazand Watershed in the southwest of Markazi Province (49°040′15″-49°520′12″ E and 33°440′42″-34°120′13″ N), in the central plateau of Iran. The total area, including 24 subwatersheds, is 1740 km 2 , of which the mountainous region accounts for about 50%. It is classified as a moderate semiarid to cold semiarid region with hot summers and cold winters. The elevation ranges between 1800 and 3350 m above mean sea level. The mean annual temperature of the watershed, based on neighboring the Arak Station in winter, is −18°C, and in summer up to +35°C, and the annual mean precipitation of 430 mm.
The Shazand Plain has fertile soil, which enables agricultural activities, and consists of five towns and 107 villages, and 10 counties Rural District Over the past decades. Considerable development of urbanization and industries and extension of irrigation and orchard fields was made between 1988 and 2000 (Darabi et al., 2014;Davudirad et al., 2016;Hazbavi & Sadeghi, 2017;Mokhtari et al., 2011) resulted in drastic changes in the social and economic conditions of the area (Darabi et al., 2014;Davudirad et al., 2016;Sadeghi et al., 2018;Sharifi Moghadam et al., 2023); this region is also the essential nature tourism hub in Markazi Province. A general feature of the study watershed and governing conditions are shown in Figure 2.

| Development of a conceptual framework for dimensions' selection
The BRIC framework (Cutter et al., 2010(Cutter et al., , 2014Moghadas et al., 2019) was selected as the theoretical foundation for primary indicator selection since it embraces four comprehensive ecological, social, economic, and infrastructural and cultural dimensions. It is divided into six dimensions: social, economic, institutional, infrastructural and cultural, cultural capital, and environmental factors (Cutter et al., 2010(Cutter et al., , 2014. It recommends a series of metrics for each dimension that can assess the baseline characteristics and current characteristics of communities to educate decision-makers about the overall degree of disaster resilience in F I G U R E 1 Stepwise methodological approach for constructing watershed resilience index. BRIC, baseline resilience dimensions for community; GIS, geographic information system; PCA, principal component analysis. the region. The criteria were summarized and analyzed in four dimensions of ecological, social, economic, and infrastructure in the current analysis, based on this context, according to the prevailing circumstances of the watershed under study, rather than the six dimensions in the BRIC framework. The explanation for the measurements in this analysis from six to four was to avoid scattering and duplication of the criteria in each dimension and adhere to the data parsimony principle in the collection of criteria and dimensions. In this framework, ecological resilience is related to the ability of an ecosystem to maintain structure and function in the face of environmental disturbances (Cutter et al., 2010;Folke et al., 2004). Social resilience is also the ability of communities to cope with social, political, and environmental disturbances and pressures (Perrings, 2006). Further, economic resilience reflects the community's viability, such as capital, housing, income, and employment. Reviewing the economic dimension allows the analysis of links that increase or decrease economic stability, particularly livelihood and welfare sustainability at the community level (Briguglio et al., 2006;Eftekhari & Sadeghlo, 2017). In the field of infrastructure resilience, the community response and recovery capacities such as shelters, housing units, communication routes, and health facilities are evaluated (Norris et al., 2008).

| Identification of valid criteria
Although this research adopts the BRIC as the basis for the composite indicator selection, it is not limited to the individual indicators presented by Cutter et al. (2010Cutter et al. ( , 2014. For this purpose, based on existing literature, the relevant, robust, and representative criteria were identified. In addition, the availability and scalability of data at the watershed scale were carefully checked. So, indicators without one of the mentioned metrics were eliminated from the preliminary list, and the final candidate set of individual indicators was virtually defined as summarized in Figure 3.

| Refining and preparing data
To analyze the data collected through the questionnaire, first, the relevant questions and periods were coded, and more than 30,000 items were entered into SPSS software. Then, Cronbach's α index was used to evaluate the questionnaire's reliability and assess the validity of the answers provided in the study items (Kleier et al., 2018;Moon, 2010). Table 1 displays the results of the reliability test. Given that the α index in the second column was more than 0.8 for all three dimensions, it is evident that the questions are internally consistent, and the questionnaire's dependability is deemed acceptable.

| Identification and weighting resilience criteria using PCA
The PCA as a multivariable statistical method was used to reduce the complexity of analyzing primary variables and better interpret information (Bai et al., 2022;Çamdevýren et al., 2005;Cozzi et al., 2019). The primary variables were converted to new and independent components (with zero correlation coefficients for both components) and then used instead of the primary variables. New components were considered as linear combinations of primary variables (Liu et al., 2003). In addition, the minimum losses of data and information of the primary variables and original data were considered by the resulting components not to miss the valuable formation of the study components (Mohan & Arumugam, 1996). The data adequacy test of the Kaiser-Meyer-Olkin (KMO) and the feasibility of selecting appropriate criteria using the Bartlet test were used to evaluate the possibility of performing PCA on the data of criteria in three study dimensions (i.e., social, economic, and infrastructure).

| Visualization and mapping
To better understand and create comparative levels of the overall resilience index, the spatial distribution of the WRI was visualized in 24 subwatersheds of the Shazand Watershed. Before visualization, the scores of the four dimensions were categorized into five main classes using ARC GIS10.6 to identify the spatial patterns of watershed resilience. This method transformed the scores to the standard deviation from the mean. Subsequently, three resilience domain maps and the final map for WRI were produced for three-time spans of 1986-1998, 1999-2016(Hazbavi et al., 2020Mirchooli et al., 2020;Moghadas et al., 2019).

| Selection of a conceptual framework and identification of valid criteria
As noted in the previous section, the BRIC approach was chosen as the conceptual framework for assessing watershed resilience in the current study. The approach can provide a resilience index based on different criteria to assess the current level of resilience by considering both characteristics and capacity of the place in four main ecological, social, economic, and infrastructural and cultural dimensions.
To move from a conceptual framework to practical evaluation, measurable individual criteria were then determined in three time periods (1986-1998, 1999-2008, and 2009-2016). Following an analysis of the quantity and consistency of available data, the initial criteria for assessing the resilience index of the Shazand Watershed was established in four dimensions: ecological (1 criterion), social (13 criteria), economic (8 criteria), and infrastructure (13 criteria), for the entire 34 criteria. Figure 4 shows the primary sets of criteria with a data source for the Shazand Watershed. In this vein, the WH index was considered for the ecological dimension resulting from the application of PSR for the same watershed. Thus, the WH concept was described using a PSR framework. To do so, 17 climatic, anthropogenic, and hydrologic criteria were used to conceptualize the watershed indices of pressure, condition (S), and response (R) (Ebrahimi Gatgash & Sadeghi, 2022;Hazbavi et al., 2019).
For the social dimension of the resilience index, 13 criteria were selected according to the capacities related to the texture of different population groups in different subwatersheds (Figure 4). According to what was stated, in this dimension, according to the limitations and needs of the Shazand Watershed residents, most of the items such as education, awareness, participation, and cooperation of local people about disasters (S 1 , S 2 , S 3 , S 4 , and S 11 ), emotional and financial support for people at risk (S 5 , S 6 , S 7 , and S 8 ), dependence on the place of residence (S 10 and S 13 ), and quality of life of people in the community (S 9 and S 12 ) in the face of natural disasters were assessed. Furthermore, in the economic dimension, according to the economic capability of the community as well as the performance of watershed residents to achieve the desired situation after the shock, eight criteria such as employment status (E 1 , E 2 , and E 8 ), economic indicators (E 3 and E 4 ), women's participation (E 5 ), and the tourism situation in the watershed (E 6 and E 7 ) were considered (Figure 4). Finally, 13 unique criteria were considered for the resilience of the infrastructure sector as the response and recovery capacity of the community as well as the number of assets that may be at risk (Bruneau et al., 2003;Cutter et al., 2010Cutter et al., , 2014 Figure 4). Therefore, issues such as critical public facilities (I 1 , I 2 , and I 3 ), ease of access to infrastructure facilities (I 4 , I 8 , and I 12 ), the impact of industrial expansion in the region (I 6 ), implementation of land use laws (I 7 , I 8 , I 9 , and I 10 ) and public relations to convey the news of infrastructure weakness to the executives (I 11 and I 13 ) were examined.

| Data reduction and weighting criteria
The data adequacy test (KMO) and the possibility of choosing appropriate criteria were performed through the Bartlet test, as indicated in Section 2.5, to evaluate the possibility of performing PCA on the criterion data in each dimension. The results for 1986-1998, 1999-2008, and 2009-2016 and different dimensions of studies are shown in Table 2.
Since the KMO statistic was greater than 0.5 (Table 2), the PCA could apply to the original data (Shirsath & Singh, 2009). The Bartlet test also confirmed that the observed correlation matrix belongs to a community with uncorrelated variables. At this stage, the effect of each criterion in estimating the relevant component was calculated after implementing the primary component test by periods in social, economic, and infrastructural and cultural dimensions. Then, each criterion was placed in a component with which there was a significant correlation. After the factor load of each criterion was determined using PCA, the factor loads of each criterion were multiplied by the ratio of the total variance explained for each factor load associated with the criterion, then to find the weight of each criterion in all dimensions, the ratio of each of the secondary values to the sum of the values were calculated, the result of this process was the weighting of each criterion in the overall relevant dimension. Finally, the criteria code and its priority are presented in Figure 3, respectively, and information about the names of the new criteria is given. Besides, in this stage, by the sum of each of the weights of effective criteria in each component, the final weight of the extracted criteria was obtained. Then the nomination of these criteria was done based on the nature of the criteria that make up the component.
As seen in Figure 3, the results of weighting by the PCA method revealed that in all three periods of the analysis, the criteria of knowledge and understanding of local populations, as well as their rationale about how to cope with natural disasters, received the highest weight (i.e., 1986-1998, 0.552; 1999-2008, 0.548; and 2009-2016, 0.540). However, the lightest criteria in the area of social dimension resilience in the period 1998-1998 were health services (0.044), welfare (0.125), and immigration in the period 2016-2009 (0.039).
According to Figure 3, the criterion of livelihood diversity and tourism (0.534) had the highest weight in the economic dimension of resilience from 1986 to 1998; the criterion of economic growth and productivity diversity (0.832) had the highest weight from 1999 to 2008, and the criterion of economic growth and tourism (0.450) had the highest weight from 2016 to 2009. The lowest weight of economic resilience was assigned to the economic growth criteria (0.245), then to women's economic participation (0.167), and finally to livelihood diversity and economic prosperity in the third era (0.224).
In the resilience of the infrastructure component in the Shazand Watershed, the most important was related to infrastructure construction and enforcement of land use laws and natural resource security (0.910) in the period 2009-2010, and the criteria of infrastructure and protection initiatives with a weight of 0.435 and 0.469 in the periods 2008-1999 and 2016-2009. In infrastructure resilience, the industry's overall development had less weight in all three time periods, with 0.032, 0.137, and 0.060, respectively.
To obtain the resilience of the desired dimension, the results from this section were combined with the weights obtained from method PCA for each criterion in each dimension, which is discussed in the following chapters. 3.3 | Combining criteria and evaluating the resilience of different dimensions Table 3 shows the results of subwatershed scale temporal and spatial resilience variations in the Shazand Watershed produced by multiplying the weighted values of the criteria in the standardized values of area and population. The resilience of each dimension in different subwatersheds of the study area was then obtained through geometric averaging. The explanation for using the geometric mean to combine the data in the preceding steps is that the values used were rated of different degrees, and each number reflects a degree of watershed resilience, so this procedure provided the best result in the end to reflect the actual value of the outcomes. Table 3 shows changes in the Shazand Watershed's resilience index in ecological, social, economic, and infrastructure dimensions throughout three periods in 24 subwatersheds. According to the results, the resilience of different study dimensions and subwatersheds varies depending on the area and population of each subwatershed. However, it is also clear that the large area or large population (such as subwatershed 16) cannot be a particular determinant reason for the lack of resilience because the existence of proper infrastructure, ecological balance, economic stability, and social cohesion, if potent, no matter how large the area and large population are. It means that if these resources are appropriately managed, resilience will be maintained. This finding can be confirmed in subwatersheds 1-3, in which despite smallness in terms of area and population, the associated resilience in ecological, social, economic, and infrastructural and cultural dimensions is very low.

| Evaluation of overall resilience for the Shazand Watershed
The resilience of the whole Shazand Watershed in different subwatersheds in three time periods (i.e., 1986-1998, 1999-2008, and 2009-2016), the four dimensions studied were combined through the geometric mean method whose results are shown in Table 4.
After calculating the resilience index in ecological, social, economic, and infrastructure dimensions, it was necessary to calculate the resilience of the whole Shazand Watershed, which is obtained by combining these four dimensions (Table 3) to create a way for comprehensive management of the Shazand Watershed (Table 4). This multidimensional composition approach provides a tool to integrate qualitative assessment into quantitative analysis. The results in Table 4 provide a better overview of resilience levels at the subwatershed scale and highlight where insight management strategies are essentially needed.

| Modeling the linear relationship between resilience and different dimensions
To identify the most critical dimension in measuring the Shazand Watershed resilience, a linear relationship was established between the four study dimensions (i.e., ecological, social, economic, and infrastructural and cultural) as independent variables and the resilience index as a dependent variable. For this purpose, a multivariable linear regression method was used, whose results are summarized in Table 5. As seen in Table 5, all study dimensions did not enter the regression equation as independent variables. So, in the first and the second time spans the social dimension, and in the third period, the infrastructural and cultural dimension did not enter the final multivariable regression and was supposed as an excluded variable. This demonstrated that, as opposed to other dimensions, the social and infrastructural and cultural dimensions had no substantial impact on the Shazand Watershed's resilience during the periods studied. However, the impact of other dimensions in different time spans on watershed resilience was that in the periods 1986-1998 and 1999-2008, the highest impact was related to the economic dimension with a relative contribution of about 66% and 58%, respectively. Then the infrastructure dimension with a relative contribution of about 29% and 39%, respectively, and the lowest contribution of some 5% were related to the ecological dimension in both periods, respectively. Nonetheless, in the third period (i.e., 2009-2016), the economic, social, and finally ecological dimensions with relative participation of about 60%, 37%, and 5%, were, respectively, prioritized in the viewpoint of their contributions to the resilience index of the Shazand Watershed, Iran. In line with this conclusion, Qiang et al. (2020) in their study to investigate the impact of environmental, socioeconomic, and industrial structure variables in the formation of recovery and resilience patterns after Hurricane Katrina through a multivariable regression test, the role of socioeconomic variables was stronger.

| Mapping of an overall resilience index
Follow-up measures were taken to understand better and create relative levels of the overall WRI. The spatial distribution of WRI was mapped for 24 subwatersheds of the Shazand Watershed based on the results obtained for four study dimensions ( resilience of the Shazand Watershed for three time periods was mapped as shown in Figures 5-7. Mapping of the obtained results illustrated the distinct spatial patterns of the drivers of resilience over three study periods and identified the hot spots of watershed resilience in the study area. For instance, in all three time periods, subwatersheds 1, 2, 3, 13, and 24 were less F I G U R E 5 Resilience index map in the period 1986-1998 for the Shazand Watershed, Markazi Province, Iran resilient and, therefore, need more attention. The underlying criteria are mostly ecological and economic attributes directly related to education and awareness, social capital, migration, land degradation, and infrastructural and cultural planning in these subwatersheds. However, an improving trend was also found in WRI for some subwatersheds. Moghadas Hazbavi et al. (2020) by mapping the results, showed clear spatial patterns of resilience and health indices, respectively, and identified hotspots in the study area that required further intervention.
According to the latest resilience map, that is, for the period of 2009-2016, subwatersheds 1, 2, 3, 13, and 24 with a total area of about 8% have very low resilience and require special attention from local managers and decision-makers. Likewise, subwatersheds 5, 6, 7, 8, and 17 have low resilience levels and can be the next planning priority for deficit management. Subwatersheds 4 and 22 have moderate resilience and can be improved at a lower cost with timely action. Also, the 9, 10, 11, 12, 15, 18, 20, and 21 subwatersheds, with about 31% of the basin area, have a high resilience F I G U R E 7 Resilience index map in the period 2009-2016 for the Shazand Watershed, Markazi Province, Iran compared to other basins. Finally, the 14, 16, 19, and 23 watersheds were evaluated as having a very high resilience according to the measured criteria. Among the maps of resilience dimensions, ecological resilience had a distinctive pattern. It showed a gradual decrease in ecological resilience levels over three study periods. Ecological resilience was a fundamental necessity in building the capacity for communities and individuals to prepare for, respond to, recover from, and adapt to the impacts of a disaster. At the same time, if appropriately performed, it may increase the resilience of other dimensions. Thus, the local stockholders could consider this to promote an equitable development process and provide fair access to critical watershed resources. Hazbavi et al. (2020) also confirmed in a study the trend of the declining health of the Shazand Watershed over time. In addition, there was a decreasing trend in resilience over time in the economic dimension, which was primarily due to economic instability and lack of livelihood diversity in the watershed. In this regard, Sharifi Moghaddam et al. (2019) showed that considering that the profession of most people living in the Shazand Watershed is agriculture, the water-energy-food nexus approach can help poverty reduction and generate employment opportunities among farmers to take steps toward economic resilience for residents. Sarker et al. (2020) also confirmed in a study on livelihood resilience in Bangladesh that the expansion of social safety net programs and coordinated participation of local government, NGOs, and public-private partnerships are vitally important to enhance the livelihood resilience of the char dwellers.
In the social dimension, due to the interaction between locals over time, the resilience of this dimension had not changed much, but the decrease in the resilience of some subwatersheds had been primarily due to a lack of knowledge and awareness and particularly migration. A strong agreement was found between the present findings and Davudirad et al. (2016) and Sadeghi et al. (2018), who notified the effects of human migration to the study region during the Iran-Iraq war.
In the infrastructure dimension, attention to industrial infrastructure and the availability of welfare and health facilities, and increasing access to communication and Internet networks have increased resilience in this dimension during study periods. This conclusion was also reached in the temporal and spatial changes in disaster resilience in US counties from 2010 to 2015, as reported by Cutter and Derakhshan (2020). In the study of Aguilar et al. (2022), physical and natural assets in the form of irrigation infrastructure and direct access to water resources were significantly associated with overall resilience (avoidance and adaptation) to water scarcity. Finally, based on the experiences of various studies (Whitfield et al., 2019), it can be emphasized that the resilience (or lack of) of different systems is partly the product of longterm historical trends as well as short-term shocks over time. The political, ecological, economic, and social lock-ups and dependencies that emanate from past experiences all contribute to the resilience of the contemporary system. It is therefore suggested that disaster shock experiences can provide a window of insight into future ecosystem responses and, when combined with historical perspectives and learning from a variety of contexts, can provide a foundation. It is important to try to develop appropriate long-term resilience strategies to mitigate the effects of change and uncertain conditions.

| Limitations and implications of the study for the managerial purposes
As mentioned in the previous sections, to provide a correct view of the resilience of a system, it is necessary to conduct an integrated analysis using various political, security, economic, social, environmental, and interactions. However, the lack of awareness of the importance of integrated approaches in watershed management has limited the possibility of employing multidimensional system approaches resulting in a lack of appropriate data. Consequently, some criteria should be omitted due to lack of access or unavailability of data and information, which may affect the comprehension of the results of such studies. Along the same line, institutional or political resilience, which plays essential roles in the preparation and planning stage for resilience to natural disasters, could not be elaborately considered in the present study solely because of the limitations mentioned above and the confidentiality of the information. However, adequate attempts were made to consider as many as possible factors in different dimensions, including the social aspects. In addition, because the validation of disaster resilience studies is often problematic due to the lack of information about the recorded effects of past natural disasters, validation based on the actual results of the previous disasters was not possible. Of course, the subjectivity and watershed-specific criteria for evaluating the resilience and weighting techniques can also be supposed to have other limitations for such studies. Despite existing limitations and barriers in the studies on watershed resilience, health, and sustainability, the results of such studies provide valuable platforms and motivations to create strategic management plans to handle controllable stressors or to strengthen the watershed systems based on spatial distribution and trend of variations in the resilience index throughout the watershed and during the lifespan under consideration. As a result, using a resilience index assessment technique helps ensure that stakeholders have the essential capability to endure and recover from a stressful event. When dreaming about stress, also assists decision-makers in shifting from reactionary to adaptive techniques. Also, despite limitations and obstacles in studies related to the evaluation of watershed resilience as shown in the current research, the findings of such studies can provide a valuable basis and incentives for developing strategic management plans to curb controllable pressure factors and strengthen watershed systems based on spatial distribution and changes over time. But, to evaluate the resilience of the watershed with high precision and accuracy, some other important criteria, such as risk perception, and early warning systems, according to the conditions of the watershed and the available data, would also be considered in future studies to effectively deal with risks by recording past experiences and ultimately improve the watershed resilience.

| CONCLUSION
As this study showed, combining four critical dimensions of resilience assessment for the Shazand Watershed in Iran verifies different statuses of the watershed. The preparation of the WRI map also proved the spatial pattern of the resilience for the study region at the subwatershed scale. Therefore, as a practical solution and based on interviews with watershed residents of the Shazand Watershed and also the results of this study, it can be suggested, considering that the resilience of this area, improving the resilience of the economic dimension can have a significant effect on capacity building of the region. Therefore, providing opportunities for increasing job diversity and education and participation of women in helping the family's livelihood and investing in the tourism sector of the constituency, to a large extent, leads to the resilience of the constituency's residents. However, according to the residents, even with employment and the increase of local production units, the possibility of reverse migration to the Shazand Watershed has intensified in recent years. Therefore, one of the reasons for the reduction of social resilience is possible. Given the above, along with the implementation of community-based natural resource projects, the intensity of use of water and soil resources, which has been one of the reasons for reducing ecological resilience in the Shazand Watershed, has been reduced, and steps will be taken to address ecological resilience and increase social resilience in the development of these projects and, consequently, bio-sustainability. Thus, as an emerging approach, resilient thinking can better understand the watersheds as complex and dynamic systems between people and nature and guide different values and interests. Capacity building for disaster adaptation by decision-makers and expansion of virtual space to increase awareness against natural disasters, considering the importance of the Shazand Watershed as the hub of agriculture and tourism in the province, improving the special position of the study watershed in terms of important industries, such as Imam Khomeini Refinery, Petrochemical, and Shazand Thermal Power Plants, and improving health care, water supply, sanitation, irrigation and drainage systems might be among the main reasons for improving the resilience of the Shazand Watershed during the studied 30 years. Contrary to watershed management approaches based on minimizing or controlling change, the resilient thinking approach is based on the principle that promotes the value of thinking about complex systems and being localized according to the national and local needs of different watersheds and being sensitive to the different conditions and characteristics of poor watersheds. Considering that this research has been conducted for the first time in Iran and so far the experience of facing disasters has not been documented to measure resilience, so the validation of the current research is more qualitative. On the other hand, according to the interaction and social interaction that the author's team had with the residents in all subwatersheds and the informal interviews that were conducted, it could be confirmed that the resilience distribution in the watershed is consistent with the obtained results. However, further insight monitoring of resilience indices and particularly in connection with WH and services using objective criteria are advised to be taken into account in future studies.
AUTHOR CONTRIBUTIONS Parisa Farzi: Data curation (equal), formal analysis (equal), investigation (equal), methodology (equal), software (equal), and writing-original draft (equal). Seyed Hamidreza Sadeghi: Conceptualization (lead), formal analysis (supporting), investigation (supporting), methodology (lead), project administration (lead), supervision (lead), writing-original draft (supporting), and writing-review and editing (lead). Mahmoud Jomehpour: Conceptualization (equal), methodology (supporting), and supervision (supporting). All authors contributed to the study's conception and design. Material preparation, data collection, and analysis were performed by Parisa Farzi. The first draft of the manuscript was written by Parisa Farzi and Seyed Hamidreza Sadeghi commented on the original versions of the manuscript. All authors read and approved the final manuscript. Seyed Hamidreza Sadeghi reviewed and edited the final manuscript.

DATA AVAILABILITY STATEMENT
The data sets used and analyzed during the present study are available from the corresponding author upon reasonable request.