Visualizing Sustainability of Selective Mountain Farming Systems from Far-eastern Himalayas to Support Decision Making

Mountain farming systems rely on both empirical and academic knowledge. Their sustainability depends on how effectively diverse knowledge is used for solution-oriented decision making. For mountains, decisions must be conducive to rural farmers whose livelihoods depend on agriculture and related activities. Adopting transdisciplinary research approach, we define a composite Sustainability Space indicator that will help decision makers better understand the ingredients for sustainability, and formulate policy and management decisions to reinforce on-the-ground sustainability. Sustainability Space was derived through analysis of the positive and negative impact factors co-defined by community and disciplinary experts, and visualized through a radar diagram. We used Principal Component Analysis to understand relationships between factors. The results on Sustainability Spaces for eight cases of farming systems from the far-Eastern Himalayas indicated that the sustainability of farming systems is strengthened if decisions holistically cater to (i) geophysical pre-requisites, (ii) ecological foundations, (iii) integrated processes and practices, (iv) resources, knowledge, and value systems, (v) stakeholders’ development and economic aspirations, (vi) well-being of farming communities, and (vii) government support mechanisms. More equitable the attention to these seven components, the higher the sustainability of farming systems in this region could be.


Introduction
Mountain farming systems are socio-ecologically dynamic. They are land spaces that define and direct people's aspirations, interests, and motivations [1], and are oriented by the ways people manage diverse resources and implement management practices within their respective socio-cultural and economic value systems [2]. For more than 70% of the rural population in the hills and mountains of South Asia, these land spaces represent not only a cultivated space, but a way of life, their knowledge system, and their means of livelihoods [3]. However, Mountain farming systems are transforming [4][5][6][7] with influence from constantly evolving global discourses on poverty reduction, natural resources management, biodiversity conservation, climate change, green revolution, sustainable intensification, globalization, and trade liberalization; along with a wide range of thematic disciplines concerning forest, soil, water, biodiversity, economics, and politics-all directing their functional pathway [7].
The sustainability context for agricultural systems picked up momentum as discourses on conservation, ecosystem services, food security, nutrient security, resilience, and climate adaptation pitched up in the global agenda [8]. Additionally, sustainability of agricultural systems was regarded as critical in the context of its "multi-functionality, changing knowledge paradigm, and social aspirations and influence from drivers of change whose future intensity and impacts remain unpredictable" [9]. Sustainability, while it needs to consider current context of being more socio-culturally equitable, environmentally sound, and economically viable [10], also has to factor in the concept of being more resilient and performing in the future [11]. There have been concerns about sustainability assessments or sustainability oriented knowledge not adequately guiding farming communities manage their farms, and other stakeholders make relevant research, management, and policy decisions [12,13]. We attempt to particularly fill this gap of appropriate knowledge interpretation and visualization of on-the-ground farms situations and farmers knowledge for decisions makers.
In this paper, we emphasize on co-defining indicators for sustainability based on the interactions between researchers who want to generate assessment-based knowledge on sustainability and other actors who make various research, management, and policy decisions at different scales [14]. We considered the key essence of transdisciplinary research [13,15] to evaluate sustainability through wider stakeholders' engagement, and make sustainability research results comprehensible and applicable to key stakeholders-particularly farming communities and policy makers. We used a composite indicator-referred here as Sustainability Space to translate farm performances into a measure for sustainability that policy makers can refer to-to facilitate appropriate management and policy decisions for sustainability. Exploring the Sustainability Spaces of selective farming systems from the far-Eastern Himalayan countries, allowed us to also understand the interplay of interdisciplinary factors affecting sustainability.

Materials and Methods
The steps in the following sections were deemed essential to define indicators that would legitimize actionable knowledge generated by communities at the local scale, and then enable decision makers relate to these in the form of broader-level decisions-ultimately aligned to address sustainability issues on the ground.

Co-defining Impact Factors
A preliminary survey was conducted in December 2016, using two sets of open-ended questions: (i) What three factors positively influence farming systems or make it better? Additionally, (ii) what three factors negatively influence farming systems or deteriorate it?-to capture factors influencing sustainability. We approached two tiers of experts: (i) Farmers and the local and traditional institutions who bring empirical and traditional knowledge were considered as Community experts, and (ii) academia, government decision makers, and stakeholders belonging to business, regional, and international firms who bring thematic discourse-based knowledge and policy decisions were considered as disciplinary experts. Disciplinary experts were reached through online survey (www.limesurvey.com) and community experts through village level key informant interviews. This exercise was crucial because wider the stakeholders' diversity in thematic knowledge (for disciplinary experts) and experiences with agricultural systems (for community experts) more comprehensive are the inputs to the impact factors. Literature also emphasizes using multidimensional indicators [16,17]. We collated responses from 100 disciplinary experts and 50 community experts and organized their responses into a set of 21 Positive Impact Factors (PIFs) and 16 Negative Impact Factors (NIFs) relating them to the three pillars of sustainability and other cross cutting pre-requisites, as shown in Table 1. Since the intention was to enumerate factors that stakeholders regarded as important for sustainability, the factors were not weighted or ranked, and all were assumed to be valid and of equal importance as they reflected experts' individual values and knowledge.

Categorizing PIFs and NIFs into Components of Composite Indicators
To provide decision makers and administrators an analytical overview of the sustainability of different farming systems and highlight areas where government interventions needed to be focused, a composite indicator referred to here as Sustainability Space was developed. It comprises seven sustainability space components-Space Organization (SO), Resource Efficiency (RE), Integrated Approach (IA), Adaptive Features (AF), Economic Prospect (EP), Social Well-being (SW), and External Support (ES). These seven components were drawn to collectively answer questions related to  [18]; what interdisciplinary actions were necessary? [19]; what builds and ensures adaptive features and resilience? [20]; what agribusiness infrastructure and services were necessary for economic viability? [21]; what societal infrastructure and knowledge base were important? [22]; and what government supports and services were necessary? [23]. These questions also reflected upon the logic experts' survey responses were sorted into. A multi-criteria influence exercise was carried out with 15 disciplinary experts to determine how each PIF and NIF could be related to these seven sustainability space components (Appendix A). This exercise led us to categorize the PIFs and NIFs into respective sustainability space components ( Figure 1).

Categorizing PIFs and NIFs into Components of Composite Indicators
To provide decision makers and administrators an analytical overview of the sustainability of different farming systems and highlight areas where government interventions needed to be focused, a composite indicator referred to here as Sustainability Space was developed. It comprises seven sustainability space components-Space Organization (SO), Resource Efficiency (RE), Integrated Approach (IA), Adaptive Features (AF), Economic Prospect (EP), Social Well-being (SW), and External Support (ES). These seven components were drawn to collectively answer questions related to key issues in agricultural systems development: what land use decisions were necessary? [18]; what interdisciplinary actions were necessary? [19]; what builds and ensures adaptive features and resilience? [20]; what agribusiness infrastructure and services were necessary for economic viability? [21]; what societal infrastructure and knowledge base were important? [22]; and what government supports and services were necessary? [23]. These questions also reflected upon the logic experts' survey responses were sorted into. A multi-criteria influence exercise was carried out with 15 disciplinary experts to determine how each PIF and NIF could be related to these seven sustainability space components (Appendix A). This exercise led us to categorize the PIFs and NIFs into respective sustainability space components ( Figure 1).

Defining Farm-Performance Indicators
The objective was to bring out a balanced set of indicators [24] that could depict farm sustainability performance-that is their strengths and challenges with respect to the PIFs and NIFs as these would either enhance or hinder sustainability. Thus, a set of 74 farm performance indicators were defined within the PIFs (Table 2). These were distilled from the community experts' survey responses on positive factors affecting sustainability. Community knowledge have been considered dynamic [25], and are valuable to reorient modern agriculture towards a more sustainable and resilient development pathway [26]. Framing of Sustainability Space building upon the farm-level determinants was thus to underscore role of primary stakeholders or the mountain farming communities in maintaining their respective farming systems; and to acknowledge their experiencebased knowledge. Disaggregation of PIFs was necessary to make farmers' day-to-day work-related

Defining Farm-Performance Indicators
The objective was to bring out a balanced set of indicators [24] that could depict farm sustainability performance-that is their strengths and challenges with respect to the PIFs and NIFs as these would either enhance or hinder sustainability. Thus, a set of 74 farm performance indicators were defined within the PIFs (Table 2). These were distilled from the community experts' survey responses on positive factors affecting sustainability. Community knowledge have been considered dynamic [25], and are valuable to reorient modern agriculture towards a more sustainable and resilient development pathway [26]. Framing of Sustainability Space building upon the farm-level determinants was thus to underscore role of primary stakeholders or the mountain farming communities in maintaining their respective farming systems; and to acknowledge their experience-based knowledge. Disaggregation of PIFs was necessary to make farmers' day-to-day work-related indicators more explicit to them and to increase their awareness about interdisciplinary factors positively affecting their farms. The intention was to place a set of performance indicators that community experts could relate to and use them to analyze their farm condition in the present, and in the future. NIFs were not disaggregated further as they adequately captured on-the-ground challenges indicated by the surveyed community experts. The 74 farm performance indicators within PIFs and 16 NIFs used in this research for community-based exercise are by no means complete. Survey with wider farming communities in different areas could refine them further.

Selecting Sites for Assessing Sustainability
Eight types of farming systems were selected in four countries in the far-eastern Himalayas ( Figure 2). The intent was to include at least one representative sample of a farming system within the traditional-commercial spectrum. The selection process also considered access to the field site; availability of financial resources; availability of country leads to facilitate community-based exercises; and time for both community and disciplinary experts.

Selecting Sites for Assessing Sustainability
Eight types of farming systems were selected in four countries in the far-eastern Himalayas ( Figure 2). The intent was to include at least one representative sample of a farming system within the traditional-commercial spectrum. The selection process also considered access to the field site; availability of financial resources; availability of country leads to facilitate community-based exercises; and time for both community and disciplinary experts.

Participatory Exercises in Different Sites to Define Sustainability Space
A cross-section of farming community members (women group, mixed group, elderly group, younger farmers group, local governing bodies and traditional institutions) totaling 530 community experts from 26 villages ( Figure 2) were engaged in the participatory ranking exercise. Community groups were formed by the village heads in respective villages. Community members in a group were asked to discuss and provide scores between 5 and 1 on each of the 75 PIFs and 16 NIFs indicators where 5 indicated best and 1 indicated unsatisfactory current condition for PIF. For NIFs, 5 indicated critical and 1 milder condition. For each site, the scores for all PIFs and NIFs indicators within each of the sustainability space component were summed up and converted to percentile score. The final performance score for each sustainability space component was derived by subtracting the total NIFs score from total PIFs score. The scores for each type of farming system were then graphically (MS Excel) plotted to develop radial charts. The assumption was that more balanced or bigger the circumference of the space, higher the system's sustainability. The balanced condition represented that all seven components were given adequate attention and had good current  performance score for each sustainability space component was derived by subtracting the total NIFs score from total PIFs score. The scores for each type of farming system were then graphically (MS Excel) plotted to develop radial charts. The assumption was that more balanced or bigger the circumference of the space, higher the system's sustainability. The balanced condition represented that all seven components were given adequate attention and had good current conditions. Greater circumference indicated that all PIFs and NIFs within the seven sustainability space components were proficiently managed-PIFs were sustained and NIFs were addressed. Since sustainability spaces were built using all PIFs and NIFs, interdisciplinary strengths of farm performance indicators were not diluted or compromised while constructing seven sustainability space components-thus, the sustainability space composite indicator.

Principal Component Analysis
Principal component analysis (PCA) was used to extract commonalities between the: (i) Farming system sites and sustainability spaces, (ii) farming system sites and Sustainability Space Components, (iii) farming systems sites and PIFs and, and (iv) farming system sites and NIFs. The analysis was conducted using SPSS v. 25. Prior to conducting the PCA, the data were tested for suitability using Kaiser-Meyer-Olklin (KMO) and Bartlett's Test of Sphericity. The extracted components were rotated using Oblimin with Kaiser Normalization to obtain the significant components. Eigen values of greater than 1.0 were used to reduce the number of variables and Scree plots were used to identify the number of components that best described the variables. Bi-plots were used to make result visualization easier for the decision makers. We opted for PCA because it is one of the most commonly used methods of reducing the dimensionality of data sets as those collected in this study. This is the oldest method used for large data sets, that allows for increasing the interpretability of relationships between variable, while minimizing information loss [27]. We regarded that other methods such as correspondence analysis and canonical correlation analysis are only loosely connected to PCA as they are based on factorial decompositions of certain matrices, although they share a common approach with PCA. Therefore, we concluded that PCA fit our purpose of analyzing our data set to enable decision makers interpret relationships between the interdisciplinary factors in our study.

Sustainability Spaces of Different Farming Systems
The sustainability spaces as visualized through a radar diagram ( Figure 3) indicated the extent of balance that currently existed among the seven sustainability components in the different farming systems studied. More uniform or bigger radial space points to a greater sustainability as it meant holistic interventions and comparatively balanced attention to the seven sustainability space components. Altogether, the sustainability spaces reflected the extent of the use of physical spaces; use of ecological and socioeconomic resources; acknowledgement, use, and promotion of local and traditional value systems; prospects for livelihoods and economic development; application of integrated farming processes and strategies; and extent of government support mechanisms.   Considering the cumulative scores of all seven sustainability components, it was noted that the Apatani heritage paddy-cum-fish farming of Ziro Valley (Figure 3a) and integrated mixed farming system of Bhutan (Figure 3f) had greater potential for sustainability with bigger and balanced radial spaces. This implied to balanced attention to all seven sustainability space components. Comparatively, the Lopil-based shifting hill agriculture of Chin Hills (Figure 3c) had uneven and smaller radial spaces, indicating lesser attention to Economic Prospects, External Support, Integrated Approach, and Social Wellbeing components. Looking at the PCA bi-plot (Figure 4), the clustering of the Traditional Shifting Cultivation of Nagaland (SCNL) and Transforming System of Dima Hasao (TSCDH) indicated that these two farming systems were similar in sustainability space scores. Likewise, the Lopil-based shifting hill agriculture (SCCH), Sikkim Organic Farming (CBOSK), and Apatani Community heritage paddy-cum-fish farming (HAP) clustered together indicating similarities between these farming systems. The Mixed Commercial farming in Lushui (MFLS) loaded more on PC1 than on PC2 and shared lesser similarities with the other systems. The farming system in Dulongjiang (MFDJ), although loaded more on PC2, showed similarity to MFDJ-both being market oriented cultivation systems complemented through 'Grain for Green' programme [28]. The Mixed Farming of Barshong (MFBA) loaded heavily on PC2 and seemed to differ from the rest of the farming systems. Considering the cumulative scores of all seven sustainability components, it was noted that the Apatani heritage paddy-cum-fish farming of Ziro Valley (Figure 3a) and integrated mixed farming system of Bhutan (Figure 3f) had greater potential for sustainability with bigger and balanced radial spaces. This implied to balanced attention to all seven sustainability space components. Comparatively, the Lopil-based shifting hill agriculture of Chin Hills (Figure 3c) had uneven and smaller radial spaces, indicating lesser attention to Economic Prospects, External Support, Integrated Approach, and Social Wellbeing components. Looking at the PCA bi-plot (Figure 4), the clustering of the Traditional Shifting Cultivation of Nagaland (SCNL) and Transforming System of Dima Hasao (TSCDH) indicated that these two farming systems were similar in sustainability space scores.
Likewise, the Lopil-based shifting hill agriculture (SCCH), Sikkim Organic Farming (CBOSK), and Apatani Community heritage paddy-cum-fish farming (HAP) clustered together indicating similarities between these farming systems. The Mixed Commercial farming in Lushui (MFLS) loaded more on PC1 than on PC2 and shared lesser similarities with the other systems. The farming system in Dulongjiang (MFDJ), although loaded more on PC2, showed similarity to MFDJ-both being market oriented cultivation systems complemented through 'Grain for Green' programme [28]. The Mixed Farming of Barshong (MFBA) loaded heavily on PC2 and seemed to differ from the rest of the farming systems. The loadings of SCNL and TSCDH in close proximity could be justified if we consider External Support (ES) component of sustainability space where communities expressed concerns on government support to conservation of traditional landraces and germ plasm, including support to organic farming. They also indicated lesser support in promoting low cost soil water management and weather technologies and climate information. Likewise, farming systems such as CBOSK, HAP, SCCH, SCNL, and TSCDH grouped together in terms of their stronger positive correlation with sustainability space components relating to Social Well-being, Space Organization, Resource Efficiency, and Adaptive Features. These farms have eco-agricultural orientations that generates both The loadings of SCNL and TSCDH in close proximity could be justified if we consider External Support (ES) component of sustainability space where communities expressed concerns on government support to conservation of traditional landraces and germ plasm, including support to organic farming. They also indicated lesser support in promoting low cost soil water management and weather technologies and climate information. Likewise, farming systems such as CBOSK, HAP, SCCH, SCNL, and TSCDH grouped together in terms of their stronger positive correlation with sustainability space components relating to Social Well-being, Space Organization, Resource Efficiency, and Adaptive Features. These farms have eco-agricultural orientations that generates both conservation and production co-benefits and enhances production-dependent livelihoods [29]. These farming systems were perceived to be also efficient in terms of RE and AF with better situation for crop/livestock germplasm conservation including equitable access and resource sharing mechanisms, and indigenous land management practices and techniques. These farming systems also share similarities in their judicious utilization of limited land spaces into various types of production spaces [30,31]. The two cases from China-MFDJ and MFLS-had more of a commercial orientation with higher inclination towards the use of hybrid species and chemical fertilizers, and market-oriented crops-therefore lesser scores for AF and RE.

General Characteristic of Mountain Farming Systems in Terms of Seven Sustainability Space Components
We extrapolated general characteristic of mountain agricultural systems (Figure 5a) from the average scores of sustainability spaces of the eight types of mountain farming systems, supplemented by a PCA bi-plot (Figure 5b) using seven sustainability space components as variables. In this case, as per the scree plot, the variability was best explained with three principal components, thus PC1 and PC2 explained only 50.2% of the variability in the data. This exercise was mainly to highlight current strength and limitations of farming systems in the mountains, so that policy and management decisions can become more mountain specific. conservation and production co-benefits and enhances production-dependent livelihoods [29]. These farming systems were perceived to be also efficient in terms of RE and AF with better situation for crop/livestock germplasm conservation including equitable access and resource sharing mechanisms, and indigenous land management practices and techniques. These farming systems also share similarities in their judicious utilization of limited land spaces into various types of production spaces [30,31]. The two cases from China-MFDJ and MFLS-had more of a commercial orientation with higher inclination towards the use of hybrid species and chemical fertilizers, and market-oriented crops-therefore lesser scores for AF and RE.

General Characteristic of Mountain Farming Systems in Terms of Seven Sustainability Space Components
We extrapolated general characteristic of mountain agricultural systems (Figure 5a) from the average scores of sustainability spaces of the eight types of mountain farming systems, supplemented by a PCA bi-plot (Figure 5b) using seven sustainability space components as variables. In this case, as per the scree plot, the variability was best explained with three principal components, thus PC1 and PC2 explained only 50.2% of the variability in the data. This exercise was mainly to highlight current strength and limitations of farming systems in the mountains, so that policy and management decisions can become more mountain specific. From Figure 5a, it is evident that: (i) Mountain farming systems are resource-efficient (RE: 68%) with the local communities managing their resources efficiently, making the most of the locally available resources, and recycling elements between the farm and other land uses [32]; (ii) Mountain farming communities make the best use of the land spaces available to them (SO: 62%) that entails making the land ownership and tenure situation as efficient as possible, demarcating areas for various purposes such as community forests, conserved areas, plantation, private forests, upland terrace, terraced fields, sloping lands, wet paddy fields, home gardens, orchards, settlements, and village areas [33]; (iii) Farming communities are custodians of site-specific cultural values and social norms, and their motivations are indispensable for sustainably shaping mountain agriculture. Often regarded as family farming, mountain agriculture co-evolves with the aspirations and well-being of the farming families and community they are part of (SW: 58%); (iv) Mountain farming systems maintain a subsistence orientation and rely on the use of microclimate-driven habitats, resulting in high agro-biodiversity, which provide the farming system with greater adaptive capacity and resilience (AF: 57%). Diversity, alongside the richness of knowledge and practices and the engagement of communities, keeps them vibrant and dynamic; (v) Mountain communities integrate an extensive set of actions (IA: 52%) such as managing forests, managing water for farm and home, building fodder resources for livestock, dealing with pests and diseases, maintaining soil nutrients, exploring cash crops and markets, promoting food crops and commodities during traditional ceremonies and festivals, building institutions and networks, and setting up collective norms and benefit-sharing mechanisms; (vi) More than money, farmers often seek an effective rural Research and Development program that is relevant to them, fulfills their basic development needs, and addresses local-level challenges (ES: 49%). They need to add value to what is already being well-managed by the farming communities; and (vii) While mountain farming has a subsistence orientation, it is a major source of livelihoods for rural farming communities. The prospects for agribusiness and income opportunities from mountain agriculture are minimally explored, including systematic support mechanisms at the organizational levels, such as cooperatives and small-scale industry developments (EP: 37%).
From the PCA bi-plot (Figure 5b) correlations among the components AF and RE (PC1 loading) with ES and EP stands true with the characteristic of mountain agriculture where lesser the economic and support outreach, the higher the tendency to rely on diversity and channelization of local resources [34]. In heritage and traditional faming systems such as HAP and SCNL, communities have long maintained their farmlands using an integrated landscape management approach by making optimal use of available local resources, also maintaining a very high diversity of agrobiodiversity making use of also wild edibles from the forests and crop relatives in the wild [31]. Additionally, They have elements of sustained soil quality maintained using rotations of crops, compost and organic manures, conservation of natural resources for long term services provisions from farmland, resilient and diverse production systems, social equity and healthy environment to sustain people, families, and communities [35]. It is evident that farming systems with greater Social Well-being (SW) and Space Organization (SO)scores-loading heavily on PC1, have higher sustainability potential, as it means greater engagement of farming communities, in terms of their motivation and affinity toward managing the systems [36]. Their engagement is crucial in strengthening the effective use of land spaces and the maintenance of agro-biodiversity resources and wider ecosystem services [37].

Positive Impact Factors Enhacing Farm Sustainability
The cumulative performance of PIFs from across eight farming cases (Figure 6a

Positive Impact Factors Enhacing Farm Sustainability
The cumulative performance of PIFs from across eight farming cases (Figure 6a) indicated that community perceptions towards current condition of most of the PIFs were good except for 3 PIFs with below average score-Use of Interdisciplinary knowledge (UIK:48%), Research and Programmatic Support (RPS: 46%), and Agricultural Entrepreneurship infrastructure (ENI: 45%). The overall scores for all three PIFs related to External Support (ES) from the government-Agriculture extension services (AES: 53%), Policy and technological infrastructure (PTS: 52%), and Research and Development programme and schemes (RDS: 46%) were the lowest compared to other PIFs.  Principal Component Analysis of the PIFs showed that PC1 and PC2 explained only 57.6% of the variability in the data. The scree plot indicated that the variability was best explained with five principal components. The factors ENI, FNI and AES loaded positively on PC2, while RDI and ABD on PC3, CIF on PC4, and ETI and ING on PC5. However, for ease in explaining the variables rotated in space, only PC1 and PC2 were considered and presented in a bi-plot (Figure 6b). The variables that shared common perceptions are circled and discussed.
Management of Local Resource (MLR: 83%) and Habitat Connectivity and Network (HCN: 82%) were among the most proficient and well-maintained PIF. Stronger correlation between HCN and MLR both loading strongly on PC1 related to better condition and extent of adjoining forests and wetlands, and ecological connectivity between farm land and non-farm spaces; better availability of food, fodder, water through-out the year; and effective channelization of inputs into farming systems from adjoining forests and wetlands. Although with differential PC loading, MLR corresponded strongly with Agrobiodiveristy (ABD: 76%) in terms of the use of various production habitats for growing diversity of crop resources. However, the extent of Agrobiodiversity (ABD) varied across different farming systems (Figure 6a).
Land tenure and ownerships (LTO: 75%) was perceived to be weaker in terms of the extent of land used for growing traditional crop species, especially farming systems that are commercially oriented. These perceptions justified the correlation between LTO and Social Equity and Cohesion (SEC: 78%)-both loading negatively on PC2. Social Equity and Cohesion stood for extent of celebration of festivals, extent of community institutions and network, interest and motivation among community towards agriculture and importance of local cuisine and culture-therefore stronger connect to the traditional farming land spaces. In the case of sites from China, the traditional upland farms were under 'Grain for Green' scheme [28], therefore no longer into traditional hill farming; traditional crops were mostly replaced by high-yielding hybrid species. The farming spaces in the newer settlement areas were also mostly occupied by collectively marketed commercial crops. Such transformation also influenced the extent of ABD.
Engagement of traditional institutions (ETI: 81%) reflected on availability of in-kind support within community, organization of labor force and extent of transfer of knowledge from elder to younger generations, across all farming systems. Majority of the community within their respective farming system still relied upon support from each other for farming activities, with elder generation still leading and guiding the farm activities. These feature strengthened both PIFs related to LTO and SEC. Community experts regarded that wherever there were inclinations or positive scores for economic orientation in farm management, there were lesser regard to opinions of or engagement of traditional institutions-this justified the negative correlation between PIFs related to Economic Prospects (EP)-Enterprise infrastructure (ENI), Financial infrastructure and process (FNI) with ETI ( Figure 6b).
Perceptions on Community interest in developing farm resources (CIF: 73%) varied across the farming systems. In the traditional farming systems community agreed on having high interest in developing local crop varieties, while in the commercially driven farming systems, scores for this indicator were low-thus showing greater correlation with LTO, ETI, and SEC-all loaded negatively on PC1.
For all farming systems, the PIFs under Economic Prospects (EP)-Enterprise infrastructure (ENI: 45%), Market infrastructure and connect (MKI: 50%), and Financial infrastructure and process (FNI: 53%) had lowest scores. Communities in general perceived that the farming systems in the mountains are not adequately equipped to enhance the economic objectives. Market connect and infrastructure was considered better in Sikkim, India, but very unsatisfactory in Chin Hills in Myanmar, where community perceived lesser promotion of local produce and it's connect to wider market beyond the villages. These three factors MKI, FNI and ENI showed closer correlation with Use of Interdisciplinary Knowledge (UIK) indicating the need to apply diversified and integrated knowledge base related to value chain mechanisms, certification and branding, microenterprise development and linkages to wider markets and private sector partnerships for agribusiness.
The three factors -Ecosystem Services Management (ESM), UIK, and Inter-sectoral Coordination (ISC) within Integrated approach (IA) correlated positively with Policy and technological support (PTS)-all loading positively on PC1. This emphasizes the need to acknowledge traditional agro-ecosystem management practices and integrate wider sectorial knowledge, while developing policy and technological supports for mountain farms. The strong correlation between ESM and UIK is well justified in terms of communities indicating that effective use of local resources calls for use of both traditional and modern knowledge base. Likewise, close collaborations of different line department of governments and research institutions with the farming communities signifies stronger correlations between ISC and Research and Development Programmes and Schemes (RPS) and between ISC and PTS. Community experts provided higher score on ISC where government research and development programs were more integrated and prompted community engagement.
Considering PIFs under Social Well-Being (SW)-Access to development facilities (ADF: 72%), and Inclusive growth (ING)-loaded strongly on PC1 reflected importance of access to basic health, local market, and market information facilities. Strengthening these factors would play an important role in mitigating farm challenges related to out-migration and farm abandonment [38]. Likewise, PIFs under Adaptive Features (AF)-Agrobiodiversity, Community Skills and Practices, and Risk Mitigation Mechanism (RMM)-all loading positively on PC2 highlight PIFs that help mountain farming community's deal with issues on crop-livestock loss, market failure, and crop depredation.

Negative Impact Factors Enhacing Farm Sustainability
The cumulative performance of NIFs from across eight farming cases (Figure 7a Labor shortage was a moderate concern only in Barshong where farming communities were migrating to cities (MFF) for other non-farm jobs, especially the younger generation who preferred other technical professions. This perception also justified strong correlations between MFF and the increasing interest towards Non-Farm Livelihoods (NFL). Likewise, MFF also related well to the fluctuation of market price (MPF) for high value farm commodities-especially lesser return of investment and unavailability of fair price. Additionally, there was a noticeable trend of less interest among younger people in taking up agriculture as a profession, especially traditional farming.
PCA of the NIFs showed that PC1 and PC2 explained only 56% of the variability in the data. Similar to the PIFs, the scree plot indicated that the variability was best explained with five principal components. However, for ease in explaining the variables rotated in space, only PC1 and PC2 were considered and presented in a bi-plot (Figure 7b). The variables that shared common perceptions are circled and discussed. A strong correlation was evident between NIFs related to Adaptive Feature (AF)-HYG and LLC indicating a common perception across all farming systems that with the ingress of hybrid and improved varieties, local crop cultivars are less likely to be grown by the farming communities. Additionally, the challenge of injudicious use of pesticides and chemical fertilizers (PCF) is also closely connected. The negative correlation between PCF and pest and diseases infestation (PAD) reiterates the challenge of extensive use of chemicals to manage cultivation of market-oriented crop species, especially true to the transforming traditional agricultural systems. PCA of the NIFs showed that PC1 and PC2 explained only 56% of the variability in the data. Similar to the PIFs, the scree plot indicated that the variability was best explained with five principal components. However, for ease in explaining the variables rotated in space, only PC1 and PC2 were considered and presented in a bi-plot (Figure 7b). The variables that shared common perceptions are circled and discussed. A strong correlation was evident between NIFs related to Adaptive Feature (AF)-HYG and LLC indicating a common perception across all farming systems that with the ingress of hybrid and improved varieties, local crop cultivars are less likely to be grown by the farming communities. Additionally, the challenge of injudicious use of pesticides and chemical fertilizers (PCF) is also closely connected. The negative correlation between PCF and pest and diseases infestation (PAD) reiterates the challenge of extensive use of chemicals to manage cultivation Across all farming systems, inadequate capacities in relation to agribusiness were the most crucial (CAB: 71), this factor had strongest positive correlation with CSW or capacities related to sustainable soil, water management.
Top soil erosion (TSE), was considered a moderate challenge; however, the perceptions showed high variation between the communities. For example, this factor was less of a concern in Dima Hasao (low altitude farming), Sikkim (perennial crop-based farming), and Apatani (valley-based farming). However, top soil erosion was considered a huge challenge by communities in Nagaland, Chin Hills, and Lushui, where farming is done in comparatively steeper lands that easily erode, especially during monsoons. In Chin inadequate and improper road infrastructure cut off the villages during the monsoon and landslides often cause erosion.
The water stress (WST) was considered as a moderate challenge, but the variations were high between the farming systems. For example, Apatani, Lushui, and Dulongjiang farmers did not consider water stress as a challenge because water was available for domestic and farm use at all times. However, in Bhutan and Chin Hills, it was considered a challenge, because of limited drainage facilities, water channeling systems, and low-cost water saving technologies.
Thus, PAD, WST, ULS, NFL with strong loading on PC2 were factors reinforcing deviation of farm practices. The NIFs indicating conflict with other land use (CLF), Lack of motivation to farming community (LMS), capacity for agribusiness (CAB) and Crop Depredations (CDP) that loaded negatively on PC2 highlight the importance of government support in terms of meeting the needs of the farming community with regard to capacity building, acknowledgements of their contribution to maintaining agricultural systems in harmony with other natural land uses.

Conclusions
The Sustainability Space: A composite index to measure sustainability as elaborated in this paper enabled the conversion of farm specific sustainability performance into seven relevant themes for decision-making. It reaffirmed that sustainable agriculture development requires a more integrated, objective oriented approach [39] and that sustainability assessment results must be socially relevant for decision making-that is helping farming community's reflect upon their knowledge and practices within the wider context of impact factors, and helping decision makers make better decisions. The decision imperatives making the mountain agricultural system more resilient in future lies on holistic strengthening of geophysical pre-requisites; ecological foundations; integrated processes and practices; resources, knowledge, and value systems; stakeholders' development and economic aspirations; well-being of farming communities; and government support mechanisms. Visualization of on-the-ground sustainability with the help of seven sustainability space components defined through 21 PIFs and 16 NIFs prompts decision makers to give attention to: (i) Minimizing external inputs to the systems and channelization of internal resources -human, social, ecological, and capacity; (ii) Planning rural development infrastructures in a way that diversified mountain production land spaces that the communities use for the cultivation of diverse crops at different times are not compromised; (iii) Considering well-being of farming communities in terms of their inclusive growth, access to basic development facilities, and societal harmony that gives continuity to communities interaction with their farmlands, evolving them further; (iv) Strengthening communities' skills and capacities through participatory and collaborative research approaches and co-learning mechanisms; (v) Diversifying economic benefits from mountain farm products and strengthening agribusiness infrastructures and investments from a wider range of stakeholders; (vi) Considering incentive mechanisms for farmers who grow local landraces and maintain in-situ farm genetic resources. It is necessary that the price of local produce incorporate the cost of maintaining the wider farm ecosystem services-the ecological, cultural and aesthetic services-adding to the benefits for the farming communities who maintain the system; (vii) Strengthening organic orientation by appropriate government schemes and policies for soil nutrient management, water management, low cost technologies, as well as for promoting certified and branded mountain farm products; and (viii) Supporting agrobiodiversity conservation-oriented policies considering in-situ conservation of mountain agro-resources and their relatives in the wild, including wild edibles, as well as ex-situ conservation infrastructure for conservation of agro-germ plasm. The future sustainability of mountain agriculture and food security will require access to a wider genetic resource base and the engagement of farming communities in the continual development of crop/ livestock germplasm.
Our primary intention behind this research was to enable solution-oriented mountain specific decision-making that brings balance among the seven interconnected sustainability space components-reinforcing holistic sustainability of mountain agricultural systems.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Appendix A. Categorization of PIFs and NIFs into Seven Sustainability Space Components
The disciplinary experts were asked to give score between 0-2. (0 = no influence; 1 = Indirect and weak influence; and 2 = Direct and strong influence) depending on how they thought each PIF/NIF influenced different sustainability space components. This was to explore which SSC was most strongly and directly affected by which PIFs or NIFs. Their opinion scores were plotted in excel to determine activeness of each impact factors. The sum of their responses for each sustainability space was converted into percentile value referred here as 'activeness' score for each PIF and NIF. Placement of PIFs/NIFs under different Sustainability Space Components were sorted by their activeness score (as highlighted) in the tables below: For example, the PIFs-Habitat Connectivity (HCN), Land tenure and Ownerships (LTO) and Rural Development Infrastructure (RDI) were categorized under Space Organization (SO) component.