Fostering sustainable agriculture in Burundi: which competencies for change-agents should vocational agriculture education prioritize?

ABSTRACT Context and purpose In a context of land shortage with high rates of land degradation and soil erosion due to unsustainable agriculture practices in Burundi, vocational agriculture education needs to prioritize competencies that enable graduates to be effective change-agents in addressing sustainable agriculture challenges. This study aimed at identifying context-relevant competencies for change-agents to be able to foster sustainable agriculture. Methodology The paper used the Delphi technique to collect data from 28 experts (comprising of 14 teachers and 14 practitioners). Findings The findings identified 11 context-relevant competencies that are considered by experts to be most important for change-agents working on sustainable agriculture. The competence ‘facilitator of change’ was found to be most relevant and the recommendation is that it should be prioritized in vocational agriculture education curriculum design. Other competencies, such as innovation and creativity, planning, system thinking, domain expertise, continuous learning, interdisciplinarity, and leadership, were also indicated as highly relevant. The competencies stewardship, self-determination and engagement were considered to be comparatively less important, although still relevant. Practical implications The findings of this study provide insight so that the curricula of vocational agriculture schools can be designed around forming a competence-based education model toward fostering sustainable agriculture. Theoretical implications The paper contributes to sustainability competence theory by taking the debate to the domain of vocational agriculture education within the context of Burundi. Originality The study utilized the Delphi process to capture the context-relevance of sustainable agriculture competencies that need to be prioritized in vocational agriculture education in Burundi.


Introduction
Agricultural sustainability is one of the most prominent issues for developing countries, where economies are widely dependent on agriculture as a source of livelihood.Under this consideration, the concept of sustainable agriculture gained more attention since 1980s with the Brundtland Report having constituted a breakthrough in the use of the concept (Edwards 2020;Velten et al. 2015).
Depending on each expert's background, various definitions of sustainable agriculture exist, but in general, definitions are centered around the interaction of ecological, economic and social sustainability elements (Cicciù, Schramm, and Schramm 2022;Heidelberger et al. 2017;Sarkar et al. 2021).In a global sense, sustainable agriculture would mean an agricultural system that considers all aspects of agriculture (Edwards 2020).This conceptualization renders a broad and ambitious definition, which leads to methodological and practical complications in the use of the concept and the setting of measurement indicators (Peng et al. 2022;Velten et al. 2015;White 2013).
Extensive literature elaborating on sustainable agriculture indicators has been recently conducted (e.g.Escribano, Díaz-Caro, and Mesias 2018;Kelly et al. 2018;Lu, Chang, and Wu 2020;Lynch et al. 2019;Sarkar et al. 2021;Srinivasa Rao et al. 2019).Most importantly, these studies have proposed indicators that comprehensively encompass environmental, economic and social aspects.The multidimensionality and complexity in assessing agricultural sustainability is evidenced at an extent that further areas for refinement of indicators may exist (Sarkar et al. 2021).
In practice, sustainable agriculture is the desired ultimate state any given community would target to achieve.Elaborating on this perspective, some authors have suggested that the definition should be related to what is required to be sustainable (Costanza and Patten 1995;Vos 2007;White 2013), hence joining the point of view of Smith and McDonald (1998, 27) where they proposed to invert the question to 'What is unsustainable agriculture?'.This enables faster realization of the observable evidence of unsustainable agriculture (Voinov and Farley 2007) and definitely orients on what should be sustainable agriculture practices and approaches.
Under this line of reasoning, sustainable agriculture can be apprehended through a variety of innovative agricultural farming practices and approaches that have been recently introduced as efforts to address agricultural challenges (Liao, Nguyen, and Sasaki 2022;Muhie 2022;Zeweld et al. 2020).Some concepts related to these approaches and practices are precision agriculture, climate smart agriculture, sustainable intensification, regenerative agriculture, organic farming, biodynamic agriculture, integrated farming, integrated nutrient management, and integrated pest management, among others (Muhie 2022).This study considers sustainable agriculture by reflecting on modern challenges and what competencies should effective change-agents be equipped with to address challenges related to sustainable agriculture in Burundi.
Although the agricultural sector is the backbone of Burundian economy, the sector fails to provide enough food due to low agricultural productivity, combined with agricultural market imperfection (Jeníček and Grofová 2016;Ndayiragije et al. 2017;Niragira et al. 2015).Due to high population pressure, plots are continuously fragmented from father to son with actual land holding per household averaging less than 0.5 hectares.In this context, farmers are forced to overexploit the land, which leads to soil loss and reduced soil fertility (Kessler et al. 2021).This is also the same for household energy resources where there is unmet demand for wood to use for cooking, resulting in high deforestation rates, decreased soil cover, and resultant soil erosion.
In the context of unsustainable agriculture, classical agricultural solutions are becoming obsolete in responding to current and future challenges (Moschitz et al. 2015).Yet, it is worldwide recognized that education can play a central role in sustainability efforts (Demssie et al. 2020;Evans 2019;Grosseck, Țîru, and Bran 2019;Rieckmann 2012;Sherman and Burns 2015), and should aim at preparing change-agents for sustainable agriculture who have context-relevant competencies.
The concept of competence has evolved through various acceptations, from being understood as job-related depictable personal attributes, to generic invariant attributes, and currently, as comprehensive and context-dependent capabilities to handle complex task-related issues (Biemans et al. 2004;Demssie et al. 2019;Sturing et al. 2011;Wesselink et al. 2015).Under this comprehensive view, recent developments elaborating on the framework by Wiek, Withycombe, and Redman (2011) conceptualize sustainability competence as an integration of knowledge, skills and attitudes needed to be sustainably performant and to be prepared for a person's future endeavors in this regard (Demssie et al. 2020).This should be the intended educational outcome for a sustainability-oriented education system.
As it was pointed out by Knickel et al. (2009), and later by Lieblein et al. (2012), the failure of the agriculture education systems to shape required sustainable agriculture competencies is because agriculture has departed from being viewed as a productionoriented industry, and should instead be regarded as a multifunctional system that encompasses environmental, economic and social aspects (Cahill 2001;Knickel et al. 2009;Seuneke and Bock 2015).This emphasizes the need for sustainable agriculture competencies to be comprehensive.
In Burundi, a network of 29 vocational agricultural schools (in French: Instituts Techniques Agricole, known as ITAB) was established to train extensionists with the intention that they eventually work closely with the rural population at the grassroots level.A national competence reference (in French: Référentiel Métier Compétence -RMC) for secondary vocational agriculture education was formally adopted in 2015.It can, therefore, be recognized that the education system at ITAB is intended to be competencebased, in line with education reforms that have been undertaken since 2000s in most Sub-Saharan African countries (Tagne and Gauthier 2016).However Tagne and Gauthier (2016) observed that such reforms were rashly introduced in Burundi without reflecting on the sustainability challenges that the educational program was intended to target.
For an educational program to fulfill its mission, the starting point would be to elaborate on a list of relevant competencies that need to be fostered (Bergsmann et al. 2015;Wesselink et al. 2007).Previous studies on sustainability competencies attempted to derive a generic sustainability competence framework (Heiskanen, Thidell, and Rodhe 2016;Mulder 2010Mulder , 2014Mulder , 2017;;Perez Salgado, Abbott, and Wilson 2018;Wals 2014;Wiek, Withycombe, and Redman 2011) and in some cases, for specific contexts such as the studies by Demssie et al. (2019) and by Tarekegne et al. (2017Tarekegne et al. ( , 2021) ) in Ethiopia.
General competence frameworks criticize competencies development in earlier decades, arguing that performance development considered 'known tasks and existing jobs' instead of adaptability to all human complex situations by looking into the future in which jobs are not even known yet (Mulder 2017).Moreover, some scholars support the idea that contextuality should be an important dimension in competence development (de Haan 2010; Demssie et al. 2019;Tarekegne et al. 2017Tarekegne et al. , 2021)).This consideration is also important for a vocational agriculture education system that aspires to deliver change-agents capable of addressing underlying sustainable agriculture challenges that may vary across regions and countries.
The studies by Demssie et al. (2019) and by Tarekegne et al. (2021) have been conducted in a somewhat comparable context as that of Burundi, but it is not clear if what was identified as sustainability competencies for the Ethiopian context is applicable for vocational agriculture education in Burundi.The overall objective of this study was to identify context-relevant competencies for change-agents to be able to foster sustainable agriculture.Two research questions guided the study: 1. What are the most relevant sustainable agriculture competencies to be prioritized at vocational agriculture schools in Burundi? 2. Are there differences or similarities in the way that practitioners and teachers in vocational agriculture schools perceive the relevance of sustainable agriculture competencies?

Methodology
This study applied the Delphi technique, a forecasting process that is used when seeking experts' consensus on a given matter for which there are no conclusive answers (Banno, Tsujimoto, and Kataoka 2020;Barrios et al. 2021;Demssie et al. 2019;Habibi, Sarafrazi, and Izadyar 2014;Hasson, Keeney, and McKenna 2000;von der Gracht 2012).The technique comprises of launching a questionnaire in successive rounds.After each round, the researcher evaluates the responses and shares the summary feedback with participants with an expectation that they adjust their views in the following round (Ballantyne, Hughes, and Bond 2016;Habibi, Sarafrazi, and Izadyar 2014;Ogbeifun et al. 2016).This increases the chance for consensual answers in the next round.The process continues until consensus is reached.The appropriateness of the Delphi technique resides in the fact that experts exchange views in an anonymous way, hence reaching a convergence towards consensual and consistent answers (Mukherjee et al. 2015;Osborne et al. 2003;Thangaratinam and Redman 2005).The technique presents an advantage of avoiding the influence of dominant participants by rigorously ensuring anonymous responses (Dalkey and Helmer 1963;Gupta and Clarke 1996;Habibi, Sarafrazi, and Izadyar 2014).However, criticism exists around the fact that the Delphi technique lacks a clear guideline regarding experts panel size, consensus determination and number of rounds (Demssie et al. 2019;Habibi, Sarafrazi, and Izadyar 2014;Hasson, Keeney, and McKenna 2000).Our prior awareness of this critique of the technique helped us to carefully progress, step-by-step, and ensure that each methodological action was sufficiently justified.The step-by-step Delphi process used in this study is illustrated in Figure 1.

Preliminary activities: identification of participants, recruitment and introductory session
It is generally known that solving sustainability challenges requires combined efforts from various experts (Caron, Durand, and Asselin 2016;Day and Bobeva 2005;Demssie et al. 2019;Rieckmann 2012).In this study, two categories of experts were considered to be knowledgeable of the requirements for ITAB laureates to become effective change-agents in their future professional orientation.On one hand, ITAB teachers are assumed to theoretically know what competencies should be prioritized for sustainability problem solving.On the other hand, practitionersindividuals from government agencies and NGOs who are working on addressing sustainability challengesare in a better position to tell what qualities and skills a competent change-agent requires in practice.
Inclusion/exclusion criteria for the selection of experts were related to knowledge and experience in the domain.We, therefore, recruited only participants with a tertiary level of education and at least five years of experience in the area of sustainable agriculture.In total, 32 experts were recruited to take part in the study (16 teachers and 16 practitioners), among which 28 experts (87.5% response rate) returned filled questionnaires for Round 1. Table 1 summarizes the characteristics of the experts who participated in this study.
Before administering the questionnaires, an introductory session with panelists was held to increase the response rate.The introductory session was also used to explain the concept of sustainability competence as 'integrated knowledge, skills and attitudes' (Demssie et al. 2019;Wiek, Withycombe, and Redman 2011) that permit agricultural school laureates to serve as effective change-agents in their future career.The introductory session was conducted from mid-December 2020 to the end of January 2021 via physical meetings, field visits, and, in some cases, telephone discussions and emails.

Round 1 data collection process
Most crucial to the Delphi technique used in this study was the 'blank paper' questionnaire that was introduced in Round 1. Instead of the usual rating of competencies from the literature (Demssie et al. 2019), panelists were asked to write down what they  perceived as relevant sustainability competencies, reflecting on their knowledge and experience in the domain.The advantage of adopting such an approach was to avoid the tendency to copy-paste theoretical competencies from literature, thereby enabling the capture of the relevance and the context-specificity of suggested competencies, as per the aim of this study.
The questionnaire was a blank sheet with three columns, headed: (i) Competence abstraction/name, (ii) Meaning of the suggested competence, and (iii) Reasons why the suggested competence should be regarded as context-relevant.Prior to the full launch of the questionnaire, it was pilot-tested on three panelists, after which minor modifications were made to the final questionnaire.Questionnaires were distributed to panelists from 10 to 20 February 2021 and a reminder (by phone call) was regularly made to reduce the risk of non-responses.The last filled questionnaire was received on 23 April 2021.

Round 1 data analysis
After recording and translation of answers provided by panelists on the blank paper questionnaire, Round 1 responses were analyzed from May to July 2021.Mainly, the task was to summarize and categorize experts' wordings/comments with a purpose to come up with a list of competencies.Content analysis was used to summarize and categorize competence elements/snippets into well-defined competencies.To remove the overlap and redundancy, regular discussions with co-authors as well as literature search on sustainability competencies were conducted.Happily the overlap of words confirmed the importance of suggested competencies (Demssie et al. 2019), hence competence wordings could be put together.The final summary feedback from Round 1 comprised a list of 11 sustainability competencies which constituted questionnaire elements for Round 2.

Round 2 data collection process
The Round 2 questionnaire featured 11 competencies identified from Round 1.In Round 2, participants were asked to rate the relevance of each competence with a brief explanation of the suggested rating.A 7-point summated scales was used for rating.Scales were defined as 1 = Very irrelevant, 2 = Irrelevant, 3 = Slightly irrelevant, 4 = Neutral, 5 = Slightly relevant, 6 = Relevant, and 7 = Very relevant.Before sending the questionnaire to the experts, it was shared with three participants for pilot-testing.No major corrections were suggested.Questionnaires for Round 2 were distributed to panelists in the last week of July 2021 and the last filled questionnaire was received by 31 August 2021.
Among the 28 respondents who received the questionnaire for Round 2, 26 respondents returned completed questionnaires, i.e. 13 respondents from each of the two categories of experts.

Round 2 data analysis
Mean rating scores and percentages of experts suggesting each score level for each competence were calculated to assess the relevance of each competence according to the panelists.To determine to what extent consensus was reached after Round 2, two consensus assessment approaches were used.First, we applied the criterion of the average percent of majority opinions -APMO (Barrios et al. 2021;von der Gracht 2012), that is, the percentage of experts who suggested a level of relevance for any of the 11 competencies.The criterion for consensus achievement was based on the argument of Barrios et al. (2021) that consensus will always increase as long as feedback indicates at least 75% group agreement; here, at least 75% of the experts suggested this competence relevance level.Second, Pearson Chi-square test statistics were computed for each competence to assess the extent to which teachers and practitioners may have scored competence relevance differently.Conclusions from this consensus assessment process provided sufficient evidence regarding the relevance of competencies; hence, the Delphi process stopped by Round 2.

Round 1 results
Results for Round 1 are presented in Table 2, in which the 11 sustainability competencies are presented.The discussion on the meaning of each competence is developed in Section 4.1.In Table 2, competencies are presented in their order of frequency of suggestion by the experts that participated in the study.For instance, 'facilitator of change competence' was at the top of the list because it was suggested by 96% of all experts participating in the study.Similarly, 'engagement competence' was listed at the bottom because it was suggested by only 11% of the experts.
As presented in Table 2, no single competence was exclusively suggested by only one specific category of experts.Most differences in percentage of suggestion between teachers and practitioners were observed for the competencies 'System thinking', 'Selfdetermination' and 'Continuous learning'.An equal percentage by both teachers and practitioners was observed for the competence 'Stewardship'.

Round 2 results
Round 2 results are presented in Table 3. Round 2 aimed at seeking consensus, therefore, in addition to percentages of experts rating competencies as relevant, Pearson Chi-square tests were used to identify differences in rating scores of each competence between teachers and practitioners, and finally the state of agreement between the two categories of experts.
Overall, it was found that eight out of the 11 competencies were rated as at least 'Relevant' by more than 75% of the experts.This finding led to the conclusion that convergence existed in ratings of the competencies' relevance if we refer to the observation by Barrios et al. (2021) that consensus will always increase as long as feedback indicates at least 75% group agreement.With exception of the 'Leadership competence', Pearson Chi-square test results for differences between both groups of experts were not statistically significant at the 5% level of significance for the other competencies (see Table 3).

Changing insights on the competencies throughout rounds
Changing insights through rounds and the move towards consensus is illustrated in Figure 2 in which arrows are used to illustrate patterns of changes from Round 1 to Round 2. Green, blue and red arrows stand respectively for increase, no-change, drop in ranking order of competencies relevance from Round 1 to Round 2. As illustrated in Figure 2, more than 50% of experts from both categories agreed that all competencies were relevant during Round 2. If we look at the changes in frequency of suggestion from Round 1 to Round 2, teachers increased the relevance they attributed to 'interdisciplinary' competence, while for practitioners, more relevance was given to the competencies 'innovation and creativity', 'continuous learning' and 'engagement'.
A drop in relevance attributed to competencies was observed for 'facilitator of change', 'domain expertise', 'planning', 'stewardship', 'engagement' and 'self-determination' by the teachers, and 'leadership', 'domain expertise' and 'stewardship' by the practitioners.Further, it can be realized that practitioners' view on competencies relevance was relatively stable; i.e. no changes observed for five competencies compared to teachers' view with no changes observed for only two competencies.

Discussion
In line with the specific research questions to be answered, the findings of this study identified a list of 11 sustainability competencies.This section addresses first the meaning that experts attached to these competencies and discusses relationships to literature on sustainability competence.Next, perceptions and scores of teachers and practitioners concerning the relevance of sustainability competencies are discussed, with implications regarding curriculum development for vocational agriculture education in Burundi.

Facilitator of change competence
This competence refers to the ability to put oneself in the position of the farmer and think and act from the farmer's perspective.Frequent wordings that were used by experts relating to this competence were empathy, heart feeling, facilitation, group/indigenous knowledge valorization.In that regard, the change-agent does not act as a teacher/knowledge-giver but as a coach who is ready to foster collaboration, knowledge sharing, and establish synergies and possible partnerships between different actors in a way that helps beneficiaries to achieve their goals.The literature relating to this competence focuses more on the aspect of collaboration.With partnerships and establishment of synergies, this competence captures aspects of stakeholders coordination (Demssie et al. 2019).As justified by the experts participating in this study and consistent with the literature, collaboration starts with empathy (Rieckmann 2012; van Dam-Mieras et al. 2008;de Haan 2006) and it also forms the foundation for facilitation, coaching, and knowledge sharing.It also requires the ability to manage indigenous and group knowledge of the targeted population (Demssie et al. 2020).

Innovation and creativity competence
Related to this competence, wordings formulated by the experts captured curiosity, creativity, innovation, reflection-in-action, and any kind of transformative initiatives for sustainability.Elaborating on these concepts, the competence would mean the ability to develop a sense of creativity and innovation in applying knowledge for sustainable transformation.It involves exercising the imagination, curiosity and reflection-in-action.It also can be concerned with discovering available opportunities in the surrounding environment and valuing them by skillfully organizing all required means and activities towards success.
The competence has comparable meaning with what Evans (2019, 9) called 'creative and strategic competence' referring to the 'ability to collectively envision, develop, implement and assess transformative interventions for sustainability'.It is also comparable to what Rieckmann (2012, 133) described 'open-mindedness and disposition to innovation'.To these, Charatsari and Lioutas (2019) add the ability to assimilate knowledge and learning from entrepreneurial experience.

Planning competence
The competence 'planning' refers to the ability to set objectives as well as organize the means, methods, tools and strategies put in place to achieve the objectives.In the view of this competence, the change-agent must be able to envision what should be undertaken to solve sustainability challenges with clearly defined goals.They have to think of best options and means to achieve sustainability goals.This capability allows the rational use of resources and, hence, optimize performance and the efficiency of the activities undertaken.Ultimately, planning competence enables self-evaluation to ensure that a task has been well-accomplished and elucidate the extent to which objectives have been reached.The meaning of this competence is comparable to what de Haan (2006, 24) called 'competence in planning and implementation skills'.It captures the ability to rationally organize and utilize resources, and the evaluation and self-assessment of performance.

System thinking competence
Main concepts that experts relate to this competence are analysis, anticipation, prediction, foreseeing the impacts of one's actions, and collective understanding of complex sustainability issues toward better propositions for sustainable solutions.This means the ability to analyze the environment and its evolution regularly and globally over time and space to predict/anticipate obstacles and propose appropriate solutions.The change-agent must be aware of the contribution of the changes induced by their interventions, including sustainability and impacts on the environment in the community.
This will help to monitor various factors likely to influence the results of the activities undertaken.
This competence has comparable descriptions as 'analysis or anticipatory thinking' competence in the literature.Some authors separate 'system thinking' from both 'analysis' and 'anticipatory' (Wiek, Withycombe, and Redman 2011) or consider 'anticipatory' or 'foresighted thinking' competence as a separate issue (de Haan 2006;Evans 2019;Gardiner and Rieckmann 2015).

Domain expertise competence
Experts suggested several domain-specific abilities in relation to students' future professional orientations such as with regard to agriculture, livestock management, soil and water management, seed production, land management, and food transformation, among others, referring to the ability to master ones' actions by having knowledge of as many best specific practices as possible in favor of sustainable agriculture.This meaning of competence is comparable to what Demssie et al. (2019, 835) called 'disciplinary competence' or 'action skill competence ' Roorda (2013, 106).

Continuous learning competence
For this competence, the central concept derived from experts' opinions relating to required capabilities for a change-agent was flexibility in the 'way of thinking and doing' to timely adapt to complex situations specifically in this era of globalization.Practically, the meaning of the competence is that graduates from vocational agriculture schools should not be overly attached to what they learned in school, but instead, must update their minds in timely ways with respect to prevailing dynamics.Comparable to the conceptualization by Demssie et al. (2019), 'continuous learning' refers to the ability to continuously adjust ways of thinking and doing with psychic and moral readiness to handle the changing contexts due to climate change, socio-economic, and ecological hazards which make life more difficult.

Interdisciplinary competence
The key concept of the competence is 'integration' of a range of complementary activities.As suggested by experts, interdisciplinarity refers to the ability to undertake diverse activities in a complementary manner and this would ultimately constitute a point of success for resilience.In practice, integration has a wide meaning such that it takes into account integration of others' ideas which impact positively the sustainability of interventions.This meaning is comparable to what (de Haan 2006, 23) called 'interdisciplinary work' competence.As justified by de Haan ( 2006), the competence is required because in today's context, a single activity/strategy does not enable complex problemsolving.

Leadership competence
Concepts suggested by experts related to leadership competence were coordination, conflict resolution, and problem solving abilities.According to the meaning attached by our experts, the competence refers to the ability to determine the path, coordinate and share visions with different stakeholders, and positively influence them with new ideas to achieve results with significant impacts.It is the ability of the team leader to mobilize people around a vision and goals by making it clear who does what and hence avoiding overlap and conflicts by prioritizing complementarity among the actors.
From the literature, leadership competence is captured by the concept of 'normative competence' in which leadership would mean the 'ability to collectively map, specify, apply, reconcile and negotiate sustainability values, principles, goals and targets' (Wiek, Withycombe, and Redman 2011, 209).The meaning of leadership competence is also captured in the 'stakeholders coordination competence'; i.e. the 'ability to organize efforts of various actors including relevant individuals, societies, groups, and experts' (Demssie et al. 2019, 835).

Engagement competence
This competence refers to the ability of an individual to independently control their future professional orientation (for instance engaging in diverse agribusiness activities that can generate income) and develop a mind of auto-promotion instead of being a passive employment demander.This meaning of the competence was derived based on the experts' suggested comments related to engagement, for instance, 'auto-promotion' and 'entrepreneurship'.The competence has an important meaning especially for the case of Burundi, where, sometimes, agricultural school graduates pursue careers that are not related to their specialization.In fact, if vocational agriculture school graduates engage in agricultural activities, it could be similar to preaching by doing (Warren 1998), i.e. providing opportunities to influence people's perceptions toward innovations.

Self-determination competence
As suggested by the experts, the central concept of competence is an individual's 'intrinsic motivation'.According to the experts' opinions, this competence refers to the selfinspiration and voluntary commitment to act with inner force and motivation for sustainability concerns without any external incentive.It is the attitude of working voluntarily with pride for better quality outcomes.The competence captures self-inspiration, inner motivation, and voluntary commitment to work for sustainability concerns.The suggested competence meaning is comparable with de Haan's (2006) description in which motivation is key for social responsibility for changes to commence at the grassroot level.

Stewardship competence
From the experts' viewpoints, this competence involves empathy for the environment and personal commitment to take care of natural resources.As explained by the experts, it is the ability to understand the interconnectedness of environmental challenges and realize the importance of taking care of the environment with a feeling of moral responsibility and personal commitment to work for the conservation of the environment.This competence is similar to what Demssie et al. (2019, 835) called 'resource utilization competence'.The competence meaning is also comparable to what Rieckmann (2012, 133) described as 'understanding the nature' and 'acting fairly and ecologically'.
4.2.Differences and similarities in the way teachers and practitioners perceived the relevance of competencies Differences can be expected in the way categories of experts reflect on the relevance of sustainability competencies.In fact, diversity of expertise is considered as a plus in solving complex sustainability concerns (Demssie et al. 2019).Moving from Round 1 to Round 2, this study revealed both differences and similarities in the way different categories of experts perceived the relevance of competencies and how their perspectives shifted after reflecting on others' views from the feedback of Round 1.

Differences and similarities in perceived competence relevance between practitioners and teachers
In Round 1, main differences were observed for competencies 'system thinking', 'selfdetermination' and 'continuous learning' in which practitioners attached the relevance of 43%, 21% and 21% respectively, but for the teachers these competencies were all suggested with 7% frequency.On the other hand, teachers attached more relevance on competencies such as 'domain expertise', 'innovation and creativity' and 'engagement' on the order of 71%, 50% and 14%, while for practitioners, the same competencies were considered as relevant by 36%, 29% and 7% respectively.
The differences between teachers and practitioners may be explained by their daily experience, in which teachers would prefer assessable competence elements when teaching, while this might not be the case for practitioners working with farmers.This can be observed for instance when we analyze the 'domain expertise' competence which was considered more relevant by teachers where they described it as a list of task/jobrelated knowledge elements.Defining a competence as task or job-related knowledge element can, however, be criticized to be limited to the knowledge level (Baartman and de Bruijn 2011), hence, missing the more comprehensive nature required for sustainability competencies.

Changing insights toward consensus achievement
From Round 2, results revealed no significant differences between teachers and practitioners in scoring the relevance levels of competencies, except for 'leadership competence', for which teachers attached more relevance compared to practitioners.Overall, more than 50% of the experts agreed that all competencies are relevant.This could be due to the shift in experts' views to converge toward consensus after reflecting on their colleagues' opinions from Round 1 feedback.
Significant changes, however, were observed among the teachers compared to the practitioners where they maintained more consistent rankings from Round 1 to Round 2 for five competencies compared to the teachers for only two competencies.Most significant shifts were observed on 'interdisciplinary competence' for teachers and 'leadership competence' for practitioners.These insights confirmed the claim by Perez Salgado, Abbott, and Wilson (2018) that the presumed relevance of sustainability competencies may vary depending on professional background of the experts.

Theoretical and practical implications of study's findings
Previous researches recognized that only scant literature on sustainability competencies exists in the Southern context (Corcoran, Weakland, and Wals 2017;Demssie et al. 2019;Rahdari, Sepasi, and Moradi 2016;Rieckmann 2012).This paper contributes to sustainability competence theory by taking the debate on sustainability competence to the domain of agriculture and vocational agriculture education within the context of Burundi.
As revealed by this study, graduates from vocational agriculture schools in Burundi need primarily to be able to act as facilitators of change for them to effectively contribute to sustainable agriculture.With such competence, the change-agent is able to work together with grassroot farmers to identify problems, opportunities and risks that they face, and determine with them the right course of actions via good collaboration and empowerment.This assumes that the change-agent understands the farmer and adopts the right attitude, ensuring that the farmer can play an active role in his own development process.In that regard, the change-agent should be innovative and creative with a farsighted and system thinking mindset.They should be self-engaged and intrinsically motivated with a sense of responsibility to be a leader and a planner for sustainability pathways.
Findings of this study raised the question about how sustainability competencies are integrated into the teaching of vocational agriculture (Mazzotti, Murphy, and Kent 2007) for the specific context of Burundi.This is even more crucial given that the competencebased education model was recently introduced in Burundi, with main critiques that such reforms were introduced rashly without a context-relevant competence model (Tagne and Gauthier 2016).Findings of this study are insightful regarding what should be the curriculum's intended learning outcomes.
With this study, teachers from ITAB and practitioners with confirmed knowledge of the context have suggested what they perceive as relevant sustainability competencies needed to foster sustainable agriculture change-agents.From that point of view, it can be considered that the suggested competencies are relevant for ITAB to graduate effective change-agents.ITAB curriculum policy makers should decide on how to prioritize identified competencies to serve as a starting point (Wesselink et al. 2007) for developing a competence-based education curriculum in ITAB.In that perspective, agriculture education stakeholders (teachers, curriculum developers and officials from the Ministry of Education) have been associated in this study from the beginning when they participated in the process of competencies identification.Now that findings of this study are available, it is planned to do a dissemination workshop with all these stakeholders with a purpose of contributing to formulating effective policies for sustainable agriculture education.

Limitations and suggestions for further studies
The Delphi technique lacks a clear guideline regarding experts panel size, consensus determination and number of rounds (Demssie et al. 2019;Habibi, Sarafrazi, and Izadyar 2014;Hasson, Keeney, and McKenna 2000).Our prior awareness of this limitation of the method helped us to carefully consider all the contours of each step of our study and ensure that every step was sufficiently justified.Future studies applying the Delphi technique can learn from this case for further refinements to the guide of the research approach.
Another limitation may be related to generalization of findings of this study.With this study conducted in Burundi, the findings may be limited to interpretation for the context of Burundi.However, the findings revealed that some competencies are comparable, either by name or by meaning, to competencies identified from the literature.Further studies should, therefore, focus on the applicability of this study's findings to similar contexts.

Conclusion
This study was conducted based on a context of unsustainable agriculture practices and related consequences, mainly ecosystem depletion, food insecurity and extreme poverty in Burundi's rural areas.The aim of this study was to identify a set of context-relevant competencies for change-agents to be able to foster sustainable agriculture.
The findings of this study revealed a list of 11 sustainable agriculture competencies suggested by experts as context relevant, with the 'facilitator of change' competence being the most relevant.In addition, more than 75% of experts agreed that sustainable agriculture competencies, such as innovation and creativity, planning, system thinking, domain expertise, continuous learning, interdisciplinarity, and leadership were relevant as well.On the bottom of the list of the 11 sustainability competencies and least often suggested by the experts were engagement, self-determination and stewardship competencies.
These findings lead to questioning the extent to which vocational agriculture education system builds up on relevant sustainability competencies to deliver graduates capable of facilitating transformational change for fostering sustainable agriculture in Burundi.As competent extensionists, graduates from vocational agriculture schools should work together with grassroot farmers to identify problems, opportunities, and risks facing them to determine the right course of actions to achieve sustainable changes.A practical implication of these study findings is that vocational agriculture education curricula designers should prioritize teaching content that fosters the aforementioned sustainability competencies and, as such, allowing students to become effective change-agents in their future professional careers.Theoretically, this study contributed to the scientific debate on the relevance of sustainability competencies for the context of vocational agriculture education in Burundi.

Figure 1 .
Figure 1.Summary of the Delphi process.

Figure 2 .
Figure 2. Illustration of changes in insights on the competencies between Rounds 1 and Round 2.

Table 2 .
Context-relevant competencies as suggested in Round 1.