Introduction

Potato late blight disease represents a serious problem for farmers in almost all major potato-growing areas (Tsedaley 2014; Mekonen et al. 2011; Kassa and Beyene 2001). On-station studies estimate yield lossFootnote 1 in Ethiopia to range from 6% to 100% depending on the level of susceptibility of the variety (Guchi 2015; Mesfin and Woldegiorgis 2007). As late blight is air-borne and the causal agent (Phytophthora infestans) travels across farm boundaries, it is not just a problem of an individual farmer but a collective problem of all farmers in a geographical area (Miller et al. 2006). The propensity of late blight pathogen to traverse farm boundaries means that control efforts by any farmer confer a benefit to other nearby farmers. It also means that lack of control on any property stimulates the development of the epidemic and thus imposes costs or losses on other properties (Graham et al. 2019; Marshall et al. 2016). Miranowski and Carlson (1986), in their seminal work, stress the critical role of the mobility of diseases and pests in designing collective management responses. Accordingly, the search for solutions and management practices should take into consideration, on one hand, technical effectiveness and, on the other hand, institutional mechanisms that stimulate and facilitate collective management at the appropriate scale (Graham et al. 2019; Zamani et al. 2018; Marshall 2016). To this end, there is a well-established research effort in designing and testing approaches aimed at enhancing collective capacity in the management of agricultural pests and diseases, generally engaging farming communities in a variety of communal management practices (Hegde et al. 2017; Christmann et al. 2015; Biedenweg and Monroe 2013a).

In the broader field of socio-ecological systems, stimulating social learning is a common recommendation for developing and sustaining the capacity of managing collective action problems(Mostert et al. 2007; Pahl-Wostl and Hare 2004; Rist et al. 2003; Schusler et al. 2003; Leeuwis and Pyburn 2002; Wollenberg et al. 2001). In this body of literature, social learning is extensively examined as a conceptual and methodological approach in facilitating knowledge generation and in forging new types of social relationships among resource users that have to deal with collective action problems. Conceptual perspectives on social learning predominantly take learning as an outcome or end result. Consequently, there is limited insight into salient processes and interactions that trigger a change in individual or collective decisions and actions in the management of socio-ecological problems (Bentley et al. 2018; Beers et al. 2016; Pahl-Wostl 2007). Furthermore, there is a tendency to limit social learning outcomes to consensus or agreements (as ‘identification of adaptation options’, ‘agreement on a common purpose’ or ‘design of governance processes and arrangements’) providing little information on whether such agreements instigated action in actual practice. (e.g. Graham et al. 2019; Hegde et al. 2017; Christmann et al. 2015; Scholtz et al. 2014).

Another significant feature of this type of research is the implicit assumption that some sort of recognition of the ecological or biophysical aspect of a collective problem exists among appropriators or public good providers (Bell et al. 2016; Cárdenas et al. 2015; Brugnach et al. 2011). With this background, the focus of research is often directed towards understanding ‘how’ or ‘in what circumstances’ social learning facilitates knowledge generation and sharing among common pool resource users as well as the social relations necessary for effective collective action on an already recognized collective problem. Such framing poses a particular challenge for the management of contagious crop diseases like late blight. The collective nature of the problem can be unclear to farmers due to the poor visibility of the life cycle and mobility of the pathogen, particularly when compared to other common-pool resources (e.g. forest, fish, irrigation management) or ‘public bads’ (e.g. invasive weeds, migratory pests). As a consequence, previous research in the study area indicated most farmers do not recognize late blight as a collective problem, mainly seeing it as a private bad (Tafesse et al. 2018). Against this background, we consider that social learning approaches for disease management need to pay sufficient attention to learning about key technical aspects of diseases that are characterized by less visible dynamics in the eyes of farmers (e.g. involving oomycete, viruses, bacteria, fungi) and this could play a crucial role in redefining the nature of the problem.

By taking a process-oriented perspective to learning, this study aims to understand how collective action in late blight management develops in a social learning process. It shows how social learning emerges in an organized learning space for potato farmers in Ethiopia and how it effects collective action of these farmers. In this way, it intends to offer fresh theoretical and practical perspectives on contagious diseases such as potato late blight as a collective bad and, correspondingly, the role of social learning in crop disease management.

The paper is organized as follows. Section 2 presents a detailed synthesis of our conceptual framework on collective action and social learning setting the scene for our research questions in the same section. Section 3 provides details of the research methodology where the case study context is also described. Our research findings are presented in section 4 and then discussed in section 5, followed by conclusions in section 6.

Conceptual Framework

Collective Action in Natural Resource Management

Drawing upon Olson (1971) notion of a collective good, Marshall (1998) defines collective action as an action taken by a group, either directly or on its behalf by an organization, in pursuit of members’ perceived shared interests. Whether a particular collective action concerns a public good or a public bad is an empirical, subjective issue because resource users may either suffer from a bad or fail to benefit from a good (Hardin 2015). This perspective entails that while contagious diseases like late blight are generally presented as a public bad that needs to be prevented, collective action provisions towards their management can also be regarded as a public good. In either perception, recognition of the collective nature of the disease is important, as was stated in the introduction.

In public good governance in the realm of agriculture and natural resource management, two types of collective action are distinguished: (1) cooperation: bottom-up, farmer-to-farmer collective action; and (2) coordination that is characterized by a top-down, agency-led collective action approach (Poteete et al. 2009; Davies et al. 2004). Broader debates around the development and success of collective action focus on key performance factors and barriers to be addressed to produce benefits from collective action. Hence, successful initiatives must overcome the most common barriers to collective action (Uetake 2012; Davies et al. 2004). In this regard, a range of substantive factors can govern the performance of collective action initiatives (Poteete and Janssen 2009; Heffernan et al. 2008). Collective action is, first, highly affected by the characteristics of the natural resource (type of good) involved and on the knowledge and the predictability of the resource. The successful integration of knowledge is key to enable communities to govern their public good/bad in a sustainable way (Pahl-Wostl 2009; Pretty 2003). Second, the social dynamics of the group involved, their shared interests and voluntary actions to pursue those shared interests are also significant. In these dynamics, community members should also improve their social relationships to overcome their collective problems (Meinzen-Dick et al. 2004; Pahl-Wostl 2000). As Ostrom (1990) highlighted, collective action is also determined by institutional arrangements, involving both locally devised, simple rules and effective monitoring and sanction systems.

In the light of the above, it can be argued that the achievement of collective action in the management of crop diseases like late blight involves significant changes in two dimensions: a change in technical knowledge about the nature of the disease and a shift in social and institutional arrangements (Deuffic and Candau 2006). As illustrated in the next sub-section, the important challenges or success factors discussed here strongly relate to the central role of social learning in the transformation of knowledge, norms and rules amongst resource users in collective action initiatives (Leeuwis and Pyburn 2002; Roling and Wagemakers 2000).

Social Learning: Technical and Relational Dimensions

Social learning is a broad concept and is operationalized differently in different disciplines and contexts (Koontz 2014). In the domain of natural resource management, social learning is essentially appreciated as occurring when people are brought together to share perspectives and experiences to develop a common framework of understanding as a basis for joint action (Reed et al. 2010; Pahl-Wostl 2009; Schusler et al. 2003). The literature broadly distinguishes between technicalFootnote 2 and relational dimensions of social learning (Phuong 2017; Koontz 2014; Scholz et al. 2014; Woodhill 2010; Pahl-Wostl et al. 2007; Bouwen and Taillieu 2004; Campilan 2002). The technical dimension concerns new knowledge or insights that relate to the natural or biophysical aspect of the problem situation. The relational aspect refers to emergent social properties, such as establishing new relations, breaking existing relations, and building trust and commitment to a group vision as groups become aware of their mutual interdependencies (Beers 2016; Koontz 2014; Reed et al. 2010). In this regard, the technical learning dimension relates to addressing the first determining factor for the success of collective action, namely knowledge about the nature of natural resource at hand. The relational aspect of social learning conforms directly with social competencies needed to deal with the second collective action performance-influencing factor. Social learning is also about creating opportunities for institutional change, relevant in the governance of collective action problems. (Gerritsen et al. 2013; Scholz et al. 2013). Correspondingly, we argue that apart from normative institutions (e.g. issues of trust, reciprocity), regulative institutions as in the form of defining or redefining of rules of engagement for collective action fall within the relational sphere of social learning.

Social Learning as a Communicative Process with Outcomes

A salient notion in the conceptualization and operationalization of social learning relates to seeing learning as a process and/or an outcome. As stated before, in the natural resource management domain, social learning is often defined in relation to a wide range of potential outcomes it may have (Biedenweg and Monroe 2013a; Scholz et al. 2014; Paul-Wostl et al. 2007). The essence of considering social learning as a process with outcomes has been highlighted in literature informed by learning theories from other fields like educational science and organizational learning (Beers et al. 2016; Biedenweg and Monroe 2013b; Senge 2006). Different learning processes can relate to different learning outcomes and an outcome is preceded by various framings, communicative interactions or coalition building processes (Van Der Stoep et al. 2017; Pahl-Wostl 2002; Joss and Brownlea 1999). Hence, one way of making claims about tangible outcomes of social learning is tracking communications and explaining the context in which it was observed (Rodela 2013). By examining communication among participants, it is possible to unravel how learning processes lead to tangible outcomes (Beers et al. 2016; Rist et al. 2006; Pahl-Wostl and Hare 2004; Lamerichs and Te Molder 2002). This constructivist approach regards communication not merely as a process of getting a message across, but as a phenomenon in which those involved construct meanings in interaction (Leeuwis and Aarts 2011; Aarts and van Woerkum 2008; Te Molder and Potter 2005). Communication itself is regarded as an (inter)action that has direct consequences for the social and material world (e.g. relations may be damaged, parties no longer meet, violence may emerge, etc.). From a communication perspective, collective action is, therefore, dependent on a change in communication, representations, and storylines that are mobilized by interaction (Bentley et al. 2018; Leeuwis and Aarts 2011).

Assuming that social learning takes place in communicative activities, means that it must become visible in, and can be identified in communication. As Beers (et al. 2016) highlighted the knowledge, relational and (collective) action dimensions are aspects of the content of the communication (i.e., what people talk about) and the social learning process is a communicative activity that aligns these three dimensions.

There is a distinction between (collective) action in communication, in the form of agreements, proposals or decisions and physical (collective) action (actual implementation of agreements on the ground) (Beers et al. 2016). While collective action in the form of agreements is treated in our study within the communicative aspect of social learning, the actual implementation (physical action) of the communicated agreement is regarded as an external outcome or impact of social learning. In view of our conceptual disposition, social learning is, therefore, a communicative process in which new or changed technical knowledge, social relations and suggested actions become aligned and may turn into collective agreements. As depicted in Fig. 1, these learning outcomes may further effect collective action in the management of late blight.

Fig. 1
figure 1

Conceptual framework with perspectives from Beers et al. (2016), Bentley et al. (2018), Ostrom (1990); Scholz et al. (2013) and Rodela (2013)

Full size image

Informed by the literature gaps highlighted in the introduction and our conceptual framing as well as the case study to which it is geared, this study addresses the following research questions:

Main research question: How does social learning as a communicative process influence collective action in late blight management by farmers? This main research question is sub-divided into:

  1. 1.

    How do aligned technical knowledge, relations and suggested actions emerge in communicative utterances and relate with collective agreements?

  2. 2.

    To what extent and how do collective agreements translate into collective actions in the management of late blight?

Methodology

Case Study Context

The case study was conducted in a village (Goro) found in Wolmera district of Ethiopia’s Oromia regional state. The village was selected considering issues such as production potential (it is designated as ‘potato cluster’ by the district), accessibility and the existence of suitable adjacent potato plots to justify collective risk situation. A total of 33 potato farmers participated in the case study. The livelihood of most farmers in the study area is primarily dependent on the cultivation of potato but also on other cereals and pulses such as wheat, barley, beans and chickpea. Most of the participant farmers are ware potato producers whereby they use a small portion of their produce for household food consumption and sell the rest in local markets as food potato. There was a seed potato producing cooperative in the area that only a few of the farmers used to participate in. The district office of agriculture was in charge of organizing the cooperatives and facilitating access to input supply and market outlets to other potato producing areas in the country. As it is further illustrated in the result section, there had been some trust issues between the farmers and the district office of agriculture, and among the farmers themselves that had consequences to the social learning process and outcomes.

Setting up a Social Learning Context

From a methodological perspective, social learning takes place in an ongoing chain of interactions, such as face-to-face meetings. These interactions offer a communication space for knowledge generation and sharing, social action and reflection (Beers et al. 2016; Leeuwis and Aarts 2011; Gee and Green 1998). The communicative interaction can be a spontaneous, self-organized natural process or something induced and guided by an external facilitator (Koontz 2014). Well-known strategies to facilitate interactions include the organization of discussion and reflection meetings. In addition, practical actions and experiments to visualize unknown biophysical processes with the help of discovery learning tools can be relevant communicative strategies for supporting social learning if they elicit communication (Leeuwis and Aarts 2011; Loorbach 2007; Leeuwis et al. 2002).

A potato scientist, with support from the community extension worker, facilitated the technical learning sessions and the first author of this paper facilitated the reflexive, follow-up meetings. Following Ford’s (1999) suggestion, the design of the learning context was arranged in such a way that it provided generic themes that guided the unfolding conversations around different technical or social issues in relation to late blight management. In this regard, the social learning context was configured through concurrent technical (conceptual and practical) sessions and reflexive meetings that were meant to inspire discussions, learning and collective action in the management of the disease. There was a total of ten such learning sessions that contained technical sessions with reflexive follow-up meetings. Each learning session took an average of two and half hours. The first two sessions were focused on disease dynamics (definition, cause, symptoms and spreading mechanisms) and opportunities for disease surveillance. The remaining eight sessions, spread across the four-month growing season of the potato crop (June–September), were designed to facilitate social learning as participants were able to exchange perspectives, negotiate, discuss success and pitfalls, and agree or disagree around different technical and relational issues in the management of the disease. As listed below, most of the learning sessions had three major parts; the first part was covered mainly in the technical sessions and the other two during reflective meetings. However, as different discussion topics were interrelated, there were apparent overlaps of the discussion agendas across the learning sessions.

  • Facilitated discussion and field practical observation on the disease and its management

  • Report and reflection session on activities since the previous learning session

  • Deliberation on activities before the next learning session

The technical sessions provided a learning space on less visible biophysical aspects of late blight, such as signs and symptoms, causal agent and spreading mechanisms. A magnifying glass, YouTube videos and potato sizing squares were used to showcase and initiate discussion around late blight spreading mechanisms, symptoms and seed selection as management practices. Practical sessions on some relevant late blight management options took place through observation and discussion on plant growth, weather conditions and on how to implement different management options. The practical sessions were partly informed by CIP (International Potato Center) previous learning material (Cáceres et al. 2008). For the technical learning sessions, a small community-managed plot (30m2 x30m2) was secured from a farmer and was planted with a third-generation potato seed provided by Holeta Agricultural Research Center.

Data Collection

Utterances from the ten learning and reflexive meeting sessions were fully captured to particularly answer our first research question on how social learning outcomes as an alignment of changed knowledge, relations and actions were developed in participants’ communicative utterances. Continuous documentation and audio recording were made on each learning session in which different learning themes about the disease and its management were covered. During these facilitated sessions, particular attention was given to participants’ reflections on new insights related to the disease; the way they made sense of their existing knowledge or practice in view of new information; and how and what social or technical practices they thought had to be unchanged or modified in the management of late blight. Plenary field discussions during the technical learning sessions and follow-up, reflexive meetings (which together constituted one learning session) were important deliberative spaces for generating this information. Data for our second research question that assessed the extent to which social learning outcomes were translated into collective actions on the ground was obtained from a monitoring sheet that was filled in and presented by a monitoring committee during the reflexive, follow-up meetings with all participants. The monitoring sheet captured information on farmer name; type of collective practice; date of monitoring; way of communication; adherence/non-adherence; measures taken. The researchers’ field observation and participation in some of the monitoring work were additional sources of information.

Data Analysis

The study report is based on analysis of extensive meeting notes or verbatim meeting transcripts. Thus, our units of analysis are transcribed utterances of participants during the learning and reflexive meetings. Each social learning event was regarded as a particular communicative space containing various utterances of participants and hence fully transcribed verbatim. As different learning themes around the nature of the disease and the management options did not necessarily begin and end with the learning sessions and usually went beyond more than one session, all discussions from the different learning sessions that related to a particular theme (e.g. fungicide spray) were openly coded on Atlas.ti software. Accordingly, we defined five themes in the communication: nature of the disease, fungicide use, late blight resistant and clean seed potato, killing/rouging volunteers and dehaulming.

The second stage, axial coding followed, drawing on our conceptual line of thinking on social learning as an alignment of new or changed knowledge, relationships and actions. In doing this, all quotes under the five codes were analyzed for text segments in which new or changed technical knowledge, relation and suggested management actions and practices were aligned. In nine data segments, the three dimensions had become aligned, leading to a collective agreement. These segments were taken as evidence of a process of social learning and are discussed as emergent sub-topics under the different codes. Collective agreements that followed upon such moments of alignment were regarded as the first outcomes of social learning. Finally, the analysis of the translation of collective agreements into physical collective action was based on the information from a monitoring report by the farmers’ committee and the notes from follow-up discussions in the reflexive meetings.

Results

The results section consists of three major parts. The first two sections (4.1 and 4.2) present how the social learning process as a communicative interaction led to social learning outcomes across the different discussion topics entertained in the learning and reflective sessions. The third section (4.3) presents a summary of the level of translation of the social learning outcomes into collective action.

Learning about the Nature of the Disease

Discussions about the disease in relation to its cause, symptoms and spreading mechanisms sparked different kinds of relational and action-oriented discussions. As can be seen from Fig. 2, utterances about the disease shifted from regarding late blight as a weather problem to regarding it as a disease with multiple infection sources. During this initial stage of the learning sessions and particularly in the first two sessions, new insights about the causeFootnote 3 and spreading mechanism brought the issue of interdependency among neighboring farmers to the center of conversations. The realization of late blight as a communal problem provided many of the participants a new lens on the way they see farmer-to-farm relationships in the management of late blight. There were marked comments on how farmers could be affected by the situation on neighboring farms and what can be done about the disease. As many stressed the importance of cooperation in managing the disease, a proposal was made by a few farmers, and endorsed by all of the other farmers present, to involve other farmers in the community that were not in the meetings. Table 1 shows how this collective agreement was related to alignment of the three dimensions of the topic in the learning process. The proposition from this group of farmers was based on articulations of their understanding of the contagious nature of the disease and their apparent practical observations of how their field had been infected directly after they had observed symptoms on adjacent fields.

Fig. 2
figure 2

Shifting conversations about the disease

Full size image
Table 1 Learning to bring on board other farmers in the community
Full size table

An interesting proposal that followed discussions on late blight transmission and the communal risk of the disease was the importance of monitoring late blight incidence beyond their individual farms. Even though all farmers inspected their own fields to a different degree, an agreement that a group of neighboring farmers should team-up for an early morning joint field scout was accompanied by statements about the communal risk nature of the disease and the need for adjusting existing individual-based practice in monitoring. The joint field scout was an idea that was particularly pushed by a few former seed-producing farmers who had experience of crop disease monitoring.

Conversely, some farmers argued they could control the disease on their field, irrespective of what is happening on neighboring farms. As can be observed in some of the statements below, their choice to continue working individually was mainly heralded by utterances of their previous experience of keeping the problem at bay by optimal spraying, despite their belief that late blight is a contagious disease: “I am always well prepared with chemicals; my field is usually safe when neighboring fields are hit with Wag [late blight]”...“As Baysa said, if I spray enough I can protect my own field”....“The owner himself can control it daily; it doesn’t cause that much problem.” Of these five farmers, all but one eventually decided not to be part of the collective monitoring work and missed the subsequent learning and reflexive sessions.

Different statements disclosed that the farmers, in general, had experience with some sort of monitoring in their diverse social endeavors, such as community health and security matters. A fascinating relational issue that was raised and discussed before an agreement was reached on joint monitoring was how to deal with the existing traditional view that strangers entering crop fields leads to a drop in productivity. This was particularly raised by some younger farmers as a potential problem for traditional or superstitious members of the community. Although none of the farmers openly said that they did not want other farmers to enter their farm, they all generally acknowledged this as an existing belief that might pose a challenge to joint late blight monitoring. As illustrated in Table 2, the discussion around this issue was concluded with a collective agreement that all farmers should welcome a monitoring committee or a neighboring farmer in the field scout team.

Table 2 Learning about team field scouting and collective monitoring of late blight incidence
Full size table

The idea of what to do with non-compliance was initially brought to the attention of farmers by the facilitator after farmers had decided to introduce team monitoring of late blight incidence. The majority of the farmers were not enthusiastic about sanctions in response to non-compliance, saying everyone should monitor late blight and that sanctions might create friction. Some farmers were less optimistic about compliance and argued that rules were needed: “we can’t be 100% sure that he [a specified farmer] will keep his promises. That is why we need a committee and also the rules.” Another farmer added, “Any kind of job which will be done in a group must have a ground-rule. If you are a problem for me, we need a common agreement that will prevent you from becoming a problem.” On the contrary, most farmers agreed with the views of two farmers who stressed “We now know the disease risk, and everyone will do it for their own benefit” ... “if there are people who don’t participate in the monitoring, it is better to advise them with love than force them.”

As illustrated here, discussions about the disease and its infectious nature led to two social learning outcomes. First, new technical knowledge about the cause and spread of the disease, reflection on their individual monitoring practices and a new way of seeing farmer-to-farm relationships in managing late blight facilitated the agreement to bring other non-participant farmers into their joint work. Second, a shift from seeing late blight as a private problem to a collective problem (new insights on the risk of late blight spread) steered discussion around cooperation and how to deal with an existing belief (bad luck) that led to a proposal for a joint field scout and establishment of a monitoring committee.

Learning about Late Blight Management

In subsequent learning sessions, deliberations on late blight management opened new windows for further reflection on the long-practiced management technique (spraying) and other late blight management options. Analysis of the different statements revealed how communications unfolded to give rise to a range of social learning outcomes.

Fungicide Use

Fungicide use for controlling late blight is a well-known practice by all the farmers and its importance was not a point of earnest discussion. Many participants reiterated the view shared by a farmer, namely that: “chemical has always been our savior.” But as late blight is now recognized as an infectious disease, a new interest surfaced among farmers in knowing if everyone is indeed spraying and if the monitoring committee can check spraying when they receive a report on late blight incidence. The five committee members agreed to share responsibility and individually check spraying on potato fields that are closer to their field. However, during a reflexive meeting after the first monitoring work on spraying, many farmers expressed their dismay that the committeeFootnote 4 members did not monitor the spraying after they were given reports on late blight incidence from the joint scouting team. This led to a debate on whether the usual representatives are the right people for the work and on the possibility of considering other committed farmers as committee members. This led to an agreement to replace four of the five committee members as shown in Table 3.

Table 3 Learning to replace members of the monitoring committee
Full size table

A new notion to most participants was the existence of different types of chemical fungicides with distinct modes of action in late blight control. While some of the farmers shared their knowledge that there are two types of fungicides (Ridomil and Mancozeb), all farmers confirmed that they only used Ridomil (a systemic fungicide) which has market dominance. During discussions, farmers learned that the two fungicides types have different modes of action. Experts considered that continuous use of Ridomil might lead to metalaxyl resistance from the pathogen while farmers questioned whether resistance might be responsible for their increased frequency of Ridomil spray over the years. Most farmers wanted to stop the continuous use of Ridomil and to alternate with Mancozeb which needs application before symptoms are observed. A few farmers, who already bought and saved enough Ridomil for the growing season, rejected the idea of alternating with Mancozeb. After the new committee reported that seven of the 25 farmers who had agreed to spray Mancozeb did not report their spraying schedule for the committee to check, the idea of enforcing rules became again a point of discussion. A consensus was reached (Table 4) that a fine should be proposed at the next meeting when everyone is present to avoid confrontation from absentees. At the next reflexive meeting, an agreement was reached to fine 100 Birr (Ethiopian currency) for those who either missed the joint team scout or failed to inform the committee on the status of late blight and follow-up of the implementation. A further 50 Birr was agreed as the fine for being absent from reflexive meetings.

Table 4 Learning about sanctioning systems on non-compliance
Full size table

After the third monitoring on spraying, farmers realized that putting a system of sanctions in place could still not guarantee full compliance. Two farmers who had failed to report on spraying said that they could not spray because they did not have money to buy fungicide. While some farmers tried to protect these two farmers from possible fines, others argued for financial support for these farmers because all farmers are at risk if two farmers are unable to spray. In this regard, the practice of mutual aid that they do in other aspects of their social life is exerted to late blight management giving a new dimension to community reciprocity in managing contagious crop diseases. Table 5 illustrates how the proposal for mutual aid was modified at the same meeting with a scheme to use the money from fines to buy fungicides for the farmers.

Table 5 Learning to share fungicide cost of ‘risk-causing’ farmers
Full size table

Analysis of communications around fungicide use as a late blight management option featured three different social learning outcomes. First, knowing late blight as a disease that spreads, as opposed to the predominant assumption that it is a weather problem, alerted farmers to bring the monitoring aspect not only to spraying but also to assess the performance of the existing monitoring committee and replace the members. Second, a new keenness to experiment with Mancozeb to avoid the potential metalaxyl resistance, encouraged the farmers to reconsider their stance on dealing with non-compliance and to propose and forge sanctioning mechanisms. Third, the practical insight that punishment could not guarantee compliance led to a decision to financially support farmers who had not complied. As can be seen from the quotes in Table 5, this decision was made because of the perceived late blight risk from unsprayed fields.

Using Late-Blight Resistant Varieties and Clean Seed Potato

Learning on the multiple sources of late blight infection in general and infected potato tuber as an inoculum source, in particular, steered discussions around the use of healthy seed from resistant varieties to deal with late blight. New insight was made into the distinction between clean/disease-free seed and late-blight resistance varieties. Many farmers initially assumed that disease-resistant varieties would remain resistant to late blight. From some of the statements, this is partly attributed to the original framing of late blight as a weather problem. The realization that their widely grown varieties, such as Gudene and Jalene once released as late blight resistant,Footnote 5 are no longer free from potato disease, led farmers to stress the importance of keeping their own seed clean from late blight. In this regard, proposals were made to undertake seed selection for the coming growing season, using a third-generation (G3) resistant variety (Gudene) for a positive selection.Footnote 6 Seed selection would be made with the help of an extensionist and a researcher from the Holeta Agricultural Research Center which would also provide the G3 seed.

The importance of keeping their own seed source free from late blight instigated reflections around the previous and potential future role of the district government agricultural office in seed quality checking. A few farmers, previously members of a seed-producing cooperative, had experience of working with the regional regulatory office and the district agricultural office in potato disease (Bacterial wilt) monitoring and seed potato marketing. They considered that this had been an unfair arrangement because only a few farmers with ‘good relations’ with the offices enjoyed the market benefits. Some of the other farmers seemed to know and embrace the stories and others appeared to be new to the details. Existing negative views about district offices and regulators were amplified by stories shared by former seed producers. The discussion led to a more restrictive proposal for a potential role for the district office in seed quality monitoring.

Communications around seed quality elicited two social learning outcomes. New insights on the distinction between healthy/clean and resistant variety and the contribution of farmers handling of seed to late blight prevalence opened new avenues for farmers to adjust their opinions about researchers from someone who failed to continually provide ‘quality’ seed to someone who can also assist in keeping farmers seed clean from the disease. This led to an agreement (Table 6) to seek researchers’ support in seed selection and handling. The other learning outcome (Table 7) aligned discussions about the importance of seed quality checking with previous relation and collaborative practices with government district offices. Farmers advocated for a principally farmer-controlled seed monitoring approach, dismissing a potential role for the district agricultural office in seed quality monitoring.

Table 6 Learning towards collaboration with researchers on seed selection
Full size table
Table 7 Learning towards a restrictive role for district government offices in seed quality monitoring
Full size table

Removing Volunteers

Social learning in relation to removing volunteers as a late blight management method is stimulated by awareness of how volunteers (Gebo, namely potatoes growing after being left in the soil from the previous harvest) can be another source of late blight infection. Some farmers removed volunteers only when they wanted to plant another crop in a subsequent growing season but not because volunteers are an inoculum source for the pathogen. When another crop is not being planted, most farmers let the volunteers grow for a second-round harvest. Having realized the role of volunteers in pathogen spread, the issue of land contractual arrangements was raised by farmers who grow potato on a leased plot in a neighboring village. In the usual contractual agreement, the leasing farmers only harvest one time and must leave behind the volunteers for harvest by the property owners. Given the role of volunteers in pathogen spread, farmers proposed (Table 8) changing their contractual agreements to reduce volunteers. Table 9 also shows how the majority of farmers agreed to collectively implement the practice of removing volunteers from their own plot, irrespective of what they cultivate after potato.

Table 8 Learning to include the issue of control over volunteer plants in land contractual agreements
Full size table
Table 9 Learning about joint monitoring of volunteer removal as a late blight management practice
Full size table

There were two social learning outcomes related to this late blight management practice. The first one relates with knowing volunteers as pathogen inoculum source that made farmers re-define existing practices and relationships in land contractual agreements. Second learning evidence is an agreement reached by more than half of the farmers to remove volunteers and to collectively monitor this practice. This time not because they see volunteers as a weed to another crop but because it can serve as inoculum source for the late blight pathogen.

Dehaulming

During the session, farmers learned that late blight can travel down from the plant (through the transfer of fungal spores) to the tuber when the haulm, comprising the aerial part of the plant, is not cut on time. However, this new insight did not lead to a learning outcome or any form of collective action in communication. Few farmers stated they had previously removed the haulm, believing it is good for tuber size and skin setting. However, they contended that the practice exposed their black soil (Koticha) which easily cracks and the tuber to direct sunlight, turning the tuber green. Given that some farmers considered that green tuber was a major reason for rejection at the market, the farmers decided not to introduce dehaulming as a management option unless there was a mechanism to avoid the change in tuber color. However, two farmers, who had no previous knowledge of the practice, said that they wanted to experiment with it on a small portion of their potato field to evaluate any effect on late blight infection or tuber color.

Linking Social Learning Outcomes with Collective Action

Analysis of the discussion around different learning themes on late blight disease and its management revealed a total of nine social learning outcomes in the form of collective agreements related to an alignment of new communicated knowledge, discussed relations and suggested actions. Table 10 illustrates the extent to which the various agreements implicated in the discussions were implemented on the ground. From a process perspective, it can be observed that there is not a one-way relationship between a collective agreement and a collective action or that collective action comes necessarily after collective agreements. For instance, agreements to change the monitoring committee, to develop sanctioning systems or to share fungicide costs are informed by reflections from practical actions. In this sense, there was an iterative learning cycle whereby farmers contentiously modified or adapted their collective action by reflecting on their collective agreements and vice versa.

Table 10 Translation of collective agreements (social learning outcomes) into collective action
Full size table

As can be seen in Table 10, two of the nine collective agreements could not be fully translated to collective action. For example, only about half of the field scout team did a regular joint visit to their farms. A possible explanation for this might be the problem of free-riding that is common in collective action problem situations. The free riding tendency by some farmers eventually led to a modification of the agreement around sanctioning mechanisms. The fact that farmers from nearby villages were not part of the collective agreement on amending land contractual arrangements for removing volunteers had influenced the translation of the agreement into action.

Discussion

The study unraveled how social learning as a communication process in which new knowledge, relations and actions become aligned leads to collective agreements (first research question) and collective action (second research question) in the management of late blight. In answering the first research question, we discerned several communicatively agreed lines of action that were attended by communication regarding new knowledge, relations and actions. In this regard, new knowledge in relation to the threat posed by late blight as a contagious disease was instrumental in instigating discussions about relations and potential collective actions, thereby leading to several social learning outcomes. Although our review failed to find previous studies with a grounded perspective on late blight as a collective action problem, empirical studies of pest and invasive weed management highlighted that collective responses were effective when communities appreciate that the problem crosses farm boundaries and where benefits might arise from cooperation (Graham et al. 2019). A change in the framing of late blight by the community from a weather problem to a contagious disease involving pathogens fundamentally shifted awareness of late blight disease from a private bad to a collective bad. This, in turn, instigated social learning as new knowledge on the mobility and the contagious nature of late blight, linked to new perspectives on interdependency and proposals for collective action to manage the disease. The empirical study thus confirmed our assumption that disease management has particular features compared to the collective action problems generally studied in social learning literature. It differs from dominant social learning literature on collective action problems in natural resource management, migratory pests and invasive weeds which implicitly assume some sort of understanding of the collective nature of the problem, hence picks the role of social learning from there (Zamani et al. 2018; Marshall 2016; Brugnach et al. 2011; Toleubayev et al. 2007). The key role of technical knowledge in determining social learning outcomes is also demonstrated in a situation where farmers did not agree to implement dehaulming as a potential management practice, owing to their technical knowledge of the impact of direct sun on tuber quality.Footnote 7 We contend that attempts to develop and sustain collective action around diseases that are characterized by less visible dynamics should be supported with a strong learning element around the delicate life cycle and spreading mechanisms, demonstrating the interdependency of farmers. In addition, the contribution of social learning can be amplified if it takes a step back and firmly integrates such learning in a situation where there is ambiguity on the very collective nature of the problem.

However, a better understanding of the mobile nature of the disease was not sufficient reason for some of the farmers to cooperate in disease monitoring or to reach agreements on the collective implementation of management practices. Their main argument was that they could individually control the late blight if they spray optimally, irrespective of the disease situation in the area. This finding is consistent with a case study on community-based resistant weed management whereby, despite knowledge of weed mobility, some community members perceived weed control as an individual problem by taking into account the economic costs (e.g. finance, labor) of managing resistance over time (Ervin and Frisvold 2016). Similarly, lessons from situational studies on banana wilt and potato late blight in Central Africa and Ethiopia demonstrated that the public good features of a disease management strategy depend not only on specific agro-ecological properties of the disease but also on the available technological options and other community-specific social conditions and resources (Damtew et al. 2018; Leeuwis et al. 2018; McCampbell et al. 2018). This has an important implication for the framing of contagious crop diseases, such as late blight, as public bad mainly based on its mobility. We hope that this study triggers more in-depth scholarly discussions on what other biophysical, social or economic attributes, together with disease mobility, could be integrated to advance current theoretical perspectives that present contagious crop diseases as a public bad.

Communication about late blight as a shared risk played a key role in reaching agreement on collective action. This stimulated discussions on new types of social relationships for collective monitoring, including a committee and neighbor-to-neighbor field scouts. In this regard, the monitoring system performed two key relational functions: joint information collection and sharing on late blight incidence not only informed collective management activities but also leading to sanctions and new reciprocity as communities strove to deal with non-compliance. This mirrors the assessment by Ervin and Frisvold (2016) that when there is a shared understanding of the collective nature of the problem, interest can develop within communities in conducting monitoring and applying appropriate penalties when some operators do not comply with locally prescribed management practices. Even though farmers initially decided to implement monitoring, despite traditional superstition around letting others into their individual fields, they were not enthusiastic about sanctions. However, the fact that they eventually developed sanctioning systems and forged new ways of reciprocity through mutual aid in buying fungicides highlighted the observation by Ibuka et al. (2014) that social learning helps farmers to amend their perceptions of risks and their collective rules of engagement. In this regard, our research provided evidence that social learning can lead to the emergence of a rudimentary form of institutionalized coercion which can help to reduce overcome non-compliance in the management of contagious crop diseases, such as late blight. In a hypothetical game scenario, Zhang et al. (2014) similarly reported that ‘toy-communities’, guided by payoff considerations, started with a clear preference for the treatment without punishment but shifted to peer-punishment after a few rounds. We consider that the emergence and evolution of regulatory institutions through social learning has significant theoretical implications for the way scholars conceptualize social learning in the management of agricultural collective action problems in general and contagious crop diseases in particular. In this respect, a great majority of the reviewed literature on the subject appreciates regulatory institutions (e.g. monitoring, incentive mechanisms, graduated sanctions) from an institutional theory perspective, principally seeing it as an externally imposed system, rather than as something that can potentially emerge through social learning (e.g. Wulandari and Inoue 2018; Hegde et al. 2017; Six et al. 2015; Christmann et al. 2015; Pahl-Wostl 2009).

In relation to our second research question, our aim was to examine the influence of social learning on collective action in the management of late blight, focusing on the actual implementation of collective agreements. As can be noted from the results, while most farmers agreed that collective actions should be implemented, a few of these proposals could not be translated into action. For instance, farmers tied to property owners from a neighboring village via contracts could not translate the agreement to modify the rights to control volunteers into action. This relates to Ostrom’s (1990) thesis on the importance of delineating clear boundaries between users and non-users of a resource. Although public good boundaries often have a gradient quality or are fuzzier (especially for highly mobile resources) than implied in the traditional public good design principles (Baggio et al. 2016; Cox et al. 2010), having a clear definition of geographical and institutional boundaries appears to be an important precondition for the translation of agreements into collective action. In this regard, unclear boundary definition for late blight management represented a missed opportunity for our social learning approach to inspire better cooperation between two seemingly different communities which are institutionally interrelated through land contractual agreements in potato cultivation. Although many of the agreements were physically implemented, there was visible non-compliance from some farmers, mainly on team scouting and on reporting on disease incidence.

Contrary to a range of evidence from public good studies, non-cooperating behavior of a few farmers did not seem to stop other farmers from complying with most of the collective agreements. This appears to provide some support for the conceptual premise of Ervin and Frisvold (2016) that farmers may still have an incentive to manage invasive or contagious crop problems and benefit from the adoption of practices, even when their neighbors do not and even when acting collectively is considered the best strategy.

This study showed how social learning, through aligned knowledge, relations and actions in communications, formed the basis for collective agreements and collective action. However, we are limited in being able to explain the translation from learning to action. Multiple factors may have influenced the extent of implementation of collective actions implicated in the social learning outcomes: levels of risk perception, information sharing/communication, production objectives, group heterogeneity, and resource boundary (Graham et al. 2019; Cox et al. 2010; Ostrom 2010). In this regard, there is considerable scope for more research to explore and detail the different factors in action around the translation of collective agreements into actual practices on the ground.

Conclusion

We conclude that social learning and associated collective actions in the management of late blight are outcomes of an ongoing communication process with a range of interlinked technical, relational and action related issues. Outcomes of such a learning process can thus be made visible through the analysis of communicative processes. In this regard, the conceptualization of social learning in the management of contagious crop diseases would benefit from a broader view than what is commonly implied in the traditional social learning literature on natural resource management. This means, on the one hand, that conceptualization needs to strongly support community learning in relation to clearing ambiguities on the collective nature of the disease and, one the other, that this conceptualization needs to widen its scope for further systematic analysis of implementation of agreed collective actions. The crucial role of regulatory institutions in the management of public bads, such as late blight, also warrants social learning to embrace the emergence or evolution of such institutions in its conceptual and methodological approaches. We contend that social learning is a necessary precondition for collective action (the implementation of agreements in our case), but may not necessarily lead to it, given potentially multiple factors in operation. This interplay between social learning and collective action, specifically the role of different factors in the translation of agreements into practical action for the management of contagious diseases, requires further exploration.