Uncovering the Landscape of Participatory Research in Agricultural Innovation: A Scoping Review

Formulating the research question

The review aimed to answer the question “How are farmers involved in research activities within agricultural systems to promote innovation in agricultural practices?”

Identify relevant studies

Two databases were considered to search for relevant papers: Scopus and Web of Science. The search string was formulated based on the concepts “participatory methods,” “agriculture,” and “innovations,” and we searched in title, abstract, and keywords. The search terms, particularly those related to innovation, were intentionally selected as potential synonyms. This approach aimed at casting a wide net to ensure the inclusion of a broad range of papers that might discuss innovation in diverse ways. The rationale was to conduct an exhaustive search and allow for a manual refinement process, whereby non-relevant studies would be excluded during the screening phase. A search was undertaken in February 2023 using the following search string:¹

TITLE-ABS-KEY (agri* OR smallholder* OR farm*) AND TITLE-ABS-KEY (“action research” OR “active citizenship” OR “citizen science” OR “collaborative research” OR “community based participatory research” OR “community-based research” OR “community-led research” OR “community-participatory research” OR “cooperative research” OR crowdsourcing OR “participat* research” OR “participatory action research” OR “participatory assessment” OR “participatory co-creation” OR “participatory co-design” OR “participatory evaluation” OR “participatory exercise” OR “participatory method” OR “participatory monitoring” OR “participatory process” OR “participatory science” OR “participatory study” OR “stakeholder-led research”) AND TITLE-ABS-KEY (“innovation model” OR “innovation framework” OR “innovation” OR restructuring OR renovation OR modernization OR remodelling OR revolution OR recasting OR reorgan*).

Study selection

The resulting unique papers were uploaded to Rayyan (Ouzzani et al. 2016), a collaborative software used for conducting blind-mode screening in systematic reviews. This enabled researchers to independently assess each paper, ensuring that at least two researchers reviewed each paper.

All papers needed to meet the following inclusion criteria:

1. Participants: Studies were included if they were conducted in an agricultural system and involved farmers (here defined as individuals who run the farm themselves, or who are employed as farm workers).

2. Outcomes: Studies aimed to enact changes in production or resource management practices.

3. Concept under investigation: Farmers or producers were involved/participating in at least 1 of the 10 operations in which citizens can be involved based on the definition of the possible models for public participation in scientific research as described by Bonney et al. (2009).

4. Study design: Publications were original research papers describing the collection of primary data and/or an intervention and were not any form of review.

5. Characteristics: Studies without temporal restrictions were considered for inclusion.

6. Language: Studies not in English were excluded. English was chosen for its global accessibility, comprehensive literature coverage, and widespread scientific convention.

The titles and abstracts of all retrieved articles were independently reviewed by two researchers who made inclusion decisions based on the predefined criteria through consensus. Subsequently, two authors reviewed all articles selected for full-text review. In cases where conflicts arose during the full-text review, a third researcher was engaged in discussions until consensus was reached.

Collating and summarizing the results

A standardized data extraction sheet was developed and completed by two researchers containing the following elements:

1. Theoretical framework used in participatory research: whether the authors explicitly mentioned a theoretical framework that guided how they participated in their project.

2. Phases of the project: when named and cited, the various phases of the project were reported.

3. Duration of the project: if timeframe was clearly declared by authors or if it could be reconstructed from the dates of the project phases.

4. Type of participants included: all the participants involved in the project.

5. Barriers: what hindered the development of the project; only problems explicitly stated by the authors were considered.

6. Facilitators: what authors stated enabled or supported project development.

7. Innovation pursued: the change in production practices sought through participatory design.

8. Modes of participation: this classification was derived from the models for public participation in scientific research as described by Bonney et al. (2009), who outline ten steps through which citizens can engage in scientific research projects. These steps are illustrated in Figure 1.

Figure 1

Analysis procedure

The researchers labeled the above eight elements on the text with the support of Atlas.ti and then extracted and reported them in a data sheet for analysis. The country where the study was conducted, project timeframe, theoretical frameworks of participation, and research project phases were mapped and described narratively. After being labeled and extracted from each paper, the ten steps defining modes of participation were summarized, and then classified into three steps, contributory, collaborative, and co-created, and were subsequently discussed in narrative form. According to the original models by Bonney et al. (2009), a project is considered contributory if citizens are involved only in stage 5, and possibly in stages 7 and 9; collaborative if citizens are involved in stages 5, 6, 7, and possibly 4, 8, and 9; and co-created if citizens participate in all ten stages.

However, this classification was adapted to better fit the projects reviewed, as Bonney’s models was developed within the context of citizen science, which generally aims to produce new knowledge. In contrast, the studies included in this review are focused on changes in production practices and innovation. This shift in focus alters the role of farmer participation, as farmers in these projects are more engaged in implementing changes and assessing outcomes with researchers than in merely collecting or analyzing data. Consequently, the task of data analysis tends to fall more naturally within the responsibilities of the scientific team, without necessarily excluding farmers from the process.

This disconnect in applying Bonney’s models could be attributed to the supportive role of science in these types of projects, where the objective is to encourage practical changes rather than to only gather data. As a result, the two pivotal phases for determining participation are phase one, defining the research question, and phase four, co-designing the methodology. If farmers are not involved in either phase, the project is classified as contributory; if they are involved in one of the two, it is collaborative; if they participate in both, it is co-created.

Although the stages of scientific research in which farmers can participate are not equivalent categories, a participation level from 1 to 10 was assigned to each study based on the extent of farmer involvement across different research phases.

Finally, barriers and facilitators were identified through a two-step coding process. First, instances explicitly labeled by the authors of the reviewed papers as barriers or facilitators to the participatory process were coded in a top-down fashion. Then, within each identified barrier or facilitator, a bottom-up coding process was conducted to extract specific descriptors from the text. These descriptors were later thematically grouped during the analysis phase.

Data extraction was made with the support of Atlas.ti (Smit and Scherman 2021) software for qualitative analysis, to increase the transparency of the coding by making them public. Those files are available on request to the authors while the extraction and analysis sheet is available in the supplemental materials and is described in the section 2 of Supplemental File 1: Appendix A.

Search results

A total of 807 papers were initially identified from Scopus and Web of Science. After the screening processes, 34 papers, published over a period from 1994 to 2022, were selected as detailed in the prisma flow chart depicted in Figure 2. Most of the papers (58) assessed for eligibility that were excluded were research projects not aimed at innovation and thus changing farmers’ production practices. Another 49 papers were excluded because they did not collect primary data or were not direct reporting of an agricultural innovation project, such as reviews or commentary papers.

Figure 2

General characteristics: countries and projects’ timeframes

Projects were conducted in 26 different countries; two projects included sites that spanned multiple countries (Andrieu et al. 2019; Dolinska et al. 2020). The majority of the countries were in Europe (12) and Africa (12), with Italy having the most studies overall (4).

The average duration of projects was approximately 5 years, with a range from a minimum of 8 days to a maximum of 22 years, although data for 4 cases could not be retrieved from the paper. In particular, 2 projects lasted less than 1 month (Dorward et al. 2003; Kimeli et al. 2017), 4 were finished in 1 year (Berthet et al. 2020; Burbi et al. 2016; Calliera et al. 2021; Chowdhury et al. 2010), 8 in 2 years (Andrieu et al. 2019; Carton et al. 2022; Krupnik et al. 2012; Osorio-García et al. 2019; Pagliarino et al. 2020; Périnelle et al. 2021; Sturdy et al. 2008; Valdivia et al. 2014), 7 in 3 years (Aare et al. 2021; Ayre and Nettle 2017; Bloch et al. 2016; Dlott et al. 1994; Dolinska et al. 2020; Orlando et al. 2020; Poussin et al. 2006), 3 in 4 years (Ciaccia et al. 2019; Duker et al. 2020; Mitchell et al. 2001), 2 in 5 years (Srinivasan et al. 2022; Temple et al. 2006), 1 in 7 years (Bhadana et al. 2015), 1 in 8 (Swiergiel et al. 2019) and 1 in 22 (Humphries et al. 2015).

It is important to note that the project duration typically does not include the preplanning phase carried out by researchers before involving other partners and participants.

Type of innovations promoted

Each research project included was conducted with the aim of achieving an innovation in farming practices or resource management. These innovations were subsequently classified thematically into five main related themes of innovation: climate change, crop choice, irrigation, pesticides and herbicides, and technical improvement.

The most frequent innovations pursued in the included projects were related to crop choice and cultivation systems, for example, including more legume cultivation in farms (Carton et al. 2022; Périnelle et al. 2021) or planting trees in crop fields (Hocking and Islam 1994). Climate change–related projects address challenges such as Greenhouse Gas (GHG) mitigation (Burbi et al. 2016), or they strengthen the adaptive capacity of farmers (Kimeli et al. 2017). Practice improvement innovations were mainly related to improvement in rice cultivation (Bhadana et al. 2015; Orlando et al. 2020; Pagliarino et al. 2020; Ussivane and Ellwood 2019). Innovations related to irrigation were, for example, aimed at improving an irrigation district (Ayre and Nettle 2017) or increasing the food production with a change in irrigation scheme (Dolinska et al. 2020). Finally, innovations related to herbicides and pesticides mainly tried to reduce their use (Calliera et al. 2021; Della Rossa et al. 2022; Swiergiel et al. 2019).

Some projects were not primarily focused on introducing specific innovations, but rather on supporting the farming community in identifying challenges and collaboratively developing solutions to address them in an innovative way (Ciaccia et al. 2019). These projects did not necessarily prioritize scalability, but rather focused on the lessons learned in terms of science-citizen collaboration. However, there were cases where innovations were directly aimed at advancing scientific knowledge in a broader sense (Aare et al. 2021; Orlando et al. 2020).

Type of involved participants

In the analyzed papers, the reporting of participants varied across the studies. Some studies provided tables that specified the type and number of participants (e.g.: Dlott et al. 1994; Hocking and Islam 1994). In other studies, participants could be identified only through descriptions of the research process (e.g.: Chowdhury et al. 2010; Dorward et al. 2003; Humphries et al. 2015; Srinivasan et al. 2022).

A total of 75 different participants were coded and classified into 15 categories, which were further grouped into five main categories: citizens, experts, nongovernmental organizations (NGOs), private companies, and public organizations, as shown in Table 1.

The citizen group consists of farmers, their associations, and members of the local community, all of which were present in every project considered. Notably, all the studies included smallholder farmers, a category of agricultural producers who typically operate on a smaller scale and are more vulnerable to socio-economic and environmental challenges. The category of experts was further divided into three main types: technicians, scientists, and facilitators. Technicians are professional advisors with specific expertise in agriculture-related areas, such as pest control advisors (Mitchell et al. 2001). Scientists refers to academics involved in the research in addition to the authors of the paper. Facilitators, the most prevalent type, operate at the intersection of the other two categories and the citizens, often serving as extension agents. However, it is noteworthy that in certain cases, researchers themselves also engage in farming activities, such as in the case of Mas Masumoto (Dlott et al. 1994).

The NGOs involved primarily focus on community development and support. Private companies encompass a wide range of entities, including processors such as bakeries, suppliers of various products such as seeds, certification bodies (for products like wine), companies within the industrial supply chain, and investment firms. Public organizations predominantly consist of government agencies and environmental protection institutions.

Theoretical framework for participation and research steps

Except for one among the included studies (Della Rossa et al. 2022), which did not specify the theoretical framework adopted to guide the participatory approach design, a total of 27 theoretical frameworks emerged from 34 papers, as shown in Table 1 in Supplemental File 1: Appendix A. However, the authors often state that while a theoretical framework inspired their work, it was adapted to suit the specific context and purpose.

In seven papers, it was not possible to clearly delineate the phases or the steps followed in the participatory research, whereas in the 27 cases in which it was possible, the majority of the participatory projects envision 3 phases (33.3%), 4 phases (18.5%), or 7 phases (18.5%), with a range that fell between 2 and 8 steps.

In two papers (Ciaccia et al. 2019; Pagliarino et al. 2020), phases were indicated as a cyclic loop, and in one case subprojects were used instead of phases (Swiergiel et al. 2019). Finally, it should be taken into account that those steps were in some cases not considered a linear process but a series of overlapping phases that alternate continuously, as in the case of the data-information-knowledge-wisdom (DIKW) hierarchy approach that guided Orlando et al. (2020).

Modes of participation

Participation in scientific research projects varied among the included papers. The average score obtained from the sum of the presence of the ten steps in which farmers could or could not be included in the research was 4.9, with a most frequent score of 6. Regarding extreme values, a score of 1 was observed in only one case, while no cases scored 10, which would represent full involvement of participants in every phase of the project, from research design to implementation. All projects included in our review, with the exception of two (Carton et al. 2022; Périnelle et al. 2021), should be defined as contributory. The two phases in which farmers participated most frequently were the co-design of methodologies for data collection or experimental changes, and the definition of research questions or project-related issues. These phases imply co-creative participatory processes, even though citizens were not involved in all phases in any of the projects considered, and participation occurred in only one case (Pagliarino et al. 2020) across 8 out of 10 steps.

Moreover, compared with the projects described by Bonney and colleagues (2009), the research questions in these projects are often less specific and more driven by the practical problems within the context, with a clear objective of bringing about change. This aligns with the principles of action research and participatory research (Cuéllar-Padilla & Calle-Collado 2011; McIntyre 2008), where addressing real-world issues and involving stakeholders are essential elements.

Applying this classification to the pool of projects included in this study shows 5 contributory, 10 collaborative, and 19 co-created projects, as shown in Table 1 in Supplemental File 1: Appendix A.

Barriers and facilitators

Facilitators to project performance were identified in 20 papers, and 79 descriptors were used to code facilitator elements, which support the success of a participatory project. Those were then classified into 15 subthemes (see Table 2 in Supplemental File 1: Appendix A). Barriers to project performance were cited in 23 papers, and 63 descriptors were used to code barriers to the success of participatory projects. These were classified into 11 sub-themes (see Table 3 in Supplemental File 1: Appendix A). The 15 subthemes of facilitators and the 11 subthemes of barriers were both grouped into three overarching themes: a) technical, b) relational, and c) cognitive, as visually represented in Figure 3.

Figure 3

Influence of participation modes on projects

Participatory approaches in agricultural innovation are widespread globally, as reflected by the diversity of nations represented in the studies. Nevertheless, a significant heterogeneity is evident among the projects included in our review, encompassing foundational rationales, methodological references, and desired outcomes. One approach to explore this diversity could be to compare the project’s participation type with the other extracted elements.

Regarding the types of innovations, there were notable co-occurrences between the modes of participation and the innovations promoted. Projects focused on technical improvements generally used contributory and collaborative approaches, reflecting the specific knowledge required for production practices. In contrast, projects addressing complex issues like climate change or pesticide management often employed fully participatory approaches. Specifically, all projects within these themes involved participants in both defining research questions and designing the experimental processes.

When comparing the modes of participation with the theoretical frameworks of the projects analyzed, several clear distinctions emerged. Notably, projects using participatory action research (PAR) methodology did not include contributory approaches, reflecting PAR’s focus on active participant involvement (McIntyre 2008). Additionally, some theoretical frameworks did not explicitly use the term “participatory” but still incorporated participatory methods, such as collective decision-making for irrigation management (Krupnik et al. 2012; Valdivia et al. 2014).

Consistent with this rationale, barriers and facilitators were less frequently reported in contributory projects. When they were mentioned, they typically related to issues such as “resistance to technical change” or “lack of training.” In contrast, most barriers and facilitators identified in this study were associated with co-created projects. The analysis also highlights a significant difference between practical or behavioral barriers and cognitive or emotional/relational barriers. Practical barriers were more commonly reported, while emotional and relational factors were more frequently mentioned as facilitators.