Crop breeding is a discipline grounded in foresight of future needs of growers, processors and consumers. Setting research and development priorities ex-ante is therefore a necessity to succeed in meeting these demands. We examined the literature around ex-ante research priority setting approaches to situate our work and identify gaps and opportunities for methodological development. To date, different approaches have been developed to support priority setting in international agricultural research programs (Wiebe et al., 2021; Alston et al., 2022). Applications span from CGIAR-level (case studies can be found in Raitzer and Kelley, 2008), to national research prioritization (e.g., EMBRAPA in Brazil as described by Avila et al., 2002). Recently, the use of international agricultural research prioritization exercises has been more sporadic, with less data intensive procedures preferred (Thornton et al., 2018). These alternative methods range from simple qualitative scoring exercises to highly complex simulation models estimating the functional relationship between inputs (research investments) and agricultural outputs while accounting for the underlying uncertainty (see Braunschweig (2000), for an overview of different priority setting methods).

We summarize existing ex-ante priority setting frameworks (Supplementary Table A1) and characterize them along three dimensions of scale of analysis, translation of benefits into dollar values, and sex disaggregation. These three dimensions represent junctions at which ex ante frameworks that are both locally relevant and gender responsive distinguish themselves from more traditional approaches. Most existing frameworks focus on benefits in economic terms, and rarely using sex-disaggregated data (Supplementary Table A1). Country-level assessments are preferred, even though frameworks allowing a flexible scale of analysis (i.e., national and regional) are common.

Guided by an interest to develop a methodology that could be more participatory and engaging for respondents, we further looked at frameworks for their level of participation and engagement with stakeholders. Participation describes the degree at which studies involve a variety of different stakeholders, beyond scientists. Engagement exemplifies the frequency at which stakeholders are involved and if feedback and validation mechanisms are put in place. Studies in the early 1990s guided by economic surplus theory and cost benefit analysis are less participatory and engaging (for a review see Braunschweig, 2000). Recent modeling approaches with heterogeneous agents are also not participatory or engaging (e.g., Endresen et al., 2011; Groot et al., 2012 in Supplementary Table A1). Participatory varietal selection (PVS)-like approaches where alternative research options are validated by non-academic stakeholders (mostly farmers) are more participatory, but the engagement of stakeholders occurs only as an a-posteriori consultation with no feedback loop, where priorities from stakeholders are then validated jointly with scientists (e.g., Randolph et al., 2001; Pemsl et al., 2022 in Supplementary Table A1). Crowdsourcing plant breeding methods (e.g., Steinke and van Etten, 2017 in Supplementary Table A1) engage stakeholders in an iterative manner with easy to implement approaches but they include primarily farmers. A rare example of a highly participatory and engaging framework was Blundo‐Canto et al. (2020) in Supplementary Table A1, but also highly complex and abstract.

Methodologies currently available for ex-ante research priority setting encompass simple interactive scoring exercises (e.g., participatory ranking scenarios) to complex simulation models estimating the functional relationship between inputs and outputs (e.g., agent-based models; Supplementary Table A1). Allocating research efficiency and selecting the most promising research activities are issues directly tied to the scarcity of resources for plant breeding in development. Therefore, most existing frameworks place an emphasis on economic efficiency and on costs and benefits that can be expressed in monetary values (Braunschweig, 2000). The economic surplus analysis and the cost–benefit analysis are the tools most frequently utilized. Despite being easy to interpret, these two techniques present a few methodological disadvantages. First, they rarely include non-quantifiable and non-marketable outcomes (e.g., the shadow price of gender-related benefits). Second, in these tools, agricultural researchers provide most of the input, and active participation of stakeholders is quite limited. Third, these tools offer a static representation of the commodity market and thus tend to underestimate longitudinal non-linear dynamics which can affect the breeding process (Petsakos et al., 2018). For example, although some implementations allow for an explicit representation of dynamics in production and consumption (HarvestChoice, 1995), economic surplus models able to analyze well-structured foresight scenarios (like those proposed by the Intergovernmental Panel on Climate Change (IPCC) Davis et al., 1987), are frequently complex and data intensive. Lastly, this economic surplus paradigm has raised concerns because externalities, distributional effects, and longer-term impacts all tend to be neglected with a narrow focus on breeding costs and benefits (Dahlberg, 1988).

To develop PEEP, we took inspiration from the systematic, quantitative ex-ante priority assessment undertaken by the CGIAR Research Program on Roots, Tubers and Bananas (RTB) in the period 2012–2014 (Pemsl et al., 2022). PEEP preserves the systematic and quantitative nature of this assessment, while integrating attention to gender and adapting for utility for national level and crop-specific breeding projects. PEEP complements the literature on methods for ex-ante priority assessment in crop breeding through three methodological advancements. First, PEEP is adaptable at the national level, building on the existing work (Mills and Karanja, 1997; Randolph et al., 2001; Singh et al., 2020; Pemsl et al., 2022). Unlike these studies, PEEP leverages the national focus to better consider the relevance of gendered preferences and access to resources, which are context dependent. Second, PEEP is designed to be consultative in all phases, resembling the frameworks of Pemsl et al. (2022) and Steinke and Van Etten (2017). PEEP includes an array of stakeholders beyond scientists, so that crop improvement priorities are set and validated by social and natural scientists as well as practitioners. Here we define practitioners as local actors directly engaged in agricultural production, processing, and marketing and representatives from, e.g., seed companies, agricultural cooperatives, agricultural women’s groups, and national or regional policy makers. PEEP is purposively combining multiple participatory tools to ensure that each category of stakeholders is involved through an approach which maximizes participation and engagement. Third, PEEP considers gender dynamics³ by seeking gender equity in the evaluation of alternative crop improvement priorities, and product concepts. To the best of our knowledge, no existing ex-ante framework thus far deliberately includes gender as a lens of analysis in prioritization.

Each of these three methodological innovations of PEEP has challenges. For example, PEEP requires a heterogenous group of stakeholders with deep contextual knowledge and local impact pathways. This may sometimes lead to conflict between national and local priorities. Establishing a process to reconcile these conflicts is a key component of the PEEP framework. Furthermore, it is particularly challenging to capture the views of the most vulnerable, but often least accessible, populations. PEEP includes nationally representative stakeholders who have an overview of the relative importance of different opportunities and barriers faced by these populations, but a risk remains that the needs of the most vulnerable respondents may still be missed. Finally, integrating a diversity of stakeholders and gender analysis requires additional expertise and creates a further level of complexity. Trade-offs hinge on who might lose or benefit from new varieties.

We leverage the use of heterodox methodological approaches and a strong emphasis on the engagement of knowledge-rich stakeholders to circumvent these challenges. PEEP ranks priorities based on a heterogenous set of benefits without quantifying impacts in terms of economic return (or net present value). Furthermore, in contrast with other scientific domains (e.g., health and medicine) where priority setting and ex-ante analysis are well-established practices and where engagement is actuated through the Delphi method (Linstone and Turoff, 1975), we utilize a less common iterative and tailored multi-stakeholder approach. Built on the principle of engaging technical as well as practice knowledge, PEEP adapts a diverse tool of engagement to each category of knowledge-rich stakeholders. This avoids multiple, highly resourced, and time intensive, rounds of consultation, especially where participants are not familiar with repetitive research-oriented routines.

How is PEEP any different from existing participatory plant breeding efforts? PEEP is interested in the “why” more than in the “what.” Linking crop improvement impacts to product concepts, PEEP generates ex-ante impact pathways. These support breeders in justifying the impact of a future variety when the variety is still simply a product concept. Having clarity on which product concept targets the set of crop improvement priorities considered relevant by a heterogenous group of stakeholders assists breeders in (i) justifying the impact of new varieties under development; (ii) ensuring that product concepts target priorities and impacts are equitable and fair (iii) justifying the investment made by national and international funders; and, finally, assists breeders in (iv) better marketing of new varieties.

In the first phase, the breeding program implementing PEEP must form a core implementing team (hereafter core team). The core team oversees the process development, analyzes the results, and compiles the final assessment report. A set of competencies underlie the selection of individuals for the core team (Box 1) to ensure that the right balance of skills and experience are represented on the team are driving this process. Once assembled, the core team decides the geographical scope of the application (national vs. regional) as well as the target crop. PEEP can accommodate evaluations at the country level or regional level on any crop product concept to be evaluated.

Assembling a team of stakeholders to engage in the PEEP framework is the second - and possibly most critical - phase of the PEEP methodology. According to their degree of familiarity with the formal (academic) research process, stakeholders are assigned to two groups: (i) group R (Research), and (ii) group P (Practice). The Research group involves any member of a cross-functional breeding team who has technical knowledge of the target crop (see Box 2). Complementarily, the Practice group includes all stakeholders involved into the targeted crop value chain and possess practical knowledge (see Box 2). Engaging the R and P groups with transparency is important, clearly informing each group member about time and resources required to attend the process. For group R, incentives to participate in PEEP might be: (i) contribution to better define national or regional breeding objectives, (ii) the possibility to prioritize gender and climate in the national breeding agenda, (iii) better allocation of resources to crop breeding programs, (iv) possibility to conduct an evaluation which attracts international donors and funding agencies, and (v) networking and round table opportunity for new projects. For group P, incentives might be: (i) steering the breeding work to account for their needs and preferences, (ii) contribute to the development of varieties which are better suited for production and selling purposes, (iii) build social networks with researchers, (iv) tighter links with the local research community, which yield learning opportunities and higher engagement.

Selecting members for the Research group should include consideration of their expertise on the chosen crop, whether they are young researchers or senior leaders, as well as if they are regional or national collaborators working at in-country international centers.

For members of group P, choosing representatives at the national or regional level may be less straightforward. The core team needs to effectively sample a sub-population of producers, traders, processors, consumers, formal and informal agricultural cooperatives, and women’s groups. Both at the national and regional level, NARs or CGIAR centers should utilize an informed stratified sampling strategy. If a list of crop producers, processors or traders for each area is available, we highly encourage the core team to use this information to calculate sampling weights

The main result of phase 3 is the prioritization of impacts from crop improvement impacts and how they map to a set of given product concepts. This helps breeders to align their prioritization of product concepts to meet impacts. During this phase, breeders, geneticists, agronomists, pathologists, and other members of the breeding team are responsible for integrating crop improvement impacts. Concretely this means answering two main questions: are existing product concepts able to address the impacts? If not, what information is missing?

A tension might emerge from this validation: on the one hand, the breeding team needs to work with product concepts that are realistic given the technical and institutional capabilities of the NAR or the CGIAR center. On the other hand, the team needs to ensure that previous efforts do not get lost in the name of “breedability” of alternatives.

We encourage the ex-ante team to repeat the analysis at every new breeding cycle. If the team perceives that a change in priority might have occurred earlier–due to a new agricultural policy, a climate event, or an expected shock like the Covid-19 pandemic–the ex-ante exercise can be repeated prior to the closing of the breeding cycle.