A scoping review on tools and methods for trait prioritization in crop breeding programmes

Public-sector crop breeding programmes are increasingly demand-led and data-driven, requiring clear understanding of trait preferences among heterogeneous stakeholders across value chains to inform target product profiles. The term ‘trait’ is defined broadly as observable crop characteristics identifiable by stakeholders. Trait prioritization translates diverse attribute preferences into actionable breeding decisions, which is critical for improving adoption, particularly among marginalized smallholder farmers. Despite the central role of trait prioritization, the underlying tools and methods are seldom examined. Historically, studies have relied on direct elicitation (ranking, choice experiments) and participatory varietal selection, within inherently multidisciplinary settings that complicate design and implementation. Prior evidence suggests systematic socio-economic biases in trait preference research (for example, low rates of sex-disaggregated reporting) and gaps in stakeholder representation. The authors hypothesize that limited attention to rigorous methods and tools hinders unbiased, representative data and ultimately impedes data-driven breeding decisions. The aim is to conduct a comprehensive scoping review of tools and methods for trait prioritization to inform robust research design for crop breeding programmes.

Prior work indicates socio-economic and gender-related biases in trait preference studies, with only about a quarter of studies historically reporting sex-disaggregated data despite evidence that gender, poverty and food security shape preferences. A scoping review focused on rice identified gaps in stakeholder representation and heterogeneity in preferences, but did not assess robustness of tools and methods. Participatory plant breeding has generated substantial trait information across contexts, yet the literature shows inconsistent definitions and taxonomies of traits and varied, often insufficient, methodological rigor in design, sampling and analysis.

Design: Systematic scoping review following PRISMA-SCR guidelines. Protocol registered 03/11/2022, updated 07/08/2022 (OSF: https://osf.io/ayw8q). Searches updated 06/23/2023. Research question: What tools and methods have been used for trait prioritization in plant breeding programmes across time, locations, crop groups, institutions and genders? Eligibility: Original research or reviews explicitly addressing trait prioritization related to crops, varieties, seeds, planting materials or germplasm within plant breeding, seed systems, or agronomy. No date or geographic restrictions; English-only. Information sources: Five scholarly databases (Scopus, Web of Science, CAB Direct, AgEcon Search, BIOSIS) and 12 grey literature sources (including CSIRO, Gardian, IFAD, JPAL/ATAI, ODI, DFID, World Bank, WHO, UNEP, WFP, FAO, AgriLinks). Search strategy: Combined population and topic terms; example Web of Science strategy used Boolean operators across stakeholder, trait preference/prioritization/selection/adoption, and agriculture/breeding domains (see paper for full strings). Grey sources searched manually with keywords. Screening and selection: 17,786 records identified; 5,808 duplicates removed; 11,978 titles/abstracts screened; 1,325 full texts assessed; 657 studies included. Exclusions included studies where trait prioritization was peripheral, lacked end-user engagement, outside scope (non-food crops or non-breeding/agronomy), unavailable full text, non-English, or focused on processing traits. Data extraction: Peer-reviewed title/abstract/full-text screening; a 40-question extraction framework pre-tested and applied. Extracted variables included tools and methods for data collection and analysis, crop group, geography, disaggregation, degree of participation, donors, institutions, and ranking. Synthesis and analysis: Descriptive synthesis in R (v4.2.3). Networks of authors/affiliations/donors mapped with VOSviewer. Crop groups aggregated into cereals, legumes, RTB (roots, tubers, bananas), and vegetables and fruits; commodity/ornamental crops (3%) excluded from synthesis. Studies could be classified in multiple categories for tools/methods when applicable.

- Corpus and distribution: 657 studies included; >50% published since 2016; 83% peer-reviewed, 17% grey literature. Geographic coverage uneven: 57% in sub-Saharan Africa; South America only 6%; Middle East and North Africa 2%. Ethiopia and India had the most studies, followed by Nigeria, Ghana, USA and Uganda. RTB, cereals and legumes were studied mainly in South Asia, sub-Saharan Africa and South America (96%), while fruits and vegetables were studied predominantly (59%) in North America and Europe. - Data collection tools: 58% elicited preferences via direct questions (surveys, focus group discussions). Direct experience tools (24%) included participatory varietal selection (PVS) and demonstration plots. About 10% used choice experiments and sensory evaluations. Over one-third used only one tool (surveys 33%, PVS 14%, FGDs 10%); about one-third used two tools (often surveys + FGDs or surveys + PVS). Use of multiple tools has increased over time. Peer-reviewed studies had a median of two tools versus one in grey literature. - Analytical approaches: Frequency counts and hypothesis testing were most common; followed by economic modelling, multivariate analyses and trait ranking. Only 5% used qualitative data analysis despite frequent use of FGDs. Mixed-methods usage was limited (descriptive + hypothesis testing: 13%; descriptive + economic modelling: 6%). - Participation and leadership: Cereal studies were more participatory (54%) than legumes (51%) and RTB (41%); participatory research was uncommon for vegetables and fruits (14%). Reporting on sampling was often weak: 12% did not report respondent counts; 52% engaged only farmers (without clarifying consumer/producer roles); fewer than 19% engaged consumers, traders or seed system actors. A de facto norm of ~250 respondents per study was observed irrespective of context or method. Of studies self-reporting as participatory, 90% were classified as formal-led; only 12.5% involved formal-led collaborative on-farm testing under farmers’ own management. - Gender and social disaggregation: Only 37% collected sex-disaggregated data; more common in RTB, legumes, and vegetables/fruits than cereals. Among sex-disaggregated studies (n≈216), the median share of women respondents was 49.5% (IQR 32–57%). Only 1.7% collected intra-household preference data. Among sex-disaggregated studies, 84% found sex-based differences; 70% explained differences via literature (40%), additional qualitative (31%) or quantitative (25%) methods. Other disaggregation was rare: 37% by state/region; 3% by both region and gender; 0.9% by age. - Trait taxonomy and synthesis: High yield remained important but showed heterogeneity by crop group (less emphasized in RTB and vegetables/fruits). Early maturity, pest/disease resistance and drought tolerance gained importance in cereals; taste ranked high in RTB; pest/disease resistance ranked high in vegetables/fruits. However, substantial heterogeneity in trait descriptions and lack of standardized taxonomies impeded cross-study synthesis; even similar rice studies in India across consecutive years produced divergent trait lists. - Networks and funding: Co-authorship mapped into ~10 isolated clusters with limited cross-cluster collaboration. Donor funding was highly centralized, with five donors supporting nearly half of studies, indicating research and funding silos.

The review reveals pronounced imbalances in trait prioritization research: cereals dominate over nutrition-sensitive crops such as RTB, fruits and vegetables, with potential consequences for dietary diversity and micronutrient security, especially under food crises. Regional gaps—particularly in South America and the Middle East/North Africa—persist despite growing breeding interest, potentially reflecting language restrictions and donor/policy priorities. Methodological weaknesses limit the utility of findings for breeding: inconsistent or unreported sampling and sample size calculations challenge representativeness and external validity; reliance on direct elicitation can bias stated preferences; and frequent use of qualitative collection tools contrasts with rare application of qualitative analytical methods. Heterogeneous trait definitions and absent taxonomies thwart aggregation and data-driven prioritization at scale, undermining links to breeding databases and crop ontologies. Limited sex-disaggregation and scant consideration of intersecting social identities risk masking distinct stakeholder preferences. Siloed author and donor networks may entrench method choices and constrain innovation. More robust participatory designs—particularly on-farm testing and novel approaches like video-based concept testing—can mitigate biases and better align varietal design with end-user demand.

Substantial evidence gaps limit demand-driven, evidence-based breeding decisions. Key recommendations: (1) Prioritize research on nutritious crops (RTB, fruits, vegetables) across sub-Saharan Africa, Asia and South America, and address the paucity of studies in the Middle East and North Africa. (2) Link trait preference studies to breeding management databases and standardized crop ontology definitions to enable cross-study comparability and synthesis. (3) Systematically collect and analyze sex-disaggregated data to inform gender-responsive breeding. (4) Implement rigorous study design, including representative sampling frames, power/sample size calculations and appropriate choice of data collection and analytical methods, while expanding participatory, experience-based tools to reduce elicitation bias.

- Language restriction to English-language studies may contribute to regional bias (e.g., underrepresentation of South America and MENA). - Heterogeneity and lack of standardized trait taxonomies across studies impeded aggregation and cross-comparison of findings. - Reporting gaps in sampling strategies and sample sizes limit assessment of representativeness and external validity. - Reliance on authors’ self-reported participation levels to classify studies may introduce classification bias. - Commodity and ornamental crops (3%) were excluded from synthesis, which may omit relevant methods from those areas.