Crops critically important for food security and nutrition such as roots, tubers and bananas, or fruits and vegetables, receive less attention in the trait prioritization literature. Prioritization of cereals in national policies reflects an emphasis on calories as the primary driver of food security, which is frequently paired with a general lack of attention to nutrition sensitivity within agricultural programs [24–27]. The few studies centered on crops that offer dietary diversity, such as fruits and vegetables, are not focused on serving populations in Asia, sub-Saharan Africa, or South America. This has consequences for vulnerable populations at risk of malnutrition and micro-nutrient deficiencies, especially during food crises. Recent findings claim that effects of food inflation and recent food shortages increase the risk of stunting, decrease diet quality and wasting for 1.2 million children in 44 low and middle – income countries [28]. A lack of trait data from nutritious crops in these settings presents a clear priority for more research to fill this gap.
The lack of research robustness in study design and taxonomy of trait preference limits the usefulness of trait prioritization data for crop breeding. It was nearly impossible to aggregate trait preferences across studies without a standardized taxonomy of traits. Standard crop breeding management databases (e.g., https://breedbase.org/) and increasingly well-developed crop ontology definitions [29] should be linked more closely to trait preference studies to enable higher cross comparability.
Taxonomies can have far-reaching impact beyond the plant science communities. Initiatives like the recently announced Vision for Adapted Crops and Soils demonstrate the government and international agencies response to identify the role of crop breeding to support agriculture’s response to climate change, especially for countries with the poorest and hungriest populations [30]. Without better methods to help guide research, development, investment and prioritization, crop breeding will not be able to deliver on its promise to support the world with healthy, nutritious, and sustainable crops.
There is very little research on trait prioritization being conducted in South America, despite a rising interest in crop breeding in the region [31]. This regional bias may reflect a choice to exclude non-English language materials. However, similar results are reported in other recent systematic scoping reviews in agriculture, and where Spanish-language publications were included [32–33]. This gap is also at odds with the centrality of participatory trait preference research for large public sector crop breeding programs in the region, such the International Potato Center [34]. Regardless of the cause, a lack of available, robust research from this part of the world may have disastrous consequences given climate change and South America’s importance to global commodities trade [35–36]. Whether the regional bias in trait prioritization studies is driven by policy and donor priorities requires further analysis. However, the network mapping offers a glimpse into differences in donor priorities for groups of crops, as well as changes in the roles of different types of donors (universities vs foundations) over time. This questions on the types of studies supported, and how these replicate dominant donor priorities [37].
Studies with robust designs, where clear research questions are accompanied by good data collection, methodological and analytical plans, are rare [38]. For example, robust sample size calculation is critical, with easily available strategies to do so [see 39]. Few papers discussed respondents’ selection and sample construction, raising serious questions on representativeness and external validity of the results they present. Sample size calculations were often overlooked by the authors, and may indicate sampling guided by habit and practicality, rather than robust study design. A lack of attention to respondent sampling in trait prioritization studies could be producing misleading and unrepresentative results for crop breeders.
Understanding how and why gender shapes trait preferences is critical for breeding programs to develop new varieties inclusively and equitably [40–41]. Gender analysis minimally requires sex-disaggregated trait prioritization data. With this data, breeders remain uninformed on sex-disaggregated trait preferences, and may over-generalize for communities based on incomplete trait preference input [6, 13]. In addition, recent literature documents how intersecting social identities and roles in crop value chains shape trait preferences [6, 13–14, 42–43]. Most studies do not report socio-economic or sex-disaggregated data on trait prioritization study respondents, though studies that do are more recently published studies [see 44–45], potentially pointing to a shift in the change in practice in trait prioritization studies. Future studies should at least sex-disaggregated data at collection, and in analysis.
Robust study design includes appropriate data collection tools, and analysis methods. Direct question-based user preference elicitation is most common in the studies we review. This could be problematic, as direct elicitation of user preferences has been shown to produce questionable results [46]. Another problematic trend we observe is the frequent use of qualitative methods (such as focus group discussions) in data collection, with very rare use of qualitative methods of analysis. Skills to collect and analyze high quality qualitative data are exceedingly rare in public sector breeding programs [47], which may explain this observation. Some of the more commonly used tools and analysis methods may also stem from the siloed research networks our analysis uncovered. Leading academic researchers playing key roles in networks may cause such siloing [48].
This scoping review captures a rich history of participatory plant breeding [11, for specific case studies see 13, 49–51]. The results show wide variability across participatory trait prioritization studies, due in part to authors conflating ‘participation’ with ‘participatory’, engaging end-users just in a consultative form. Formal-led and consultative participatory breeding is more common in public sector breeding [23]. Our findings confirm this trend, with only a small number of studies reporting trait prioritization data from farmers testing varieties in their own fields and with their own management practices. Studies that use on-farm testing (direct experience) to capture trait prioritization partially circumvent biases that arise from direct questions.