학술

Farmers’ trait preferences and its implications for demand-driven yam breeding in Uganda

PLOS ONE
CC BY
이 매체는 공공·자유 라이선스로 본문을 직접 표시합니다.

Figures
Abstract
Yam is a critical staple and income source in tropical regions, such as Uganda. In Uganda, data on yam production systems, farmer typologies, and trait preferences remain scarce. This knowledge gap hampers the development and adoption of improved varieties because breeders lack clear targets aligned with farmers’ needs. To address this gap, this study characterized yam farming systems and farmer typologies in Uganda and identified farmers’ priority traits to provide demand-side insights to guide breeding programs toward developing end-user-preferred, locally adapted varieties. A participatory survey of 540 smallholder yam farmers across six districts (covering three agro-ecological zones) captured data on their production practices, constraints, and trait preferences. Trait priorities were elicited using a 1000minds conjoint analysis of the Potentially All Pairwise Ranking of All Possible Alternatives (PAPRIKA) algorithm, which presents farmers with pairwise trait trade-offs to rank their preferences. White yam was the predominant species grown. The main production constraints included a lack of quality planting materials (~40% of the respondents) and limited access to improved varieties (~26%). Farmers selected high yield as their top trait priority (73% of respondents), followed by tuber mealiness (42%), canopy vigor (29%), disease resistance (26%), and long shelf life (25%). The conjoint analysis confirmed strong preferences for high dry matter content (contributing to mealiness), shorter time to maturity, extended storability (long shelf-life), high total crop value, attractive tuber flesh color, and high fresh yield. These findings provide valuable insights into yam utilization, production constraints, and farmers’ trait preferences, which can inform yam improvement. Breeding programs should integrate farmers’ preferred traits and address seed system constraints to ensure the adoption of new varieties.
Citation: Adjei EA, Angudubo S, Odong TL, Esuma W, Chamba EB, Ochwo-Ssemakula M, et al. (2026) Farmers’ trait preferences and its implications for demand-driven yam breeding in Uganda. PLoS One 21(11): e0362574. https://doi.org/10.1371/journal.pone.0362574
Editor: Mehdi Rahimi, KGUT: Graduate University of Advanced Technology, IRAN, ISLAMIC REPUBLIC OF
Received: July 1, 2025; Accepted: October 27, 2026; Published: November 20, 2026
Copyright: © 2026 Adjei et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: A repository for the datasets and analysis scripts has been created, this be accessed through the DOI: 10.6084/m9.figshare.31429544.
Funding: Since the corresponding author is affiliated to the Alliance of Bioversity International and CIAT, the organization is willing to pay the publication fee upon PLOS ONE sharing an invoice. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors declare that they have no competing interests and received no external funding for this study.
Introduction
Yam (Dioscorea species), a monocotyledonous crop belonging to the genus Dioscorea in the family Dioscoreaceae [1]. It is one of the most important root and tuber crops in the tropics and subtropics, where it contributes significantly to food, nutrition, and income security [2–4]. Globally, yam production exceeds 60 million tons, cultivated on approximately 5 million hectares, marking an 80% increase in the cultivated area since 1961 [5]. West Africa accounts for more than 90% of global production, with Nigeria, Ghana, Benin, and Côte d’Ivoire being the leading producers [6]. The crop’s cultural and economic significance in this region has driven sustained research investment and the establishment of yam breeding programs. In contrast, yam remains a minor and under-researched crop in East Africa, particularly in Uganda.
Yam cultivation in Uganda is widespread across the Central, Northern, and Eastern regions, supported by the country’s bimodal rainfall pattern [7–9]. The crop fits well into smallholder farming systems as a food security buffer and a source of income, particularly during the lean season. The most cited estimates suggest that Uganda produces approximately 220,000 metric tons of yam on 18,000 hectares (ha), with an average yield of 5–8 t/ha, far below the potential of 20–30 t/ha [7–9]. However, production is constrained by several factors, including limited access to high-quality planting materials, lack of improved varieties, pest and disease pressures, and reliance on traditional agronomic practices [8,10,11]. The lack of formal breeding investment reflects low prioritization of yam within the national research agenda.
Recent efforts by the National Agricultural Research Organization (NARO) have started to revive interest in yam through the introduction and evaluation of breeding materials from West Africa [12]. These early efforts have created a foundation for local breeding [13] but must be complemented by a clear understanding of the production environment, farmer typologies, and user-preferred traits to ensure the relevance and adoption of future varieties. The demand-led breeding (DLB) framework provides a systematic approach to link breeding objectives with the needs of end users across the value chain [14].
Previous studies on yam trait preferences have been conducted primarily in West Africa, where yam production and consumption are deeply embedded in culture and trade. For example, research in Nigeria, Ghana, and Benin has identified key farmer-preferred traits such as high yield, tuber size, tuber shape, ease of peeling, storability, and resistance to pests and diseases [15]. While these findings provide useful insights, they cannot be directly generalized to Uganda. Production environments, species composition, and consumer markets in East Africa differ substantially from those in West Africa. Moreover, Uganda’s yam farmers operate under unique socio-economic conditions characterized by small landholdings, low input use, limited access to markets, and high varietal diversity from local landraces. Yet, to date, no study has comprehensively documented yam production systems, farmer typologies, or end users’ trait preferences in Uganda or the wider East African region.
Addressing this gap is therefore a prerequisite for establishing a functional yam breeding pipeline in Uganda that can deliver varieties with high adoption potential and market relevance. Without a clear understanding of what farmers value most, whether it is early maturity, high yield, tuber appearance, or cooking qualities, breeding efforts risk producing varieties that are better performing compared to existing varieties but socially irrelevant, not preferred by the end users.
This study applies the demand-led breeding framework to generate new empirical evidence on yam production systems and trait preferences in Uganda. Specifically, it aims to (i) characterize yam production systems across major agro-ecological zones, (ii) identify farmer typologies and their distinguishing features, and (iii) elicit and prioritize the traits that farmers value most in new yam varieties. By combining structured surveys with participatory trait prioritization methods, the study captures the diversity of end-user perspectives and links them to breeding priorities. This integrated approach ensures that the resulting data are both quantitative and context-specific, providing a strong foundation for subsequent trait discovery and selection processes [16].
The findings are expected to guide the formulation of Uganda’s yam product profiles and influence the prioritization of traits in ongoing breeding programs. By integrating farmer perspectives at the trait discovery stage (product design stage), the study helps ensure that future yam varieties align with both production realities and market preferences. This research also provides a reference for similar efforts in other underutilized crops where participatory and demand-led breeding approaches are still emerging. Ultimately, the study contributes to repositioning yam as a strategic crop for Uganda’s food, nutrition, and income security agenda, supporting broader goals of agricultural diversification and climate resilience within the agrifood system.
Materials and methods
Description of the study areas
The study was conducted in three agro-ecological zones in Uganda: Northwestern savannah grasslands, Lake Victoria Crescent, and the Eastern highland ranges, as presented in Fig 1. These regions span 85,391.7 Km2, 61,403.2 Km2, and 39,478.8 Km2, respectively [18]. The zones experience bimodal rainfall with interspersed dry spells and variations in temperature and soil conditions. In these zones, yams are primarily cultivated for subsistence purposes and are commonly intercropped with banana and tree crops. As a result, yam cultivation is distributed across various districts within the selected agro-ecological zones. The selection of the study districts was based on the importance of the crop to farmers’ livelihoods, existing cropping patterns, and the diversity of yam varieties.
Sampling selection and sample size
A scoping study on yam production systems and disease prevalence informed the purposive selection of two districts per agro-ecological zone: Adjumani and Arua (Northwestern), Kayunga and Masaka (Lake Victoria Crescent), and Mbale and Tororo (Eastern highland). A proportionate sample size was calculated using Cochran’s formula (Equation 1) for categorical data [19], assuming a 5% margin of error, a 95% confidence interval, and an estimated population proportion (p) of 0.35. According to de Bekker-Grob et al [20], this approach is feasible for determining the optimal sample size in choice analysis studies if the initial parameter values are available. However, preliminary research on a small sample to obtain the initial parameter values was not possible in our study, owing to practical constraints. The t-value (t) at an alpha level of 0.05 (0.025 in a two-tailed test) for a sample size greater than 120 was determined to be 1.96. The variance estimate in Equation 1 is represented by the product of p and q, where p represents the highest potential sample size based on the proportion of yam-farming households in the district at 0.35.
(1)Where;
- (p)(q) in this study = (maximum possible proportion (0.35) * (1-maximum possible proportion) = 0.2275
- d accounts for the tolerable margin of error for the estimated proportion (0.05), which represents the error level that the study was willing to accept.
The returned sample size () did not exceed 5% of the number of households; therefore, this study used the returned sample size to collect the data. However, the response rate in household surveys is generally below 100% [19] because of partial responses, incorrect responses, and data quality criteria of the research procedure. This study employed an oversampling approach to eliminate measurement errors. Based on the success of the planned face-to-face interviews, this study projected a response rate of approximately 65%.
(Approximately 540 households were included in the survey).
The 540 targeted households were distributed proportionately across districts. The respondents were selected purposively within each district based on their involvement in yam cultivation.
Ethical considerations
Ethical clearance for field activities was obtained from the internal Research Ethics Committee (REC) of the National Agricultural Research Organization (NARO), Uganda’s apex body for public-funded agricultural research. The NARO REC operates as a semi-autonomous committee under the oversight of the Uganda National Council for Science and Technology (UNCST). The REC provided approval through a formal confirmation letter signed by the Director of Research (DoR) at the National Crops Resources Research Institute (NaCRRI), who serves as a signatory under the NARO REC. Additional administrative permissions were obtained from the relevant district agricultural offices to facilitate access to farmer groups in the selected sub-counties and parishes. These permissions were mainly granted through verbal approval and follow-up phone communication with the Agricultural Extension Officers responsible for agricultural production and marketing in the study areas.
Socio-economic data collection procedure
A desk review and key informant interviews (KIIs) with farmers, extension agents, and a multidisciplinary yam breeding team informed the development of a structured questionnaire (S2 Text). The insights obtained from these interviews were crucial for developing a well-structured questionnaire. Following peer review and pretesting, the questionnaire was digitized using Alchemer, an online enterprise survey software (<http://www.surveygizmo.com/terms/>), and administered through the same platform (S2 Text).
Trait preference trade-off tool
Drawing from the yam-breeding community of practice, 11 traits relevant to production, processing, and consumer use were selected (Table 1). The cross-functional research team identified trait levels relevant to the breeding objectives.
Each trait was defined with a measurable unit and change level for use in a multi-criteria decision-making (MCDM) framework. The 1000minds software (https://www.1000Minds.com) was used to operationalize the Potentially All Pairwise Rankings of all Possible Alternatives (PAPRIKA) algorithm. The software generates trait trade-off questions. Farmers compared pairs of alternatives (Fig 2) in a structured preference elicitation process. Holistic trade-offs between alternatives offer valuable information for market-oriented breeding programs.
Data analysis
Data were analyzed using Stata v.16 [21] and R statistical software [22]. Descriptive statistics were used for the socio-economic variables. The Kruskal-Wallis test was used to assess trait ranking differences across groups [23]. Chi-square tests were used to test associations between categorical variables and farmer typologies, whereas correlation analysis among traits was performed using the “corr” function in R. Trait preference clusters were derived via k-means clustering using the HCPC procedure in the “FactoMineR” package. Bar plots with error bars were generated using “ggplot2” in R statistical software [22].
Results
Demographic characteristics of yam farmers
Figs 3 and 4 show the percentage distribution of the demographics of the farmers used for the study. Chi-square tests for gender, marital status, education level, and age group revealed statistically significant differences in these variables across agro-ecological zones. The proportion of men (50.2%) was slightly higher than that of women, with variations in the gender distribution observed in the agro-ecological zones (Fig 3A). The highest age categories ranged from 31 to 64 years across agroecologies (Fig 3B). Most farmers had low education levels, with the majority (59.1%) having attained only primary education (Fig 4A) and being married (Fig 4B). Additionally, most farmers (71.5%) were within the productive age group (31–64 years), which has important implications for their participation in decision-making processes related to yam production, consumption, and marketing.
L.C.V: Lake Victoria Crescent; E.H.R: Eastern Highland Range; N.S.G: Northwestern Savannah Grassland.
Socio-economic characteristics of yam farmers
The summary statistics of the socio-economic characteristics that affect the choice of the yam production system and trait preferences are displayed in Table 2. F-tests indicated statistically significant differences in the variables across the agro-ecological zones. On average, farmers had access to 0.91 ± 0.650 hectares, of which 0.76 ± 0.620 hectares were dedicated to crop production. Approximately 0.44 ± 0.320 hectares of land used for crop production were allocated for yam cultivation (Table 2). Regarding access to institutional services that enhance production and marketing, the average distance to markets and extension agents was 1.76 ± 1.149 km and 2.4 ± 1.608 km, respectively. However, access to better markets and extension services varies, regardless of the distance between these services.
Importance of yam production, farm-level performance and types of yams cultivated
There was no statistically significant relationship (ꭓ2 = 4.66, p = 0.32) between the gender of the farmers and the importance of yams (Table 3). However, 33.6% of the farmers identified yams as an important food security and income crop. The results in Table 4 indicate that farmers primarily cultivate four varieties of yams (white, yellow, purple, and aerial) across the agro-ecological zones. White (33.3%) and yellow (33.3%) yams were predominantly grown in farming households. In the Eastern highland range, the most cultivated types were aerial yams (48.9%) and white yams (31.1%), with adult men (57.8%) being the primary cultivators. In the Northwestern Savannah grasslands, most adult men cultivate white yams (35.7%), whereas adult women prefer to grow white and yellow yam cultivars.
The findings in Table 3 reveal distinct variations in yield performance related to gender and agro-ecological zone. The data demonstrate that high, medium, and low yield performance is prevalent across various agro-ecological zones, highlighting substantial variations in yield levels. The disparities in yield performance among agro-ecological zones by gender suggest dissimilarities in production systems, thereby emphasizing the importance of implementing suitable agronomic practices and seed systems. To enhance the yield performance of yams, it is crucial to consider and integrate the unique characteristics of crop systems across gender and agro-ecological zones.
The findings depicted in Figs 5–7 reveal that one-third (33.3%) of farmers cultivated both white and yellow yams across diverse agro-ecological zones. Adult women cultivate white, purple, and yellow yams in the Lake Victoria Crescent and Northern Savannah grasslands, whereas adult men primarily grow white and purple yams. Adult women and men cultivate white, yellow, and aerial yams in the Eastern highland ranges. These results suggest diverse preferences for yam types across the different agro-ecological zones. Furthermore, the difference in ranking of the importance of yams across gender categories indicates diverse preferences for crops in terms of production and marketing among farming households. Consequently, effective dissemination strategies require an understanding of the traits that influence the preferences and performance of distinct types of yams.
L.C.V. = Lake Victoria Crescent, E.H.R. Eastern Highland Range and N.S.G. = Northwestern Savannah Grassland.
L.C.V. = Lake Victoria Crescent, E.H.R. Eastern Highland Range and N.S.G. = Northwestern Savannah Grassland.
L.C.V. = Lake Victoria Crescent, E.H.R. Eastern Highland Range and N.S.G. = Northwestern Savannah Grassland.
Factors affecting yam production and marketing
The findings revealed in Table 4 suggest that yam production in Uganda is primarily hindered by a lack of access to planting materials (40.4%) and improved varieties (26.5%) across different agro-ecological zones. In addition to seed access, farmers in the Northwestern Savannah grasslands reported that pests and diseases (21.7%) posed significant production constraints. As yams are mainly produced for domestic consumption, marketing constraints are considered relatively subtle, except in the Lake Victoria zone where the crop is marketed. These findings emphasize the need to initiate yam seed systems in Uganda.
Farmers’ trait preferences by agro-ecological zones and Trait prioritization using 1000Minds
To evaluate preferences for yam traits, three groups of traits were presented to farmers: plant and storage (S1 Fig), tuber/yield (S2 Fig), and cooking/culinary (S3 Fig). The preferences for plant and storage traits varied significantly among farmers in different agro-ecological zones (ꭓ2 = 23.8; p = 0.05). Most farmers prefer yam cultivars with good canopy formation, disease resistance, and a long shelf life. In the Northern Savannah grassland, 41.1% of farmers highly preferred yam cultivars with a long shelf life, whereas in the Eastern highland range and Lake Victoria crescent, farmers’ choice of yam cultivars depended on good canopy formation and disease resistance (S1 Fig). Regarding preferences for yield and tuber characteristics, the majority (73.1%) of farmers preferred high-yielding varieties, with large tuber sizes and high dry matter content being the other important considerations (S2 Fig). Concerning cooking/culinary traits, most farmers prioritized mealiness, sweet taste, and short cooking time, with differences in preferences across agro-ecological zones (S3 Fig).
The overall ranking of the trait preferences of the 540 farmers, as determined through the 1000Minds choice experimentation survey design, followed the approach presented by Martin-Collado et al. (2015). This overall ranking provides a comprehensive perspective on preferences, regardless of the socio-economic characteristics of the respondents. The Kruskal–Wallis test was used to evaluate variance in trait preferences (Fig 8). The findings revealed a diverse range of preferences for yam traits, with most farmers prioritizing agronomic traits that affect yield, such as months to maturity, crop value, and yield. Although farmers seemed indifferent towards the preference for disease-resistant yams, this is due to limited knowledge and ability to identify the effect of YMV and YAD on yam production. Farmers also prioritized tuber quality traits, including flesh tuber color and shape (Fig 8). Concerning culinary preferences, farmers, who act as consumers, generally prioritize dry matter content, followed by shelf life and taste. In contrast, softness was the least preferred trait after cooking. These results emphasize the importance of acknowledging the heterogeneity of trait preference in yam improvement initiatives.
[NB: Friedman test, X2(10) = 1226.78, p < 0.0001, n = 540].
Cluster analysis
Cluster analysis facilitated the formation of homogeneous groups of respondents based on their responses to the selected variables. Principal component analysis indicated that all survey respondents, particularly those with similar trait preference rankings, were grouped (Fig 9). The key traits influencing the variation in preferences among groups include fresh tuber yield, cooked taste, maturity time, resistance to Y.M.V, dry matter, and flesh color. Significant variations in the preferences for these traits among respondents resulted in their categorization into distinct groups.
Plot separating respondents based on districts within agro-ecological zones.
Traits correlations
The heatmap in Fig 10 shows the correlations between the trait preferences and potential trait combinations. These findings suggest a moderate association between the traits, particularly between tuber yield (TFY) and total crop value (TVC). This suggests limited interdependency among the traits, implying that traits, whether considered independently or as a bundle, influence farmers’ preferences for yam varieties.
The Principal Component Analysis (PCA) results of the clusters represented typologies, which were assigned based on the pattern of the preferred traits: total crop value (n = 124), crop output (n = 101), product quality (n = 246), and plant survival (n = 25). The traits that significantly accounted for variation within each group exhibited significantly higher preference ranks, as shown in Table 5. Lower ranks indicated a higher preference compared to the ranks of the same traits in the other groups.
The results in Table 5 reveal a strong preference for the total crop value, with a mean of 2.9, compared to the overall mean of 3.3 within the crop value and output groups, respectively. A notable preference was also observed for dry matter, which had a mean of 6.0 against the overall mean of 6.5 in the product quality group. Yam Anthracnose Resistance (YAD) demonstrated a high preference, with a mean of 6.1 compared to the overall mean of 6.5 in the crop output group. Other traits showing high preference include tuber flesh, with means of 6.9 and 7.3 in both crop value and output groups, and softness after cooking, with means of 7.2 and 7.9 in the crop value group. Additionally, there was a significant preference for tuber shape, with means of 8.1 and 8.3 in the product quality group and plant survival group, which recorded means of 8.0 and 8.3.
The crop value group demonstrated a strong preference for total crop value, tuber flesh color, and softness after cooking, which did not significantly differ from the other groups in terms of total crop value. The crop output group exhibited notably high preferences for total crop value, YAD, and tuber flesh color. The respondents in the product quality group expressed a strong preference for dry matter and tuber shape. Meanwhile, the respondents in the plant survival group assigned significantly higher trait preferences for Yam Anthracnose resistance and tuber shape. The results, as presented in Table 5, indicated statistically significant differences (p ≤ 0.05) in total crop value, dry matter, tuber flesh color, softness after cooking, and tuber shape among the various respondent typologies. The associations between these typologies and the socio-economic characteristics are presented in Table 5.
Relationship between typologies and socio-economic characteristics
The results in Table 6, based on the approach by Martin-Collado et al. (2015a), show the association between typologies and socio-economic factors. We examined the typologies against various factor groups to determine which factors may account for the differences among the typologies. To identify the differences in frequency distribution by gender, age, education, land access, access to extension, access to market, land under yam, and agro-ecological zones, we used a chi-square test. The results indicate significant differences in the distribution of men and women across typologies, with men having more representation in the product quality group. Additionally, the analysis revealed differences in the distribution across age categories among the typologies, with a higher representation of the younger age group (18–30 years) in the crop output group, and the elderly (>65 years) in the product quality group.
We found differences in the distribution across education levels of the respondents, with a higher representation of those who attained primary and secondary education in the crop output group. Respondents who attained tertiary education were more represented in the product-quality group. Furthermore, the respondents with access to land of five acres or more had higher representation in the crop output group. Table 6 indicates that respondents who had access to farm-gate markets had a higher representation of the crop output group. Those who accessed neighboring markets showed high representation in both the crop output and product quality groups. Significant differences were observed in the distribution of land under yam production and agro-ecological zones across typologies. A greater proportion of respondents cultivating yams on five acres and above belonged to the crop output group. We also found that respondents in the central agro-ecological zone were more represented by the crop output group, followed by those in the eastern agro-ecological zone.
Discussion
Piloting 1000minds survey for profiling yam trait preferences
This study initiated the application of the 1000minds conjoint survey to ascertain yam trait preferences among smallholder farmers in Uganda. The PAPRIKA algorithm facilitated farmers in making significant trade-offs among features, offering a systematic and interactive method for prioritizing breeding objectives. This approach not only enumerated preferences but also illuminated how farmers assess intricate combinations of production, quality, and market-oriented characteristics [24]., It has been used to prioritize cassava and sweetpotato traits in efforts to improve demand-led crop breeding [25,26]. Critically, this study contributes new evidence that demand-led breeding in yam can be quantified using adaptive preference tools traditionally used in livestock and other crops.
Such tools have been successfully applied in cassava and sweetpotato (Balogun et al., 2022; Okello et al., 2022), but their application in yam is limited. Thus, this pilot introduces methodological innovation to a neglected crop and sets the stage for the integration of farmer-derived economic weights into yam breeding pipelines. Importantly, research shows that smallholder farmers with little digital proficiency can use sophisticated preference elicitation methods if the process is contextualized and effectively supported. The successful piloting of the 1000minds survey illustrates that smallholders with little experience of formal surveys may participate effectively in sophisticated preference elicitation when the process is well assisted.
This is consistent with data from other research that show that adaptive conjoint methods, such as 1000minds, may be used to extract breeding priorities from stakeholders, yielding quantitative trait weights and revealing preference heterogeneity [27]. In our study, our approach generated a data-driven ranking of yam features based on farmer values, providing benchmark information for demand-led breeding. Importantly, the study validates trends in participatory breeding. For instance, yam farmers in Nigeria consistently rank high yield and food quality (e.g., poundability or texture) among the most important traits [27]. Convergence across regions implies that the 1000minds survey can adequately capture relevant farmer preferences. Overall, our pilot study established that the preference-profiling method can substantially inform breeding programs by pinpointing what farmers want in new yam varieties in Uganda.
Adapting the 1000minds survey for yam smallholder farmers in Uganda
Data capture.
The implementation of the 1000minds preference survey with yam farmers in Uganda involved careful adaptation of data capture methods, trait selection, and trait definition. We employed a participatory data collection approach; rather than expecting farmers to complete the online survey independently, trained enumerators visited the villages with tablets and facilitated the survey in the local language. Face-to-face administration was essential because of limited internet access and varying literacy levels. Enumerators presented pairwise trait comparisons both verbally and visually, ensuring that the respondents understood each choice.
Trait selection.
The trait selection was customized through a consultative process. Due to the lack of established “industry standards” for yam characteristics in Uganda, the study team identified essential attributes by engaging with yam experts via key informant interviews and examining relevant literature on yams and other tuberous crops. This methodology guaranteed that the characteristics featured in the survey (such as fresh tuber output, maturation duration, culinary quality, and disease resistance) authentically represented the selection criteria of farmers. We narrowed the list to 11 traits to keep the survey cognitively feasible while covering the spectrum of production, processing, marketing, and consumption traits that are important in yam cultivation.
Trait level and unit definition.
Each trait was defined in terms that were easy for farmers to understand by using concrete units or levels. For instance, instead of abstract units, the yield was described as kilograms of tubers per acre, and maturity was framed as months from planting to harvest. Qualitative traits were given practical descriptions: shelf-life as “how long tubers can be stored after harvest without rotting.” Contextual definitions helped farmers evaluate the trade-offs. We piloted the survey with farmers to refine their wording and ensure that the questions were clear and culturally appropriate. We grouped traits into categories (plant growth, yield/tuber attributes, cooking quality) during the initial discussions, allowing farmers to rank category importance before specific trait comparisons, which reduced complexity. The procedure showed the difficulties of explaining hypothetical trade-offs but the use of local languages and accessible examples in the interviews helped to overcome these difficulties. The results are consistent with other participatory breeding programs where survey design adapted to the local environment results in credible data. For example, a similar conjoint survey on cassava used local characteristic levels with almost 800 Nigerian stakeholders finding significant preference clusters [25]. Our experience indicates that with sufficient training and localization, even advanced preference elicitation techniques such as 1000minds may be successfully used among smallholder farmers to capture implicit trait values.
Typologies and implications for yam breeding
The analysis of the 1000minds survey identified several farmer typologies, groups of respondents who have similar trait preference profiles. Four major typologies have formed in Uganda’s yam farming community, namely: crop value, crop output, product quality, and plant survival-oriented groupings. Farmers in the crop value group were very market-oriented, giving the closest attention to overall crop value (yield and market prices combined). They also looked at tuber appearance and cooking characteristics that influence marketability, such as tuber flesh color and softness of the yam after cooking. This implies that these farmers, potentially those with stronger market access or more commercial orientation, desire varieties that optimize economic returns and consumer appeal.
The crop output group was more oriented toward production. In this cluster, respondents ranked high qualities that contribute to overall productivity, total yield, and resistance to main yam diseases. Tuber flesh color was still sought, but qualities that guaranteed a large, dependable yield were more important. On the other hand, the product quality typology was made up of farmers who emphasized end-use quality attributes. They liked high dry matter (mealy yam with dense texture) and tuber shape, which may be associated with ease of processing or cultural preference for certain recipes. This group is mainly concerned with how the yam does on the platform and in storage, not just yield. Finally, the plant survival group was directed at survival and resilience, with a strong emphasis on disease resistance (especially anthracnose) and tuber form being key. The emphasis is on hardy yams that can resist biotic challenges, and which can be stored or replanted.
These four typologies highlight that Ugandan yam farmers are not a homogeneous group, but rather there are clear sub-groups with various “ideal variety” profiles. This variability has important implications for the breeding of yams. It is doubtful that one new variety will be suitable for all farmers. Instead, breeders may have to balance features or establish a range of types to fulfill their various needs. For example, a breeding program may develop one line that is focused on yield and disease resistance for the crop output segment, and another line that is focused on cooking quality (e.g., dry matter and taste) for the product quality segment, but with broadly acceptable traits (e.g., basic yield levels and moderate disease resistance) across all segments. Recognizing such heterogeneity is vital to enhancing adoption: if a variety lacks the feature most important to a particular group (e.g., poor storability for individuals who value shelf-life), that group will be less likely to adopt, even if other traits are improved. Similar results for other crops validate this method. Parallel research on cassava showed separate clusters of trait preferences that are associated with gender, food security status, and area, indicating that various user groups value agronomic features in systematically different ways [28].
Such disparities, which we call end-user segmentation, enable breeding programs to establish more informed breeding goals. Breeders may define distinct profiles of products that correlate to each typology [28], instead of just using average preferences. This might include the development of selection indices that incorporate farmer preference weights or stage-gate decisions to move breeding lines forward. Also, the typology method allows for targeted positioning of the diversity: extension services and seed systems can align new yam varieties with the people that value them most. For example, a high-dry-matter, long-storage type might be targeted to food-insecure areas where household usage and in-ground storage are important.
Meanwhile, a high-yielding, disease-resistant cultivar can be pushed in places where farmers want to optimize production for sale. Finally, the use of crop value, yield, quality, and survival-oriented farmer typologies can be a strategic lens for yam improvement. This ensures that breeding is demand-driven, consistent with the different goals of end users, and exposes breeders to potential trade-offs (e.g., selection for very high dry matter can mistakenly limit yield, or vice versa) that must be addressed by detailed breeding and product placement. Ultimately, the adoption of these typologies will lead to more inclusive genetic improvements providing diverse forms of benefit (yield, income, food quality, risk reduction) to different sectors of yam growers.
Adding value to yam variety improvement, diffusion, and adoption
The findings of this study provide value to yam improvement efforts in Uganda, especially on breeding priorities and seed system development. Knowledge of production systems and restrictions in agro-ecological zones gives a context for breeding and delivery. Farmers highlighted lack of access to quality planting material (40.4%) as a key limitation, while 26.5% said there was unavailability of better varieties. Even the best new variety will have a limited impact if the seed system problems are not addressed. Investing in seed networks for yams will ensure that newly bred high-performing varieties can reach farmers more efficiently. Therefore, breeding programs must work with seed multiplication initiatives, and establishing a sustainable yam seed system through community producers or private partnerships is critical for farmers to obtain improved varieties. Equally important is focusing breeding objectives on the traits that farmers prioritize to make new varieties attractive to end-users. By aligning breeding targets with farmers expressed needs, yam breeding programs can develop end-user-preferred climate-adapted varieties that address on-farm constraints and production objectives. This demand-driven orientation is expected to accelerate the replacement of landraces with superior varieties, thereby improving yam productivity and resilience.
Our research makes breeding goals relevant: creating cultivars that farmers choose is useless if they stay at the research sites. Farmer typologies and trait preferences are important to guide breeding aims and selection criteria. The study results indicate that farmers are not a homogeneous group and different sectors of farmers value qualities differently. This acknowledgment of varied tastes allows breeders to more precisely define their objectives for certain target populations. Data on trait preference can be translated into economic values and selection indices that can be used to influence breeding decisions. By assigning weights to traits based on farmers ranked importance, breeders can construct selection indices to rank breeding lines objectively for advancement. This ensures that selection during variety development reflects the actual needs of the end-users. For example, if farmers in one cluster prioritize tuber quality or storability, whereas another group prioritizes yield under drought, a breeding program can develop selection indices that balance these traits appropriately. Such data-driven selection indices make the breeding process more efficient and more demand-responsive. Ultimately, integrating farmer typologies and preferences into breeding pipelines helps to define breeding goals that are more relevant and inclusive, thereby increasing the likelihood that released yam varieties will be widely adopted.
This study underscores the importance of participatory processes and market intelligence in sustaining the impacts of variety adoption and scaling. Engaging end-users through participatory variety evaluation ensures that crop improvement remains grounded in farmers’ realities. Approaches such as stage-gate variety development, which integrates end-user needs, will facilitate higher adoption by fostering ownership and relevance. This participatory survey demonstrates how end-user input shapes breeding priorities. However, maximizing the impact requires looking beyond the farmers. The demand-led framework should incorporate market intelligence to capture the preferences of value chain actors including seed producers, consumers, processors, and traders [29]. Recent research has shown that gathering trait preferences across the value chain provides a comprehensive perspective for breeding programs [26]. These insights demonstrate differences in preferences across different groups and assist breeders to develop varieties meeting both agronomic and market needs. As such, market knowledge from crowdsourcing is important, as preferences amongst actors differ. The integration of participatory breeding and market knowledge can help yam improvement projects to develop varieties with broad appeal. This method will facilitate adoption rates as varieties are moved from research to market. The more they adopt, the bigger the effect on food security and income in the community [6]. In conclusion, including participatory methods and market feedback in yam breeding would improve acceptance and optimize the developmental benefit in Uganda.
Conclusion
Yam has strong potential to contribute to food, nutrition, and income security in Uganda, yet remains underutilized due to limited research, absence of improved varieties, and weak seed systems. This study filled part of that gap by profiling yam production systems, farmer typologies, and trait preferences across major yam-growing zones. Farmers prioritized traits such as high fresh tuber yield, good mealiness (linked to dry matter), early maturity, disease resistance, and long shelf-life. They also highlighted poor access to quality seed yam as a major constraint. By applying the 1000minds multi-criteria tool, the study quantified trait importance and revealed four distinct farmer typologies, showing that preferences are not uniform. These insights provide a foundation for demand-led breeding that aligns with farmer realities and can improve adoption rates. The study also offers a scalable framework for integrating farmer-driven preferences into breeding pipelines. However, future work should expand beyond producers to include consumers, traders, and processors, and analyze the socio-economic factors that shape trait priorities.
Policy recommendation
The study highlighted important farmer needs for yam in Uganda, including high yield, early maturity, good mealiness, disease resistance, and long shelf-life. Major production obstacles notably included limited access to quality seed yam and dependence on landraces. The findings provide an evidentiary framework for guiding policy and breeding initiatives. First, Uganda’s National Agricultural Research System (NARS) should incorporate farmer-preferred features into yam breeding objectives and product profiles. The integration of these criteria in breeding pipelines and participatory selection will help to ensure new varieties are aligned with user demands and have higher adoption potential.
Second, although this study did not assess seed multiplication methods, the reported unavailability of planting material highlights the need for robust seed systems. Policymakers and research institutes should invest in projects to improve the production and distribution of clean planting material. Techniques such as yam miniset and vine propagation (not evaluated here) could be studied as feasible remedies to the seed limitation reported by farmers. Thirdly, yam should be acknowledged as a strategic crop in Uganda’s agricultural policy framework. Investment in breeding, seed systems and value chain development will further improve productivity, resilience and market integration.
Finally, future studies and interventions should broaden trait preference assessments to include consumers, merchants and processors. Knowledge of the whole value chain will boost breeding design and ensure that new yam varieties are not only agronomically superior, but also commercially and socially relevant.
Supporting information
S1 Fig. Farmer preference of plant and storage traits across agroecological zones.
https://doi.org/10.1371/journal.pone.0362574.s003
(TIFF)
S2 Fig. Farmer preference of yield and tuber traits across agroecological zones.
https://doi.org/10.1371/journal.pone.0362574.s004
(TIFF)
S3 Fig. Farmer preference of cooking/ culinary traits across agroecological zones.
https://doi.org/10.1371/journal.pone.0362574.s005
(TIFF)
Acknowledgments
We acknowledge the support and resources of Scientists in Crop Improvement for Food Security in Africa (SCIFSA), National Agricultural Research Organization – National Crop Resource Research Institute (NARO-NaCRRI), Makerere Regional Center for Crop Improvement (MaRCCI), Council for Scientific and Industrial Research – Savannah Agricultural Research Institute (CSIR-SARI), and the International Institute of Tropical Agriculture (IITA), Ibadan.
References
- 1. Liu L, Huang Y, Huang X, Yang J, Wu W, Xu Y, et al. Characterization of the dioscorin gene family in dioscorea alata reveals a role in tuber development and environmental response. Int J Mol Sci. 2017;18(7):1579. pmid:28726732
- 2. Souza JT de, Santos ACOS, Monteiro FP, Batista MR, Santos PO dos, Carvalho LF, et al. Genetic diversity and population densities of endophytic Bacillus spp. in yam plants. Bragantia. 2017;76(2):203–8.
- 3. Tamiru M, Becker HC, Maass BL. Diversity, distribution and management of yam landraces (Dioscorea spp.) in Southern Ethiopia. Genet Resour Crop Evol. 2008;55:115–31.
- 4. Barton H. Yams: origins and development. In: Smith C, editor. Encyclopedia of Global Archaeology. New York, NY: Springer. 2014.
- 5. Girma G, Hyma KE, Asiedu R, Mitchell SE, Gedil M, Spillane C. Next-generation sequencing based genotyping, cytometry and phenotyping for understanding diversity and evolution of Guinea yams. Theor Appl Genet. 2014;127(8):1783–94. pmid:24981608
- 6. Ifeoma I, Agwu A. Assessment of food security situation among farming households in rural areas of Kano State, Nigeria. j cent eur agric. 2014;15(1):94–107.
- 7.
Akello B, Biso G, Odongkara K. Towards increasing and sustaining production of yam for food security and improved livelihoods in Uganda. In: 10th African Crop Science Conference Proceedings, Maputo, Mozambique, 2011. https://www.cabdirect.org/cabdirect/abstract/20133200290
- 8. Kagoda F, Coyne DL, Asiedu J, Wanyera N, Mudiope J, Dusabe J. Yam germplasm conservation and screening in Uganda. In: Tenywa JS, Adipala E, Nampala P, Tusiime G, Okori P, Kyamuhangire W, editors. Seventh African Crop Science Conference Proceedings. 2005. p. 245–9.
- 9.
Kilimo Trust. Analysis of Demand and Value Chains of Yams Sub-Sectors in Tanzania and Uganda. 2013. www.kilimotrust.org
- 10.
FAOSTAT. Africa Root Crop Production Statistics. https://www.fao.org/faostat/en/#home Accessed 2022 May 16.
- 11. Coyne DL, Talwana HAL, Maslen NR. Plant-parasitic nematodes associated with root and tuber crops in Uganda. Afr Plant Prot. 2003;9:87–98.
- 12. Adjei EA, Esuma W, Alicai T, Bhattacharjee R, Dramadri IO, Agaba R, et al. Phenotypic Diversity within Ugandan Yam (Dioscorea species) Germplasm Collection. Int J Agronomy. 2022;2022:1–10.
- 13. Amponsah Adjei E, Esuma W, Alicai T, Bhattacharjee R, Dramadri IO, Edema R. Genetic diversity and population structure of Uganda’s yam (Dioscorea spp.) genetic resource based on DArTseq. PLoS One. 2023;18.
- 14. Persley G, Kimani P, Yao NK, Tongoona P. The Business of Plant Breeding: Market-led Approaches to New Variety Design in Africa. 2017.
- 15. Balogun I, Garner E, Amer P, Fennessy P, Teeken B, Olaosebikan O, et al. From traits to typologies: Piloting new approaches to profiling trait preferences along the cassava value chain in Nigeria. Crop Science. 2022;62(1):259–74.
- 16. Varshney RK, Terauchi R, McCouch SR. Harvesting the promising fruits of genomics: applying genome sequencing technologies to crop breeding. PLoS Biol. 2014;12(6):e1001883. pmid:24914810
- 17.
Environmental Systems Research Institute ESRI. ArcGIS Desktop. Redlands, CA: Environmental Systems Research Institute. 2020.
- 18.
Uganda Bureau of Statistics. The Annual Labour Force Survey 2018/19 – Main Report. Kampala, Uganda. 2019. https://www.ubos.org/
- 19. Bartlett JE, Kotrlik JW, Higgins CC. Organizational research: determining appropriate sample size in survey research. Information Technology, Learning, and Performance Journal. 2001;19:43–50.
- 20. de Bekker-Grob EW, Donkers B, Jonker MF, Stolk EA. Sample size requirements for discrete-choice experiments in healthcare: a practical guide. Patient. 2015;8(5):373–84. pmid:25726010
- 21.
Stata statistical software. Release 18. College Station: StataCorp LLC. 2022.
- 22.
R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. 2022.
- 23. McKight PE, Najab J. Kruskal‐Wallis Test. The Corsini Encyclopedia of Psychology. Wiley. 2010. p. 1–1.
- 24. Martin-Collado D, Byrne TJ, Amer PR, Santos BFS, Axford M, Pryce JE. Analyzing the heterogeneity of farmers’ preferences for improvements in dairy cow traits using farmer typologies. J Dairy Sci. 2015;98(6):4148–61. pmid:25864048
- 25. Balogun I, Garner E, Amer P, Fennessy P, Teeken B, Olaosebikan O, et al. From traits to typologies: Piloting new approaches to profiling trait preferences along the cassava value chain in Nigeria. Crop Science. 2022;62(1):259–74.
- 26. Okello JJ, Swanckaert J, Martin-Collado D, Santos B, Yada B, Mwanga ROM, et al. Market intelligence and incentive-based trait ranking for plant breeding: a sweetpotato pilot in Uganda. Front Plant Sci. 2022;13:808597. pmid:35317017
- 27. Byrne TJ, Amer PR, Fennessy PF, Hansen P, Wickham BW. A preference-based approach to deriving breeding objectives: applied to sheep breeding. Animal. 2012;6(5):778–88.
- 28. Teeken B, Garner E, Agbona A, Balogun I, Olaosebikan O, Bello A, et al. Beyond “women’s traits”: exploring how gender, social difference, and household characteristics influence trait preferences. Front Sustain Food Syst. 2021;5.
- 29.
Tools, services and expertize to accelerate genetic gains of breeding programs targeting the developing world. EiB. 2019.

관련 뉴스