Diversity, trait preferences, management and utilization of yams landraces (Dioscorea species): an orphan crop in DR Congo

Yam (Dioscorea spp.) is cultivated in many villages of DR Congo as a means to sustain food security and alleviate poverty. However, the extent of the existing diversity has not been studied in details thus, considered as an orphan. A survey covering 540 farmers in 54 villages was conducted in six major yam growing territories covering three provinces in DR Congo to investigate the diversity, management and utilization of yam landraces using pre-elaborate questionnaires. Subject to synonymy, a total of 67 landraces from five different species were recorded. Farmers’ challenges limiting yam production were poor tuber qualities (69%), harvest pest attack (7%), difficulty in harvesting (6%), poor soil status (6%). The overall diversity was moderate among the recorded yam germplasm maintained at the household level (1.32) and variability exist in diversity amongst the territories and provinces. Farmers’ in territories of Tshopo and Mongala provinces maintained higher level of germplasm diversity (2.79 and 2.77) compared to the farmers in territories of Bas-Uélé (1.67). Some yam landraces had limited abundance and distribution due to loss of production interest in many villages attributable to poisons contained hence, resulting in possible extinction. Farmers’ most preferred seed source for cultivation were backyard (43%) and exchange with neighboring farmers (31%) with the objective of meeting food security and generating income. In villages where yam production is expanding, farmers are relying on landraces with good tuber qualities and high yield even though they are late maturing. This study revealed the knowledge of yam landrace diversity, constraints to production and farmers’ preferences criteria as a guide for collection and conservation of yam germplasm for yam improvement intervention.

Statistical analysis. Descriptive statistics was used in generating summary tables and means were tested for significance using 95% confidence interval where necessary. Spearman's correlation was used to assess the relationship among the socio-demographic characteristics and the yam diversity maintained at the household level using corrplot library R package 25 . Shannon-Weiner diversity index (H') 26 , species richness (number of unique yam landrace in an area) and equitability (E) were used to quantify the diversity of yam at the village, territory and province levels as shown below: where N is the number of yam species in villages, territories and provinces, pi the proportional abundance of the ith yam species.
where H′ is Shannon diversity index and H max represent the theorectical maximum diversity.

Results
Sociodemographic characteristics of the study areas. Among the 540 farmers, 80.60% were male while 19.40% were female with an average experience in yam production being 13 years. Besides, 1.30% of the farmers surveyed were teenagers (< 20 years), 73.30% were adults class (20-50 years) and 25 www.nature.com/scientificreports/ (> 50 years). Secondary and primary educations (59.80% and 34.60%, respectively) were the most common form of education. The major activity of the farmers is farming (98.70%). The average family size of the survey participants was approximately seven members with a minimum of one member and a maximum of 24 members in the case of extended family. The average farm size was 1.49 hectare with a recorded minimum farm size of 0.10 hectare and maximum farm size of 15 hectares. The average size of yam field under cultivation was 0.10 hectare with a maximum of 1.50 hectares. With respect to yam cultivation relative to other crops (food and cash), yam took an approximate of 7% of the total available land for crop cultivation in general (Table 1).

Constraints to yam production in DR Congo.
Generally, tuber quality forms the highest proportion (69%) of the farmers' constraints to yam production. The traits reported by farmers under the tuber quality were poor postharvest shelf life (30.58%), high tuber flesh oxidation (14.54%), poor taste (~ 14%) and rapid hardiness of tubers (10%) ( Table 2). The biotic factors (9.56%) followed after tuber quality. This was largely influenced by the proportion of farmers who reported harvest pests' problem (~ 7%) while theft was only ~ 3%. The abiotic factors form the third constraint reported by farmers (8.96%). This constraint was influenced by farmers who reported poor soil (6%) compared to other factors in this category.
Agronomic quality (7.26%) was also identified by farmers as a constraint to production influenced by difficulty in harvesting (6%). The last production constraint reported by farmers was marketing (~ 5%). This forms the least of the farmers' production constraints and is influenced by poor market price for yam tubers (~ 4%). In summary, the major factors reported by farmers as constraints to yam production in DR Congo are: poor post-harvest shelf life, high tuber flesh oxidation, poor tuber taste, rapid tuber hardiness, pests, difficulty in harvesting, poor soil, and poor market price.
Yam landrace diversity. The mean landrace diversity expressed as Shannon index (H') and the H max (the maximum possible in study areas) were 1.32 and 1.58 respectively (Sup Table 2). Landrace richness was observed to significantly differ between the provinces. Tshopo province has the highest landrace richness (35) which is statistically different from that of Mongala (26) and Bas-Uélé (10). The territories within respectives provinces have statistically similar landrace richness however different from territories of other provinces ( Fig. 2A 2C).
Considering the possibility of synonyms, the numbers of yam landraces recorded ranged from two to ten per village. The minimum number of landrace (2) was observed in the villages of Bambesa (Adiwaya, Bongenge, Dingima, and Mendigba) and Buta (Bobomale, Bonzo and Boyelia) territories while the maximum number of landrace (10) was observed in the villages of Isangi territory (Q. Bangala and Q. Lumumba) (Sup Table 2).
The relationship between socio-demographic characteristics and the number of landraces cultivated at individual household level is presented in Fig. 3. Yam experience in cultivation, farm size, and farmer age had significantly positive relationship with the number of landraces cultivated at household level.
Throughout the entire survey period, five different species of yams (D. rotundata, D. cayenensis, D. alata, D. dumetorum and D. bulbifera) were recorded (Fig. 4). These species form the major landraces that were cultivated by farmers as well as in the forest in the survey areas. The quantity of landrace varieties per species also varies from one village to another with number varying from one to three (Dioscorea rotundata, Dioscorea cayenensis, and Dioscorea dumetorum), one to four (Dioscorea alata), and one to two (Dioscorea bulbifera). Only two villages: Quartier Bangala and Yalinga in Isangi territory (Tshopo province) were found to have all the yam species represented ( Table 3).
The maturity classification of the yam landraces in DR Congo was also assessed ( Table 4). 74% of the farmers reported that the yams are of the late maturity class (> 10 months), 17% reported intermediate maturity class (8-10 months) and 9% reported early maturity class (< 8 months). At the territory level, majority of the farmers The relationship among the unique 67 yam landraces observed during the survey period with respect to agronomic characters (tuber shape and maturity duration), tuber quality parameters (tuber colour, tuber taste, and tuber oxidation (browning) after cooking) and yam species is represented in Fig. 5. The cluster analysis    Cluster four comprised of landraces of the D. cayenensis, D. alata, D. dumetorum having cylindrical and oblong tuber shape, white, yellow and purple colour, early to late maturity, sweet to bitter taste, and no-oxidation to heavy browning after cooking. Following the observed diversity of yams in the study areas, approximately 74% of the farmers also reported landraces experiencing decline in attention by farmers or extinction (Table 5). A total of 14 yam landrace names were reported by farmers that fall in this category of event. Of the proportion of farmers that reported varietal loss, 78% reported poisons as the principal reason for extinction, followed by poor tuber quality attributes (10%) and late maturity (8%). Fig. 6 showed that achieving food security is generally of significant priority followed by revenue in all the territories considered. With respect to food security, farmers in Bambesa showed higher preference (71%) significantly different from all other territories. Farmers in Bumba showed the least preference (52%) though statistically similar to that of Lisala and Isangi territories. With respect to revenue, farmers in Bumba showed the highest preference (47%) significantly different from other all other territories while farmers in Bambesa showed the least preference (29%) (Fig. 6).

Yam cultivation and cultural practices. The major reason for yam cultivation presented in
Cropping system mostly observed in DR Congo was intercropping pattern (90%) where yam is mostly intercropped with other arable crops. Approximately 10% of the farmers practice sole cropping system (Table 6). At the province and territory level, similar trend of results was obtained except for Bumba territory where the gap between these two cropping systems was significantly reduced. In this territory, over 40% of the farmers practice sole cropping system for yams as compared to other territories where less than 15% of the farmers were observed.
During the production period of yam (Table 6), approximately 58% of the farmers provide staking support for their yams. At the level of the province, only 18% responded positively to supporting their yams with stakes during the production cycle in Bas-Uélé. Provinces of Mongala and Tshopo however have appreciable proportion (above 70%) of farmers that incorporated staking as part of their cultural practice. At the level of the territory,  While the majority of the farmers in these territories have neglected staking, farmers in the territory Bumba have made it a necessary requirement for yam production ( Table 6). Throughout the survey, the harvest signature used by farmers is total senescence of leaves. Harvesting is only done once as farmers do not practice milking. Once harvesting of the tubers is completed, approximately 89% of the farmers prefer to go for storage of their produces while 11% of the farmers send their produce directly to the market for sale. In all the provinces and territories, at least 11% of the farmers send their produces directly from the field to the market after harvest except for the famers in the province Bas-Uélé (Bambesa and Buta territories) that prefer to go for storage (Table 6).

Seedyam production system. Of four different sources of seedyam presented to the farmers, overall
results showed that the backyard source (43%) (retention from previous season harvest) was the most used means to obtain seedyam. Neighbor source (31%) (exchange with neighboring farmers and friends) followed in significant ranking. Forest source (11%) is the least used means (Fig. 7A). At the level of the province, farmers in the Bas-Uélé province significantly preferred the use of Backyard source and Neighbor sources (35% and 33% respectively). Farmers in Mongala province significanty prefer the use of backyard source (46%) with neighbor and market coming as second significant preferences. Farmers in Tshopo province significantly preferred backyard source (53%) followed by neighbor source (Fig. 7A).
At the level of the territory, farmers in Bambesa statiscally preferred the use of the forest source (50%) compared to all other sources. Farmers in Buta significantly showed preference for neighbor (46%) and backyard (42%) sources. In Bumba and Lislala, significant preference was reported for backyard source (48% and 45% respectively). Farmers in Isangi territory however reported significant preference for backyard and neighbor  Bamboo stick - www.nature.com/scientificreports/ sources (52% and 48% respectively). In Kisangani, farmers reported the significant preference for backyard sources (Fig. 7B).
Farmers' preference criteria for yam selection and utilization in DR Congo. The assessment of farmers' preference criterial for yam selection and utilization in DR Congo is presented in Table 7. The prefer-  www.nature.com/scientificreports/ ence for yams variety with good tuber quality (sweet taste and no tuber browning after cooking) was reported by 53% of the farmers, high yield preference was reported by 38% of the farmers, and preference for earliness was reported by 7% of the farmers. Thus, these three criterial accounted for over 97% of the farmers' preference in DR Congo. At the level of the province, similar trend of results was observed across the three province however, this was not totally the case at the level of the territory. Farmers in Bambesa territory only reported the preferences for two criteria only (high yield = 50% and good tuber quality = 50%). Farmers in Buta showed preference for three criteria with importance placed on two (good tuber quality = 57% and high yield = 42%). Farmers in Bumba reported four criteria with importance placed on two criteria (high yield = 47% and good tuber quality = 47%). Farmers in Lisala reported the preference for all the traits with emphasis on good tuber quality (59%) and high yield (30%). Farmers in Kisangani reported the preference for three criteria (good tuber quality = 47%; earliness = 37%; high yield = 16%). Farmers in Isangi also reported preference for three traits with emphasis on two (good tuber quality = 56% and high yield = 42%). The major form of consumption assessed amongst the survey respondents (Table 7) showed that boiled form is the most preferred method of consumption (76%). Value addition form remain unknown to the survey respondents hence remain unexploited. With respect to the territories, farmers in Bambesa and Buta do not consume yam in any other form than boiled form which maybe attributable to the challenge of low tuber quality prevalence in these territories. Farmers in Bumba reported four forms of consumption (boiled yam = 33%; grilled yam = 25%; pounded yam = 22%; fried yam = 19). This territory has some of the best landrace varieties of yams in the survey regions. Farmers in Lisala preferred boiled form (~ 58%) and pounded form (41%). Farmers in Isangi had preference for boiled yam (84%) compared to other forms.

Discussions
Constraints linked to yam production in DR Congo. In general, our study reveal the presence of a moderate diversity of landrace that could support the collection and conservation of yam germplasm for future use. Though when this diversity is compared to similar findings from other yam producing countries, it is lower. The provinces (including villages) of the forest agro-ecology holds larger landrace diversity than that of the transition zone. Thus, we could infer that the variation in the number of landrcaes at household level could be attributed to agro-ecology, climatic and human factors. Increase in the diversity of yam species in the forest has also been reported by 28,29 in Togo and 30 in cultivated species in Ghana. Our study observed five different species of Dioscorea, however, we cannot totally ascertain that all the landrace morphotypes within each D. species are truly genetically distinct due to the possibility of linguistic polymorphism. Hence, the likelihood that this study has underestimated or overestimated the actual number of landraces cannot be ruled out. Similar studies have also reported the influence of linguistic polymorphism in bush yam in the central region of Ghana 31 ; bitter yam in Benin 32,33 ; and Dioscorea species in Southern Ethiopia 10 . The proposed challenge with linguistic polymorphism in this study could be easily clarified with further study on morphological and molecular characterization of the landraces observed.
Aside comparably lower landrace diversity to other major yam producing countries, yam production is faced with many constraints in DR Congo with the principal being tuber quality attributes of the landraces (poor postharvest shelf life, high tuber flesh oxidation, poor taste, and rapid tuber hardiness). This has largely discouraged a lot of farmers from the cultivation of yam, affected market demand leading to poor pricing by the consumers and thus, reducing farmers profit margins. The influence of tuber quality attributes on the adoption and abandonment of yams has also been reported in many studies. For example, It was reported as a contributing factor for yam varietal loss 34 as well as abandonment of bush yam 28 in Togo. Other important constraints identified by this study include difficulty in harvesting, poor soil, and pest and disease.
Exisiting landrace maturity is mostly the late type. We infer this could be the consequence of the lack of genetic improvement through breeding and selection. Even with the presence of the National Agricultural Study and Research Institute (INERA) and numerous higher institutions, yams have received insignificant attention with respect to varietal improvement. Such programs could have assisted in proper collection, documentation, and conservation of yam germpalsm to prevent loss of genepools. The insignifant attention has also enhanced the loss of some landraces. Farmers reported 14 landraces that have been abandoned in cultivation and/or usage primarily due to poisons. Of the observed species D. bulbifera has been mostly implicated in this regard. However, not all the morphotypes of this species are poisonous as it is still being consumed by some people as observed during the survey. The question remains the proper differentiation of the morphotypes safe for consumption from the genepool. Modern breeding techniques such as detailed morphological characterization, molecular tools (DNA markers e.g. SSRs and SNPs) and DNA sequencing could help to tackle this challenge.
The practice of generating seedyam from previous harvests is a common phenomenon for seasonal yam cultivators. Farmers also engage in trade by batter (changing other food crops for seedyam for field establishment). These methods has contributed to low viability and inadequacy of seeds availability. This system is currently putting the yam producers at a disadvantage unknown to many of the farmers considering the weight of the setts for planting, lack of the knowledge and/or zeal to practice double harvesting, and high tuber losses due to poor storage. Different methods of generating seedyam have been researched and proposed to yam farmers in many yam producing countries 35,36 . Of these methods, the miniset system appeared to be most successfully adopted by farmers in many yam producing countries and thus could be attempted in DR Congo.
Farmers' preferences for selection and utilization of yams were tuber quality attributes (good taste and nonoxidizing flesh color) and agronomic characteristics (high-yield and earliness). According to the farmers, realization of these criteria will spark a new line of interest in the mind of many farmers. Thus, establishing a yam improvement program with the objective of assisting the farmers' should put these criteria into consideration. Similar study have also reported good tuber qualities to enhance yam marketability and cultivation 28  www.nature.com/scientificreports/ Perspectives for yam improvement in DR Congo. Enobblement effort and seasonal cultivation by farmers have kept yam diversity in DR Congo from total loss. The establishment of yam improvement program would go a long way to providing a substancial solution to majority of the constraints linked to yam production. Such program would facilitates the collection and conservation of germpslams to prevent loss of yam gene pool, increase yam genetic diversity through hybridization and introduction, facilitates the proper characterization of yams to distinguish the consumable and the non-consumable morphotypes, and develop new and improved yams that will meet farmers' and consumers' requirements through selection. These have been observed in countries where yam improvement program is currently existing such as Nigeria, Ghana, etc. Another important perspective in ensuring yam improvememt in DR Congo is the dissemination of information or technology transfer. This is principally important for generating good and quality seedyam. In the presence of yam improvement program, efforts sould be made to organize trainings on technology transfer (yam minisett to begin with) on seedyam production for farmers' through the extension experts as they are the closest to the farmers and most trusted by the farmers. Participatory plant breeding approach would also create an atmosphere for close relationship with the yam farmers as a means to rapidly understand their challenges.

Conclusion
The study revealed a moderate diversity for yams across five different species in DR Congo. The diversity was relatively higher in Tshopo and Mongala provinces than Bas-Uélé province. The principal challenges limiting yam production in DR Congo surrounds the tuber quality attributes (poor post-harvest shelf life, tuber oxidation, poor taste, and rapid tuber hardiness) of the available varieties. Yam cultivation is targeted to meet the food and financial demands of the populace. Yam farmers preferred yams varieties with good tuber qualities, high yield and early maturing varieties. In the absence of formal seedyam production practice, farmers practiced the system of producing seeds by themselves as well as informal exchange of seeds with neighbors and friends. The establishment of a yam improvement program to meet farmers' selection criteria, collection and conservation of yam germplasm, and the development of an effective seed delivery system to meet the seed availability and viability needs could increase yam production and profitability in DR Congo.