White Guinea yam (Dioscorea rotundata Poir.) landraces trait profiling and setting benchmark for breeding programs in the Republic of Benin

To meet the high demand for white Guinea yam, there is a need to develop and release improved varieties to farmers. Unfortunately, low rate of adoption of most of the improved yam varieties by both producers and consumers was observed. Information regarding agronomic characteristics and food qualities of popular white Guinea yam landraces with high market value are not available to establish minimum standards to be considered by breeding programs. To fill this gap, surveys using rural appraisal tools were carried out in 20 villages and 16 markets throughout Benin. Data on the agronomic performance suggested that for an improved variety to be adopted by Beninese farmers it should have a minimum yield of 4.16 ± 0.15 kg per mound, and average number of marketable tubers of 1.23 ± 0.05, a mean tuber length of 36.41 ± 1.22 cm, and a minimum diameter of 25.44 ± 1.16 cm. The sensorial attributes for boiled and pounded tubers of this improved variety should have minimum score of 3.16 for texture, 0.75 for softness, 3.75 for elasticity, and 1.34 for colour during the sensory evaluation. The improved variety must also have a minimum average severity score of 1.1 for yam mosaic virus disease, 1.33 for anthracnose and 1 for nematodes. Landraces Amoula, Laboko, and Djilaadja should be considered as the standard for yield, sensory attributes, and tolerance to pest and diseases while landraces Danwari, Kodjewe, Mondji, and Gnidou should be characterized as possessing good flowering and fruit setting capacities for breeding programs.


Introduction
Yam (Dioscorea spp.) is a climbing crop species that produces edible underground tubers, which contribute to the food security and poverty alleviation in many developing countries  [1]. About 94% of the global yam production is from West Africa with the Republic of Benin as the fourth world producer [2]. In this region, among the cultivated yam species, the white Guinea yam (Dioscorea rotundata Poir.) remains the most produced and preferred [3,4]. Its tubers appear to be a good source of carbohydrates and essential minerals such as magnesium, zinc, iron and phosphorus [5]. White Guinea yam plays an important role in the socio-cultural life of local farming communities through festivals dedicated to the release of new yam tubers [6], and in the rural household economy [7]. Unfortunately, white Guinea yam production is constrained by numerous biotic and abiotic constraints resulting in yield reduction and varietal diversity loss [8]. According to Hounnou et al. [9], climate change could lead to significant reduction in Benin yam production up to 6.87% by 2025. To meet future demands for white Guinea yam and to ensure food security, there is a need to develop and release improved yam varieties for farmers [10]. However, the adoption rate of improved yam varieties introduced in the past in the Republic of Benin is low [11]. For example, the adoption rate of improved yam varieties (TDr 131 and TDr 205) introduced and popularized by the International Institute of Tropical Agriculture (IITA) in Benin was estimated at 37% in 2007 [12]. This low adoption rate could be due to several factors including poor match with consumer needs and preferences, poor dissemination, lack of seeds, but also under local conditions to the small adaptation of improved varieties over landraces [11,13]. Although, improved yam varieties are known to have superior agronomic and food quality traits [14], unfortunately there is no defined minimum breeding standards for tuber yield, food quality, pest and disease tolerance to be considered when developing new white Guinea yam varieties.
Beninese farmers perceive popular and elite yam landraces as those having high yield potential, resistance or tolerance to many biotic and abiotic constraints [8,15]. Therefore, it is important to determine the average agronomic performance across production zones of these popular yam landraces, which could serve as minimum breeding standards for acceptability of improved white Guinea yams by Beninese farmers. Presently, there is lack of information on the agronomic characteristics and food qualities of popular white Guinea yam landraces with high market value to allow the establishment of minimum quality standards for the national yam breeding program of Benin.
The selection of white Guinea yam landraces is usually based on a set of farmers' preference criteria of which good yield, good quality and high market value are the most important [8]. Customer needs are now the focus of yam breeding programs in Africa [14]. Although efforts of several research programs have been devoted to white Guinea yam in the Republic of Benin, there is very limited information on the yield and market value of the popular landraces. However, this information is crucial to the development of new yam varieties with improved agronomic characteristics with a greater probability of adoption in the markets. To fill these gaps, it is important to explore the yam producing villages and local markets in order to identify the popular yam landraces and document their market value and agronomic performance including their on-farm phytosanitary status as white Guinea yam landraces are highly susceptible to pathogens [8,16], which affect their yield and market values [17].
In addition to the good agronomic characteristics required in any improved yam variety, the sensory attributes of dishes based on this variety are major determinants of its adoption by both producers and consumers [18][19][20]. Indeed, the texture and taste of pounded and boiled yam, which are the most popular yam-based dishes in Benin Republic are the main sensory quality attributes of Beninese farmers, processors, and consumers [20,21]. Therefore, it is important to determine the minimum standards of these sensory attributes of popular white Guinea yam in order to be locally adopted.
Sixteen closest rural markets to the surveyed villages were selected for the estimation of the market value of the identified popular yam landraces. In each market, 5 to 12 yam wholesalers and retailers of yam were randomly selected for the survey. A total of 128 traders were surveyed in the study area. In the markets, yams were sold by retailers in small lots of 3 to 5 tubers per landrace, while the wholesalers sold them in lots of 50 to 100 tubers. It was observed in some markets such as Glazoué market that some traders were selling yam in kilograms using baskets and an adapted weighing scale. The market value of tubers of each yam landrace was estimated regarding the price per kg, the number of sellers and available quantity (in tons). For each yam landrace, five representative lots of tubers were selected and weighed to provide the peak price per kilogram for both wholesale and retail sales. The average price per kg of tubers of all popular yam landraces found in the markets was used as a standard to identify those considered to have high prices. A popular white Guinea yam landrace was considered as sold by many sellers when, Sc � Ns/2 (Sc is the number of local yam variety sellers and Ns total number of randomly selected sellers in the market). A popular white Guinea yam landrace sold by many sellers with high price and available in large quantities (� 1 ton) were considered as having high market value.

On-farm agronomic performance of popular yam landraces with high market value
In each surveyed village, five yam producers were chosen for field visits with the help of the village chief and/or heads of farmer associations. These farmers were selected on the basis of similarity in their farming practices (monoculture, staking, mulching, and no use of fertilizers and pesticides), and in order to include all popular yam landraces with high market values identified at the village level. Three visits were made in each field at different yam development stages. During the first visit (yam germination stage), five plants of each identified popular and high market value landraces were randomly selected and tagged with white tape for data collection. During the second visit, at the vegetative yam development stage (10 to 12 weeks after planting), the vigour, flowering intensity and fruiting intensity were recorded according to the yam descriptors [25]. In addition, the presence/absence of yam mosaic virus (YMV) and anthracnose symptoms was assessed on the leaves using the standard descriptor for yam variety performance evaluation trials [26]. Disease symptom severity for anthracnose and YMV were scored using the scale 1 = no visible symptoms of disease; 2 = few symptoms of 1 to~25% on the plant; 3 = symptoms covering~26 to~50% of the plant; 4 = symptoms on > 51% of the plant; 5 = severe necrosis and death of the plant [26]. At the last visit, at the yam harvest stage (6 to 8 months after planting depending on the landrace), the yield (kg per mound) and yield components (total number of tubers harvested per mound, weight, length and diameter of the longest tuber per mound) were evaluated. In addition, nematode (Scutellonema bradys Steiner and LeHew) damage on harvested yam tubers were assessed using the standard descriptor of Asfaw [26].

Sensory evaluation of popular yam landraces with high market value
Fresh matured tubers of each popular white Guinea yam landrace with high market value were collected from the surveyed villages. The yam tubers were washed with clean water to remove adhering soil and other undesirable materials in the laboratory of the Biotechnology, Genetic Resources and Animal and Plant Breeding (BIORAVE). The yam samples were sorted, handpeeled using kitchen knives and then cut into slices of about 10 cm each. About 2 kg of yam tuber slices from each landrace were cooked (temperature of 100˚C) with 200 ml of water. Cooking times were recorded as they varied from one yam landrace to the other. Qasa yam pounding machine (Model QYP-6000) was used for the pounding of the boiled yam samples for 3 min.
Four randomly selected boiled and pounded yam samples from each landrace were served to consumers to determine consumer acceptability and assess sensory characteristics. The yam dishes prepared from each landrace was coded. To determine the consumer acceptability of the dishes prepared from each landrace, 60 untrained consumer panellists from the students and staff members of the National High School of Applied Biosciences and Biotechnologies (ENSBBA) were selected as panellist based on their willingness to participate. A 5-level hedonic scale ranging from very unpleasant (1) to very pleasant (5) were used for the evaluation [20]. Subsequently, 15 panellists previously trained on the principles of scoring and assessment were recruited to assess the sensory characteristics (hardness, sweetness, colour, aroma, hardness and elasticity) of boiled and pounded tubers of each white Guinea yam landrace. A glass of water was provided to each panellist to rinse the mouth between two tasting sessions.

Ethics declarations
The research protocol was approved by the ethic committee of the National University of Sciences, Technologies, Engineering and Mathematics (UNSTIM). Interviews were carried out in accordance with the guidelines of the Declaration of Helsinki. Written informed consent was obtained from all participants prior to the interviews.

Data analysis
Descriptive (mean, percentage, covariance, etc.) and inferential statistics were used to analyse the data collected. The Hierarchical Cluster Analysis (HCA) was used to establish the relationship between the popular yam landraces with high market value and their agronomic and culinary characteristics as perceived by the surveyed farmers. The data analysis was performed using the R software [27] with the ade4 [28,29], and cluster [30] packages. The dist. binary function was used to calculate Jaccard's similarity and the hclust function was used to construct a dendrogram. Analysis of variance (ANOVA) was performed using the IBM SPSS software version 23. Before the ANOVA test, the data normality was tested using Levene's test. To achieve normality and homogeneity of variances, yield and yield component data were logtransformed (log (x + 1)). Significant differences between means were separated using the Student Newman Keuls test (p < 0.05). Principal Component Analysis (PCA) was performed to describe the relationship between the popular yam landraces and their yield and yield components. Pearson correlation analysis was used to measure the degree of association between the total number of tubers harvested, yield in kg per mound, length, diameter and weight of marketable tubers per mound.
To classify and characterize the popular yam landraces with high market value, the data collected including agro-morphological and sensorial attributes (Table 1) were considered for analysis. The data were analysed using classification performed in two stages. First, dimension reduction was performed on the data matrix containing yam landraces in rows and groups of variables in columns using multiple factor analysis (MFA) to maximise variation between popular yam landraces in the study area. MFA is a factor analysis which builds not only on the variation between yam landraces, but also the influence of groups of variables on the variation [31]. Second, hierarchical cluster analysis (HCA) was performed on the output from the MFA with popular yam landraces grouped such that the within-group variability in landrace characteristic was minimized while between-group variability was maximized. Both MFA and HCA were performed in R software [32] using the Factoshiny package [33]. Finally, the average of each quantitative character in each of the resulting clusters were compared to the global mean for that characteristic using the v-test (Decision statistic) and Mann-Whitney test.
Disease incidence (DI) and severity index (MS) were calculated using the following formula [34]: with IP: Number of infested plants and PS: total number of plants.
with S score of infested plants and IP Infested plants.

Identification of the popular white Guinea yam landraces with high market value
The number of white Guinea yam landraces identified per village varied from 2 to 16 with 8.7 landraces in average (Table 2). Subject to local names, 213 white Guinea yam landraces were inventoried throughout the surveyed villages with 116 landraces identified as popular at the village level. The great majority (62.1%) of the identified popular white Guinea yam landraces were found for sale in the markets surrounding the surveyed villages (Table 2). In the markets, the price of tubers of the identified popular yam varied from one landrace to the other (Table 3). Forty-four popular white Guinea yam landraces which had a high price (� 350 CFA franc � 0.62 $ per kg), were available in high quantity (� 1 tone), and sold by many sellers (� 5) in the markets ( Table 3). The landrace Laboko was the most expensive (650 CFA franc � 1.14 $ per kg) and Gnidou landrace was the cheapest (350 CFA franc/kg). Majority (72.7%) of these white Guinea yam landraces are early maturing with double harvest per year. Subject to synonymy, the Donga department had the highest number (31) of popular yam landraces with high market value, followed by the Collines (30), Borgou (22), and Atacora (18) departments (Table 3).

Participatory evaluation
According to the surveyed farmers of the 44 popular white Guinea yam landraces with high market value recorded, 40 were perceived as possessing very good pounding ability, 34 were very good as boiled yam, 32 were very good as fried yam while 34 produced high quality yam chips (Table 4). Farmers perceived most of these white Guinea yam landraces as having relatively long post-harvest storage ability (32), high productivity (36), tolerance to weeds (35), tolerance to drought (20), and adaptation to poor soils (29). Some elite white Guinea yam landraces have been reported to be tolerant to anthracnose (25), yam mosaic virus (24), and nematodes (29). Using the hierarchical clustering method, the 44 popular yam landraces with high market value were classified into three groups (Fig 1). The first group included 12 latematurating yam landraces (10 to 12 months to maturity), namely "Kokoro group of varieties" that have a single harvest and produce many small tubers, commonly used for the manufacture of yam chips. The second group included 11 early-maturing yam landraces (6 to 8 months to maturity) found to be highly susceptible to low soil moisture conditions, poor soil fertility, anthracnose disease, virus, and weeds. However, some landraces in this second group, namely Gnidou, Okagan and Agatou, were perceived by farmers as resistant to nematodes and having high productivity with very good quality as boiled yam. The twenty-one popular yam landraces of group 3 have early maturity are moderately productive, tolerant to high soil moisture but highly susceptible to nematodes, anthracnose and the yam mosaic virus.  Table 5). The yield of the popular white Guinea yam landraces ranged from 2.0 ± 0.5 kg/mound (Kpanhoura landrace) to 8.6 ± 0.8 kg/ mound (Amoula landrace) with an average of 4.6 ± 1.8 kg/mound. The average yam tuber length and width were 39.4 ± 11.3 cm, and 28.6 ± 11.4 cm, respectively. The Déba landrace had significantly lower marketable tuber weights per mound (1.5 ± 0.2 kg) and higher number of harvested marketable tubers per mound (2.6 ± 0.9). Tognibo landrace had the smallest marketable tubers ( Table 5). Results of correlation analysis revealed a positive and highly significant association between yield and marketable tuber weight (Fig 2). Similarly, tuber length, width, and diameter displayed positive correlations with the weight yield per plant (Fig 2).
Results of the principal component analysis showed that the first axis explained 85.24% of the total variability. The length (LtubL), diameter (DtubL), and weight (PtubL) of tubers and yields were found to be positively correlated with the first axis, while the number of harvested tubers per mound (Ntub) was negatively correlated with the first axis (Fig 2). The principal component analysis clustered the popular white Guinea yam landraces with high market value into three groups with the 12 yam landraces in the first group producing a relatively higher number of tubers per mound. The second group comprising 23 yam landraces had moderate yield performance. The third group comprised nine yam landraces with higher yield as well as large and long yam tubers.

Flowering and fructification.
Twenty-nine elite yam landraces with high market values had low flowering capacity. Of these yam landraces, 24 yam landraces had flowering capacity but no fruit production, whereas five landraces were found to be completely sterile ( Table 6). Two popular yam landraces (Dodo and Effourou) exhibited half flowering and fructification capacities. Seven of the popular yam landraces had high flowering capacity (Table 6). However, only Danwari, Kodjewe, Mondji, and Gnidou landraces had very good potential for flowering and fruit production.

Sensory evaluation
Majority (81.8%) of popular yam landraces with high market values were preferred in both boiled and pounded forms. Only 6.8% and 11.4% landraces were preferred in the boiled and pounded forms, respectively. Ahimon, Baniouré, Déba, Douroubayésirou, and Kratchi landraces were most preferred in the pounded form while, Adoro, Ossoukpana, and Singou landraces were the most preferred in the boiled form. The hedonic tests of boiled and pounded yam of the 44 popular white Guinean landraces with high market value showed that more than half of them were classified as pleasant (95.6%), and 4.5% as very pleasant by the tasters. Boiled and pounded form of Laboko and Djiladja landraces were considered as very pleasant by all the taste panellists. A significant variability in the scores obtained for pounded tubers of the different yam landraces was observed in terms of colour (CV = 29.8%), sweetness (CV = 33.1%), hardness (CV = 34.6%), aroma (CV = 168.1%), presence of lumps (CV = 199.1%) and elasticity (CV = 30.6%). The sweetness of Laboko and Tchee landraces were most appreciated by the panellists, while Tamsam and Baniouré landraces were the least appreciated for this attribute. Okogan, Tamsam, and Gnarambo landraces were considered by panellists as producing pounded yam with good texture attributes. In terms of the elasticity of pounded yam, Laboko had the highest score by the panellists (Table 8). Deba, Djiladja, Laboko, and Tabane landraces were adjudged by the panellists as having the most attractive aroma. The unpleasant presence of lumps in pounded yam, attracted a high score (1) in the Douroubayésirou, Effourou, Tchée, and Tognibo landraces.

Contribution of each category of variables to the explanation of the variations between yam landraces.
The multiple factor analysis showed that the first factor (Dim 1) explained 13.3% of the total variation among the yam landraces, the second factor (Dim 2) 12.4% while the first two axis explained 25.7% of the total variation among the yam landraces (Fig 3). The percent contribution to the inertia of the first factor was highest for resistance to yam viral infection and resistance to anthracnose traits (Fig 3). However, the percent contribution to the inertia of the second factor was highest for the sensorial character followed by the yam yield components, resistance to nematodes and the flowering characters. Fig 4 showed the projection of yam landraces on the first and second axes (Dim 1 and 2) derived from the multiple factor analysis. Table 9 showed the mean values for tuber yield and yield components and the mean scores of the qualitative variables for popular yam landraces within clusters derived from the hierarchical cluster analysis. The cluster 1 included yam landraces with high yield (4.9 kg/mound), big tubers with height of about 43.3 cm and weight of about 4.7 kg and width of about 31.4 cm as well as good taste and good elasticity of the pounded yam. Additionally, the cluster 1 yams exhibited tolerance to pests and diseases (low incidence of viral infection and anthracnose, and low severity of viral infection and nematodes). The cluster 2 included yam landraces characterized by high number of tubers (1.2), good plant vigour, and low incidence of anthracnose, but had the smallest tuber sizes, low mean yield (3.8 kg/mound), high incidence and severity of nematodes, low elasticity, and lump of pounded yam. The cluster 3 included yam landraces characterized by the highest incidence of anthracnose, severity of anthracnose and viral infection, and lump of pounded yam.

Discussion
Our results showed that, subject to local yam landraces only 44 of the 116 popular white Guinea yam landraces recorded in the study area had a high market value. Yam production in the study area is mainly intended for family consumption [35]. Consequently, farmers cultivate several yam landraces of low market value with the major goal of meeting their varietal   preference criteria. Majority of the popular yam landraces with high market value recorded have early maturity and are known for their large tubers. Indeed, Banson and Danso [36] have shown that the price per kilogram of yam increased with the size of the tuber. The majority of the late yam landraces identified with high market value were mainly found in Donga department, which is the preferred area for producing yam chips for which these landraces are known for [23]. In general, farmers perceived most of the identified popular yam landraces with high market value as having high agronomic and culinary performances. Indeed, according to Zannou et al. [13], the price of different yam landraces in the market reflected their technological or taste characteristics. In addition, previous studies reported that, in general, yam landraces grown on a large scale by many households in a region tended to have better agronomic performance [8,23]. Knowing that, in many cases synonymy exists among cultivated yam landraces in the Republic of Benin [3,37,38], the identified duplicates among this popular yam landraces with high market values by Agre et al. [39] is useful for their better conservation and use.
Our results showed that, the yield of the popular yam landraces with high market value varied significantly from one mound to another. This variation of yam yield could be due to an important interaction between the environment and the expression of their phenotypic potential [40]. In addition, it is known that yam yield is highly sensitive to climatic and pedological parameters such as precipitation, temperature, light, photoperiod and soil type [41]. Indeed, the rate of accumulation of photosynthetic reserves depends not only on the available vegetative mass [42], but also on geophysical conditions [40]. Multi-local tests should be carried out to identify growing-areas with optimal production conditions for each recorded landraces to boost yam yields in Republic of Benin. Most of popular yam landraces with high market value recorded in this study showed poor flowering and fructification capacities. Indeed, it is known that natural pollination of white Guinea yam occurs very rarely in the field [43,44], and this species has a very poor potential https://doi.org/10.1371/journal.pone.0273043.g004 Table 9. Mean values of yield and yield components, and median values from scores of qualitative variables of popular yam landraces within clusters derived from the hierarchical cluster analysis (median values are in the square brackets).

Quantitative traits Mean [Median]
for flowering and fructification [45]. Landraces such as Danwari, Kodjewe, Mondji, and Gnidou, which presented a high flowering intensity with high fructification capacity, could be used in further yam genetic improvement.
Our results have revealed that the identified popular yam landraces with high market value in the Republic of Benin offer enormous potential for yam breeding programs. In fact, the recorded yam landraces that showed tolerance to diseases and nematodes, and had good agronomic and culinary characteristics should be prioritized for demonstration and use in yam breeding programs [1,46]. Laboko and Djiladja landraces, which have organoleptic characteristics that satisfy consumer preferences could be used as outstanding varieties in yam breeding programs. In fact, Fakorede et al. [20] identified these landraces as satisfying the preferences of Beninese consumers and processors and are among the best parents for white Guinea yam breeding programs.
Our results suggested that a minimum yield of 4.16 ± 0.15 kg/mound, and tubers of minimum length and width of 36.41 ± 1.22 cm, and 25.44 ± 1.16 cm respectively must be the minimum standards that an improved variety of yam must have to be accepted by Beninese farmers. Furthermore, in addition to resistance to various biotic and abiotic stresses the improved variety should reveal minimum scores of 3.16 for texture, 0.75 for softness, 3.75 for elasticity, and 1.34 for colour preference during sensory evaluation to be acceptable by consumers. In the framework of a varietal improvement program for early maturity yam landraces, the MFA showed that landraces in cluster 1 would be good parents while the white Guinea yam landraces in cluster 2 would be ideal parents for a late-maturing yam for a breeding program.

Conclusions
Our study identified 44 popular white Guinea yam landraces with high market value and agronomic and culinary performance that could be very useful for market-oriented breeding programs. Among them, Danwari, Kodjewe, Mondji, and Gnidou landraces should be integrated in the national breeding programs because of their high flowering intensity, and fruit setting capacity. Amoula, Laboko, and Djilaadja landraces should be considered the standard landraces for high yield, good sensory attributes, and resistance to diseases and nematodes, respectively. This study allowed to identify basic breeding standards, which used by breeders could increase the acceptability of improved white Guinea yam varieties by farmers and consumers in the Republic of Benin.