Yam is cultivated mainly for its tuber. It is a major source of carbohydrates and nutrient energy for many people in tropical countries including east and West Africa, the Carribean, South Africa, India and South East Asia  . Ethiopia is the fifth largest yam producing country of Africa and its annual production is estimated at about 1,191,809 metric tons . The crop plays a vital role in local livelihood particularly in densely populated areas of south, southwestern, and western parts of the country  . About nine types of yam have been reported to grow in Ethiopia indicating the diversity of the species in the country .
Yams have become an important cash crop in most localities. Yams are also served during the traditional festival which coincides with the peak harvesting time thus allowing farmers to earn profit from the market. It grows in the altitude range of 1140 to 2200 and in a wide range of soils mainly in clay, clay loam, sandy and sandy loam types. It is planted in October (in most parts of South Ethiopia), November and December (in South Western and Western part of the country) . The present work is to identify Ethiopian yam collections from major growing areas of the country into their Dioscorea species group and study their phenotypic diversity.
2. Materials and Methods
2.1. Morphological Characterization and Yam Landraces
Sixty yam accessions from ten different geographic origins of Southern Nations Nationalities and People’s Regional State and Oromiya region of the country were used for this study. The sprouted tubers of some accessions were obtained from Research Centers whereas others were directly collected from farmers’ fields during early March, 2010 as indicated in Table 1. The samples collected were planted on April 2010 in complete randomized block design with three replication at two research sites (namely at Hawassa and Wonago) which are found under South Agricultural Research Institute at the end of April 2010 as indicated in Figure 1. All important cultural practices such as staking, weeding and irrigation were done starting from planting till harvesting.
Among 45 morphological characters (Table 3) observed in the study presence and absence of spines on stems and roots, number of male and female inflorescence, stem length, twining direction, and flesh color were the major traits considered for species identification. The species identification was done through observing some earlier developed identification keys .
Table 1. Ethiopian yam germplasms with their origin site, vernacular name and respective altitudes used in the study.
Figure 1. General appearance of the field experiment of yam (Dioscorea spp.) at Hawassa Agricultural Research Center.
The identified Dioscorea species were subjected to qualitative morphological characterization. The qualitative morphological characterizations were carried out to determine the level of variability and the relationship among accessions using IPGRI/IITA descriptors . Characters were observed for five different healthy plants per accession. The color aspect of the plant was recorded using Munsell color chart for plant tissues. The farmers perception were taken at the time of collection and sensory evaluation during harvesting for both boiled (eight characters) and un-boiled yam tubers.
The total forty five morphological characters were recorded at the time of young, flowering, maturity stage of the plant and at the time of harvesting using IPGR descriptor list  (Table 2 and Table 3). The collected raw data were subjected to statistical analysis using computer software packages for the diversity and relatedness study among yams (Dioscorea spp.).
Table 2. The qualitative morphological characters and their score code of young and mature leaf, stem and tuber of yam (Dioscorea spp.) germplasms collected during field experiment.
2.2. Characters Evaluated by Farmers
Fifteen farmers were selected from Gedio zone (Wonago district) near the research site. Following their consent an introduction was given to them by the researcher on the sensory parameters of interest. Yam tubers from each accession were taken washed and boiled by farmers as their method of preparation. The sensory values decided by farmers were recorded by the researcher as indicated in Table 3 using IPGRI descriptor list.
Table 3. Characters assessed by farmers.
2.3. Statistical Analysis for Morphological Characterization
Statistical analysis was done using Numerical Taxonomy and Multivariate Analysis System Version 2.02 NTSYpc software program  and the data reduction function of SPSS for windows (version 12.0, 2003). The standardized data for qualitative characters were subjected to multivariate analysis and principal component analysis to identify the most discriminating morphological characters. The matrix of similarity was generated based on distance coeficents. The distance matrix was subjected to hierarchical cluster analysis using Shan through an un-weighted pair group method average (UPGMA) . Cophenetic value (ultrametric) matrix was used to compute the cophenetic correlation as a measure of goodness of fit.
3.1. Morphological Variability Based Species Identification
Based on presence and absence of spines on stems and roots, number of male and female inflorescence, stem length, twining direction, and flesh color sixty yam germplams from ten major growing areas of the country were grouped into six species namely D. alata, D. bulbifera, D. abyssinica, D. praehensilis, D. rotundata and D. cayenensis. The two species D. rotundata and D. cayenensis were observed to have 1 - 3 male inflorescences per spike and after cutting revealed white and yelowish flesh color. During this study the yellowish colour was grouped to D. cayenensis whereas the white flesh colour was grouped to D. rotundata as indicated in Figure 2(b) and Figure 2(d) respectively.
The species D. alata was differentiated from other groups based on its four angled stem. The average stem length of this species under study were 10 m long, it had different tuber shape apperance (Figure 2(f)). Whereas D. abyssinica had an average climber stem of 2 - 5 m and the number of male infloresences observed was 3 - 6. Under this study most of the D. abyssinica species had male infloresecence (Figure 2(a)). Spines were observed on stem of this species. Whereas, the average stem length recorded from D. praehensilis was 10 m and spines were observed both on roots and stems of this species. The number of male and female inflorescences observed in D. praehensilis was 3 - 5 and 1 - 2 respectively (Figure 2(e)). Generaly D. abyssinica and D. praehensilis revealed three types of flesh color which were purplish colour, a central white colour surrounded by purplish colour and the mixure of white and purplish colours (Figure 2(a) and Figure 2(e)).
The species whose stem was climbing to the left or in a clock wise direction was grouped in to D. bulbifera (Figure 2 (c)). The average stem length of this species was 3 - 10 m with 3 - 5 male inflorescence. Its areal tuber showed both white and purple color. While, all the other species under study twining direction was anti clock wise.
(a) (b) (c) (d) (e) (f)
Figure 2. Morphological variations (Male and female inflorescences, fruits, twining habit and tuber flesh color) among six yam (Dioscorea spps.) germplasm observed in this study. (a) Collection area: Jima (South west Ethiopia) Altitude: 1753 masl, Species: D. abyssinica; (b) Collection area: Wonago (South Ethiopia) Altitude: 1940 masl, Species: D. rotundata (White yam); (c) Collection area: Wolega (Western Ethiopia) Vernacular name: Dhokunma, Species: D. bulbifera; (d) Collection: Kefa (South West Ethiopia), Species: D. cayenensis; (e) Collection area: Gamogofa (South Ethiopia) Altitude: −1140, Vernacular name: Hatiya-2, Species: D. praehensilis; (f) Collection area: Jima (South West Ethiopia), Species: D. alata.
3.2. Morphological Diversity Data of Yam Determined Based on Cluster Analysis
Among 45 qualitative morphological characters recorded on six Dioscorea species namely D. abyssinica, D. praeihensilis, D. cayenensis, D. rotundata, D. alata and D. bulbifera 12 traits did not reveal phenotypic variation among the 60 yam germplasm studied. These traits were not included for clustering and principal component analysis. Based on the relative magnitude of distance similarity matrix and dendrogram using Shan UPGMA cluster analysis, all sixty yam genotypes included in the study were grouped into five clusters (Figure 3).
At the similarity distance 1.81, the dendrogram identified three clusters 1 and 2 which contain eight to eighteen accessions per cluster (Table 4 and Table 5). Cluster 1 is the second largest group and contains 18 yam accessions with green mature petiole colour. Cluster 2 contained accessions with pale purple young petiole colour, green mature petiole colour, high leaf density, presence of cracks on tuber surface, intermediate corm size, highly coloured tuber after cooking and cooked tuber sweetness.
The dendrogram at the similarity distance 1.6 identified clusters 3. This cluster group is characterized by dark brown stem color after emergency, green mature petiole color, intermediate corm size and presence of cracks on the cooked tuber.
The dendrogram with similarity distance 1.71 identifies one cluster group 4. This cluster is the largest and composed of 19 accessions. This group is characterized
Table 4. Clustering groups of six different Ethiopian Dioscorea species derived from distance similarity coefficient based on their morphological diversity data.
by yellow mature petiole color, green stem color and absence of cracks on the tuber. Cluster 5 contained two cluster groups cluster 5a and cluster 5b. Cluster 5a is a group of D. bulbifera species with yellow mature petiole color and green stem color. Cluster 5b is a group of D. rotundata composed from three yam accessions. This group is characterized by absence of barky patches on the stem, green stem color, and yellow mature petiole color, small tuber width at middle part, absence of tuber cracks, late maturity group and white tuber color after cooking. In this analysis the cophenetic correlation coefficient result was r = 80 which revealed the efficiency of the dendrogram.
Figure 3. Relationship of six yams (Dioscorea spp.) species collected from ten major growing areas of Ethiopia based on distance UPGMA hierarchical cluster analysis.
Table 5. Clustering pattern of yam (Dioscorea spp.) accessions derived from the similarity matrix of 60 accessions based on their morphological traits.
Note: Ji-Jima, Ga-Gamogofa, , Da-Dauro, Wo-Wolayita, Si-Sidama, Ar-Areka, Km-Kembata, Ge-Gedio, Wg-Wolega, Kf-Kefa.
3.3. Principal Component Analysis
The PCA results revealed that the first axis largely accounted for the variation among yam accessions (24.87%) followed by the second axis (14.96%) and third axis (11.63%). The first five axes with eighteen values greater than unity accounted for the total variations among 33 characters describing 69.31% and the first three axes accounted for 51.48% (Table 6 and Table 7). Traits with comparatively greater weight in PC1 and PC3 were young petiole color. Correspondingly greater weight in PC2 was color of tuber after cooking. Similarly, flesh color at the lower part of tuber and flavor of cooked tuber resulted in larger weight in the first three principal components while tendency of tuber to branch gave greater weight to the first two principal components.
Table 6. Eigen values, variability percentage and accumulated variation with respect to five character in 60 yam accessions.
Note: SN = Serial number.
Table 7. First 3 principal components (PCs) scores of 33 morphological traits across yam genotypes collected from 11 major growing areas of Ethiopia.
Note: SN = Serial number; *Bolded values revealed highly correlated morphological traits to respective principal components.
Morphological characterization is advisable to be taken as the first step before biochemical or molecular studies . In the present study, tuber flesh colour and tuber shape revealed differences among different species (Figure 2 and Figure 4). Similarly, it is stated that tubers vary greatly in size, colour and shape depending on species, cultivar and environment   . In this study, yam tubers of D. praehensilis and D. alata had finger like appearance while tubers of D. rotundata, D. cayenensis and D. abyssinica were long with oval shape (Figure 4). It is also reported that the tuber shape of D. alata is often irregular which increases harvest time and thus is labour intensive .
In the present study pink and red purplish colors were observed in D. abyssinica and D. praehensilis, whereas D. cayenensis and D. alata revealed yellow and white tuber flesh colors respectively. It is indicated that anthocyanins and carotenoids are pigments found in yams (Dioscorea spp.) which give their distinctive tuber flesh colors . B-carotene and Xanthophyll esters are responsible for the yellow flesh color of D. cayenensis.
(a-1) (a-2) (a-3) (a-4) (a-5) (b-1) (b-2) (c) (d-1) (d-2) (e-1) (e-2) (e-3)
Figure 4. Tuber shape similarities and differences among six different species of yam (Dioscorea spp.) germplasm with their respective collection area (vernacular name and altitude). (a-1) Wolayita (Chocha, 1850 masl), (a-2) Gedio (Gedio), (a-3) Sidama (Gellawcho, 1940 masl), (a-4) Sidama (Adame Ado) and (a-5) Jima Species: D. rotunda; (b-1) Jima (Kembata, 1753 masl); (b-2) (1630 masl), Species: D. abyssinica. (c) Gedio (Feres kotae); Areal tuber of species: D. bulbifera; (3d-1) Wolayita and; (d-2) Jima, Species: D. alata; (e-1) Gamogofa (Bunne 2, 1655); (e-2) Awasa (Bunne 1); (e-3) Gamogofa (Bunne 1), Species: D. perhensilies.
Morphological characterization indicated only one cluster which grouped yam accessions based on their geographic origin where most accessions of Wolega origin were in cluster 3 (Table 5). It is supported and recommended that geographic diversity may serve as an indicator of genetic diversity in parental selection . Yam genotypes of Wolayita origin showed more morphological diversity and were distributed in all clustering groups (Table 5). Earlier studies showed a presence of more morphological diversity in Wolayita yam cultivars . Species of D. abyssinica, D. rotundata, D. cayenensis and D. praehensilis revealed relatedness. This relationship could be due to the likelihood of D. praehensilis and D. abyssinica being one of the wild parents of several cultivars of D. cayenensis and D. rotundata .
The efficiency of different cluster algorithms can be compared through estimation of the cophenetic correlation coefficient. It is a product moment correlation coefficient measuring agreement between the dissimilarity and similarity indicated by the dendrogram which is output of the dissimilarity and similarity matrix . In this analysis the cophenetic correlation coefficient r = 80 revealed the efficiency of the dendrogram.
In this study, morphological traits that best discriminate between the landraces were young petiole color, flesh color at lower part of tuber, tendency of tuber to branch, color of tuber after cooking and flavor of cooked tuber. In the same way, it was reported that D. alata that scores morphological variability on the first principal component (PC-1) was highly correlated with the characters related to the tuber flesh colours and petiole colour . Similarly, it was indicated in morphological variability study of Kenyan yam that Pc3 was mainly correlated to characters related to the tuber flesh colour . It is also showed that the most effectively used characters in classification of D. alata are tuber, leaf and growth characteristics .
Generally the present study indicated that, the presence of important heritable phenotypic trait of yam to achieve genetic improvement of the crop. Clustering the accessions helped in identifying parents with diverse characters. Most of the morphological variations among the yam genotypes were contributed by young petiole color, tendency of tuber to branch, tuber flesh color at lower part, color of tuber after cooking and flavor of cooked tuber. Hence, these morphological and organoleptic traits can be taken as useful characters for identification of yam cultivars. Morphological characterization of Ethiopian yam accessions showed that Wolayita regions had wide genetic diversity and most yam landraces from Wolega revealed different genetic base. Therefore, it is important to consider yam germplasms from Wolega and Wolayita regions during breeding and improvement activities of the crop.
We would like to thank Rural Capacity Building Project (RCBP) for a financial support during the execution of this experiment. The project was launched by Ethiopia Ministry of Agriculture and Rural Development (MoARD). The major aim of the project was to make clients aware of economically viable and environmentally sustainable practices.
 Yang, D.J., Lu, T.J. and Huang, L.S. (2009) Effect of Endogenous Glycosidase on Stability of Steroidal Saponins in Taiwanese Yam (Dioscorea pseudojaponica yamamoto) during Drying Processes. Food Chemistry, 113, 155-159.
 Food and Agriculture Organization (2015) FAOSTAT Agriculture Database, Agricultural Production, Crops Primary. Yams. Food and Agriculture Organization, Rome.
 Hildebrand, E. (2003) Motives and Opportunities for Domestication: Anthropological Study in Southwest Ethiopia. Journal of Anthropological Archeology, 22, 358-375.
 Muluneh, T. (2006) Assessing Diversity in Yams (Dioscorea spp.) from Ethiopia Based on, Morphology, AFLP Markers and Tuber Quality, and Farmers Management of Landraces. Culvillier Verlag, Göttingen.
 Miège, J. and Sebsebe, D. (1997) Dioscoreaceae. In: Edwards, S., Sebsebe, D. and Hedberg, I., Eds., Flora of Ethiopia and Eriterea, National Herbarium, Biology Department, AAU, Ethiopia, Department of Systematic Botany, Uppsala University, Sweden.
 International Plant Genetic Resource Institute and International Institute for Tropical Agriculture (1997) Description for Yam (Dioscorea spp.). International Institute for Tropical Agriculture, Ibadan, International Plant Genetic Resource Institute, Rome.
 Otoo, E., Akromah, R., Kolesnik, A. and Astedu, R. (2009) Ethno-Botany Morphological Characterization of the Yam Pona Complex in Ghana. African Crop Science Conference Proceedings, Vol.9, Cape Town, 28 September-1 October 2009, 407-414.
 Martin, F. and Ruberté, R. (1976) The Polyphenol of Dioscorea alata (Yam) Tubers Associated with Oxidative Browning. Journal of Agricultural and Food Chemistry, 24, 67-70.
 Lebot, V., Trilles, B., Noyer, J. and Modesto, J. (1998) Genetic Relationship between Dioscorea alata L. Cultivars. Genetic Resources and Crop Evolution, 45, 499-509.
 Hasan, S., Ngadin, A., Shah, R. and Mohamad, N. (2008) Morphological Variability of Greater Yam (Dioscorea alata L.) in Malaysia. Plant Genetic Resources, 6, 52-61.