Bivariate analysis of the genetic variability among some accessions of African Yam Bean ( Sphenostylis stenocarpa ( Hochst ex A . Rich ) Harms )

Variability is an important factor to consider in crop improvement programmes. This study was conducted in two years to assess genetic variability and determine relationship between seed yield, its components and tuber production characters among twelve accessions of African yam bean. Data collected were subjected to combined analysis of variance (ANOVA), Principal Component Analysis (PCA), hierarchical and K-means clustering analyses. Results obtained revealed that genotype by year (G × Y) interaction had significant effects on some of variables measured (days to first flowering, days to 50 % flowering, number of pod per plant, pod length, seed yield and tuber yield per plant) in this study.The first five principal components (PC) with Eigen values greater than 1.0 accounted for about 66.70 % of the total variation, where PC1 and PC 2 accounted for 39.48 % of variation and were associated with seed and tuber yield variables. Three heterotic groups were clearly delineated among genotypes with accessions AY03 and AY10 identified for high seed yield and tuber yield respectively. Nonsignificant relationship that existed between tuber and seed yield per plant of these accessions was recommended for further test in various agro-ecologies for their suitability, adaptability and possible exploitation of heterosis to further improve the accessions.


INTRODUCTION
Global food security is being threatened with increasing dependence on a few major staple crops and this has resulted in an alarming reduction in crop diversity and variability (Ikhajiagbe and Mensah, 2012).African yam bean (Sphenostylis stenocarpa) (AYB) is one of the under-utilized legumes grown in Nigeria, Central African Republic, Gabon, Zaire and Ethiopia.It is used as food or food components and provides two consumable products; the tuber which grows as the root sink and the actual yam beans which develop in pods above ground (Olasoji et al., 2011).AYB performs better when intercropped than when grown as sole crop (Adeniyan et al., 2007).Grain yield of about 248 -4,130.46kg/ha has been reported for this crop (Adewale, 2011).Like other grain legumes, AYB is an excellent food, low in fats and rich in protein, carbohydrate, fibre, minerals and vitamins (Akande et al., 2012).
Information on the genetic variability among genotypes and its potential use is vital in the crop genetic improvement.This will enhance the identification of useful genes and their behaviour in breeding programmes (Ogunniyan and Olakojo, 2014).However, numerous techniques have been proposed to estimate genetic divergence in crop accessions.Among such techniques are principal component analysis (PCA) proposed by Gower, 1966) and numerical taxonomic (Sneath, and Sokal, 1973).PCA is useful for reducing and interpreting large multivariate data sets with underlying linear structures and for discovering previously unsuspected relationships among crop characters (Ogunniyan and Olakojo, 2014).
Research has been carried out on AYB concerning morphological characteristics of the crop (Akande, 2009;Popoola et al., 2011); genetic variability in its seed quality (Olasoji et al., 2011); nutritive and antinutritive factors in the beans (Ajibade et al., 2005).However, there is little baseline information on the relationships between seed yield, tuber production and their components coupled with its improvement.Also, presently, there is no known released variety of the crop in Nigeria, it is only found in the hands of old farmers (Saka et al.,2004;Adewale et al., 2012;Ojuederie et al., 2014), thus make it vulnerable to extinction.Therefore, in order to accelerate the general acceptance of AYB as one of the means to proffer solution to the problem of malnutrition among children and nursing mothers in Sub-Saharan Africa, there is need to genetically improve AYB.The objectives of this research are therefore to assess genetic variability among twelve accessions of AYB from the forest agro-ecology of Nigeria using Principal Component Analysis (PCA) and also, determine the relationship among seed yield, its components and tuber production characters of the AYB accessions.

MATERIALS AND METHODS
Twelve accessions of African yam bean (AYB) coded as AY01, AY02, AY03, AY04, AY05, AY06, AY07, AY08, AY09, AY10, AY11 and AY12 that were previously collected from six states of Nigeria (Figure 1) were obtained from the germplasm collection of the  Twenty (20) morphological characters namely; days to first flowering, days to 50 % flowering, terminal leaf length (cm), terminal leaf width (cm), petiole length (cm), number of seeds per pod, number of pod per peduncle, number of pod per plant, pod mass per plant (g), pod length (cm), seed yield per plant (g), 100 seed yield per plant (g), days to physiological maturity, seed length (cm), seed width (cm), seed thickness (cm), tuber yield per plant (g), tuber smallest mass (g), tuber biggest mass (g), and number of tubers per plant were collected based on AYB descriptors (Adewale and Dumet, 2010).
Data obtained were subjected to combined analyses of variance (ANOVA).Difference between the treatments were separated using Duncan Multiple Range Test (DMRT) at 5 % or 1 % levels of significance.Principal component analysis was carried out and components with Eigen values > 1.0 were considered.Contributing characters with values > 0.6 were considered relevant for principal components (Matus et al., 1999).
Accessions were clustered into groups based on hierarchical clustering using squared Euclidean distance and k-means clustering analysis was used to evaluate character contribution to clustering groups.Pearson's moment correlations between morpho-agronomic characters and tuber yield characters were determined using IBM SPSS (Version 23.0, 2015).

Analysis of variance (ANOVA) for morphoagronomic variables of AYB accessions
Combined analysis of variance (ANOVA) for morphoagronomic variables of the 12 AYB accessions evaluated in 2015 and 2016 are presented in Table 1.
The results of the combined analysis revealed that genotype exhibited significant differences in all of the variables except terminal leaflet length, petiole length and number of seeds per pod.There were significant differences in some of the variables based on years of evaluation while results of variables such as terminal leaf length, terminal leaf width, petiole length, number of seeds per pod, number of pod per peduncle, 100 seed mass, seed length and seed width were not significantly different from each other in both years (across the years).Also, genotype by year (G ×Y) interaction had significant effects on some of the variables (days to first flowering, days to 50 % flowering, number of pod per plant, pod length, seed yield per plant and tuber yield per plant) in this study.2d).

Mean performance of AYB accessions based on morpho-agronomic variables across the year of evaluation
Significant genotypic differences were observed among the genotypes evaluated across the years; days to first flowering varied from 131 to 149 days in accessions (AY07, AY08) and AY12 respectively with mean of 139 days after planting.Accessions AY08 and AY12 attained 50 % flowering between 134 to 157 days respectively with mean of 147 days (Table 2a).AY01 and AY03 had higher seed yield per plant with values of 46.32 g and 47.50 g respectively compared to the other accessions.Number of pod per plant ranged from 4 to 21 with mean of 12. Pod mass among these accessions with mean of 41.63 g ranged from 18.86 g (AY12) to 69.26 g (AY03).AY05 had the longest pods (20.36 cm) while the shortest pod (12.98 cm) was obtained in AY08 with long and recorded mean value of 16.23 cm (Table 2b).Also, AY03 and AY12 attained physiological maturity between 197 to 210 days with mean value of 203 days (Table 2c).All of the AYB accessions produced tubers at maturity except AY11 and AY12 (Table 2d, Figure 2).Accessions AY10 and AY02 produced the highest average tuber yield per plant (150.28 g and 149.47 g respectively) (Table 2d).A high level of variability existed in the tuber characters of the accessions evaluated as reflected by coefficient of variation (CV) with the highest variability found in the smallest tuber masses, followed by tuber yield per plant, the biggest tubers mass per plant, seed yield per plant and pod mass per plant while the least variability was found in days to physiological maturity.Means with the same letter(s) in the same column or row are not significantly different from each other at p = 0.05; ACC: Accessions; CV: coefficient of variation Means with the same letter(s) in the same column or row are not significantly different from each other at p = 0.05; ACC: Accessions; CV: coefficient of variation Means with the same letter(s) in the same column or row are not significantly different from each other at p = 0.05; ACC: Accessions; CV: coefficient of variation Means with the same letter(s) in the same column or row are not significantly different from each other at p = 0.05; ACC: Accessions; CV: coefficient of variation

Principal Component Analysis (PCA)
The results of PCA of morpho-agronomic variables of the accessions revealed that five component axes had Eigen values that are greater than 1.0 and accounted for 66.70 % of the total variation.Relative discriminating power of the PCA as revealed by Eigen value was 4.96, 2.94, 2.17, 1.88 and 1.40 for PC1, PC2, PC3, PC3 and PC5 respectively (Table 3).PC 1 accounted for 24.79 % of the variation and is associated with days to first flowering, days to50 % flowering, petiole length, number of pod per peduncle, number of pod per plant, pod mass per plant, and seed yield per plant while the second principal component (PC2) was responsible for 14.69 % of the variation and was mainly associated with tuber yield per plant, seed width, seed thickness, days to physiological maturity and terminal leaflet length.The third principal component (PC3) accounted for 10.85 % of the variation and was mainly associated with tuber variables (tuber yield per plant, tuber smallest and biggest mass, and number of tuber per plant).The fourth and fifth principal components explained 9.38 % and 6.99 % of the variation respectively (Table 3).

Clustering of AYB accessions
Accessions were grouped into three clusters at the rescaled distance of 15 units (Fig 3 ); cluster 1 comprised of two accessions (AY11 and AY12).The members are late flowering, low seed and non-tuber producing genotypes.Cluster 2 contained seven accessions (AY03, AY05, AY01, AY04, AY06, AY07 and AY09); the members of this cluster are made up of high seed yield and medium tuber yield.Also, cluster 3 made up of three accessions (AY02, AY10 and AY08) that exhibited early to late flowering and with highest tuber yield (Fig. 3).Inter-cluster distance revealed that AYB clustered in groups 1 and 2 were closely related with the least inter-cluster distance of 65.97 units while groups 1 and 3 were distantly related with inter-cluster distance of 117.80 units (Table 4).Five AYB accessions initially sourced from Ondo state were found in groups 2 and 3 while the three AYB accessions initially sourced from Oyo state were found spread in groups 1and 2 while other accessions were one per state.

Correlation between seed yield and tuber production variables
Correlation between pairs of seed yield variable and tuber production variables showed that some pairs of the variables evaluated were significant at 5 % probability level or highly significant at 1 % probability level (Table 5).Tuber yield per plant was positively and significantly correlated with seed width (0.26), seed thickness (0.38) and terminal leaflet length (0.32).Nonsignificant relationship with low magnitude of Pearson coefficient of correlation was obtained between tuber yield per plant and seed yield per plant (0.07).On the other hand, tuber yield per plant was inversely or negatively correlated with days to first flowering (-0.11), days to 50 % flowering (-0.09), number of seed per pod (-0.11), pod length (-0.15) (Table 5).Also, seed yield was positively and significantly correlated with number of pods per plant (0.84), pod mass per plant (0.84), number of seeds per pod (0.35) and pod length (0.27).Similarly, positive and significant association existed in tuber yield per plant with the biggest tuber mass (0.65), the smallest tuber mass (0.36) and number of tubers per plant (0.67) (Table 5).*Means found significant at 5 %, ** significant at 1 % significant level A: first flowing; B: days to 50 % flowering; C: terminal leaf length (cm); D: terminal leaf width (cm); E: petiole length (cm); F: number of seeds per pod; G: number of pod per peduncle; H: number of pod per plant; I: pod mass per plant (g); J: pod length (cm); K: seed yield per plant (g); L: 100 seed yield per plant (g); M: days to physiological maturity; N: seed length (cm); O: seed width (cm); P: seed thickness (cm); Q: tuber yield per plant (g); R: the smallest tuber mass (g); S: the biggest tuber mass (g); T: number of tubers per plant

DISCUSSION
Genetic variation is a prerequisite for successful crop improvement programme.Knowledge of genetic variation and relationships between accessions or genotypes is important to understand the available variability and its potential use in breeding programs (Yoseph et al., 2005).If the genetic variability of any crop specie is carefully exploited, same is capable of providing critical bases for improving the grain yield as well as other economic and important traits (Mahmood et al., 2003;Inamullah et al., 2006).
Africa yam bean has dual crop advantage because it produces both seeds and tubers; however, tuber formation seems to be dependent on the genetic makeup of the accessions.In this study, ten accessions produced tubers while two did not produce tubers.Significant difference among the accessions is an indication of the existence of genetic variations.Also, significant difference in the results of each year as well as their interaction in some of the traits, implies that the performances of the AYB accessions might be influenced by environmental factors and this would enable selection for the traits of interest in different agro-ecologies.Nwofia et al. (2014) had reported that genotype and year of evaluation had significant effects on some of the agronomic traits of African yam bean.
Hierarchical clustering based on twenty morphoagronomic variables evaluated in this study, clustered the AYB accessions into three groups, where genotypes within the same cluster exhibit high homogeneity and high heterogeneity between the clusters.Also, the result obtained from the PCA showed that PC1 and PC2 accounted for 39.48 % of the variation, which was associated with seed and tuber yield variables.This indicates that days to first flowering, days to 50 % flowering, petiole length, number of pods per plant, pod mass per plant, seed yield per plant, seed width and tuber yield per plant were considered as major contributors to the total variation having PC values > 0.6 (Matus et al.,1999) in this study.These identified variables could facilitate effective selection in AYB improvement programmes.
Although, significant differences were obtained among the AYB accessions for some of the traits, 74.99 % of the accessions were grouped in two clusters of the three clusters formed by K-means analysis.These two clusters were the most closely related with the least inter-cluster distance of 65.97 units.These results justify the low genetic diversity observed among the African yam bean accessions in forest agro-ecology of Nigeria which needs to be improved.Akande (2007) had anticipated that through mutation breeding, introduction, recombination and selection, improved varieties which are more productive than those currently grown by farmers can be developed.
Non-significant relationship obtained between tuber yield per plant and seed yield per plant; could be a function of genetic make-up of the AYB accession.Several researchers (Marcelis, 1996;Farrar, 1996;Geigner et al.,1996) had reported that such occurrence could be competitiveness between the storage sinks (seeds and tubers) to import photo-assimilates, which is a function of environmental factors such as water, temperature and strength of the competing sinks.
Identifying the molecular determinant of sinks strength would further increase yield.In addition, it could be observed from the Pearson coefficients of correlation, that number of seeds per pod, number of pods per plant, pods mass per plant, pod length and 100-seed mass greatly influenced seed yield.Also, the smallest tuber mass and the biggest tuber mass contributed to tuber yield in this study; this could be attributed to the positive and significant relationship obtained among these traits with seed yield and tuber yield.Ikhajiagbe et al. (2007) earlier reported that yield characters such as number of pods per plant, number of seeds per pod, 100-seed mass variables strongly influence grain yield of AYB.Hence, these variables could be used for indirect selection for seed yield and tuber yield.Significant positive correlations between yield and other agronomic characters that can improve yield are quite desirable in plant breeding, because it facilitates selection process and gains from selection (Bello and Olawuyi, 2015).Therefore, efforts should be geared towards improving them for better yield.

CONCLUSION
This study revealed that some of the traits of African yam bean genotypes like seed yield and tuber yield were significantly influenced by prevailing environmental factors across the years of evaluation.For instance, environmental factors favoured high seed yield in 2016 and high tuber yield in 2015.Also, significant genotypic differences were obtained among the genotypes evaluated across the years; AY03 and AY01 had the highest seed yield while AY10 and AY02 gave the highest tuber yield.On the other hand, AY01 and AY03 gave the highest seed yield in 2016 and 2015 respectively while highest tuber yield was obtained in AY02 and AY10 in 2015 and 2016 respectively.Three heterotic groups were clearly delineated among the genotypes with accessions AY03 and AY10 identified for high seed yield and tuber yield respectively.
However, non-significant relationship between tuber yield and seed yield per plant of these accessions, which could be attributed to their genetic make-up can be further tested in various agro-ecologies for their suitability, adaptability and possible exploitation of heterosis to further improve the accessions.

ACKNOWLEDGMENT
The effort and contributions of Late Prof. (Mrs) S.R.
Akande to this research will forever be remembered.She collected and provided most of the African Yam Bean accessions used for this research work.May her gentle soul rest in peace.

Figure 1 :
Figure 1: Locations and distribution of AYB accessions across collection sites

Figure 2 :
Figure 2: Tubers of ten accessions of the African yam bean (AYB) harvested in this study

Figure 3 :
Figure 3: Dendrogram cluster of the 12 African yam bean accessions evaluated based on hierarchical clustering using squared Euclidean distance show the three clusters formed at the rescaled distance of 15 units

Table 1 :
Combined analysis of variance (ANOVA) of AYB accessions based on morpho-agronomic characters evaluated in 2015 and 2016

.2 Effect of years of evaluation on the performance of AYB accessions AYB
accessions flowered earlier in 2016 (134 days after planting) than in 2015 (144 DAP) and attained 50 % flowering with mean values of 142 and 152 DAP in 2016 and 2015 respectively (Table 2a).Higher seed yield per plant and pod mass were recorded in 2016 (34.10 g and 50.29 g) than in 2015 (21.76 g and 32.96 g) respectively while 2015 significantly favoured tubers production with tuber yield per plant mean value of 102.36 g than in 2016 (46.03 g) (Table 2b; 2c; 2d).obtained in AY08 (8.26 g) and AY12 (14.65 g) in 2015 and 2016 respectively (Table 2b; 2c).On the other hand, AY02 gave the highest tuber yield per plant (265.58g) in 2015 and AY10 gave the highest tuber yield per plant (150.28 g) in 2016 while AY11 and AY12 did not produce tubers in both years (Table

Table 2a :
Mean values of the AYB accessions for morpho-agronomic characters evaluated in 2015 and 2016

Table 2b :
Mean values of the AYB accessions for morpho-agronomic characters evaluated in 2015 and 2016

Table 2c :
Mean values of the AYB accessions for morpho-agronomic characters evaluated in 2015 and 2016

Table 2d :
Mean values of the AYB accessions for morpho-agronomic characters evaluated in 2015 and 2016

Table 3 :
Characters with respect to its principal component, Eigen values and variation

Table 4 :
Inter-cluster distance, according to K-mean clustering analysi

Table 5 :
Pearson coefficient of correlation between pairs of morpho-agronomic characters