Multivariate Analysis and Its Application for Characterizing Cowpea Landraces

Cowpea [Vigna unguiculata (L.) Walpers] is widely cultivated legume in tropics and sub tropics especially Africa, Asia, Central and South America (Muchero et al., 2013). It contributes to food and nutritional security, fodder, soil fertility, income for farmers and food vendors (Fatokun et al., 2018). According to FAO 5.8 million tons of dry cowpea cereal is produced annually from 11 million hectares planted all over the world.
       
Cowpea is a composition of carbohydrates (63%), proteins (25%), fat (1.5%), vitamins, minerals, folate, thiamin and riboflavin (Xiong et al., 2016). In Pakistan it is cultivated in Khyber Pakhtun Khaw and Punjab on an area of 16.9 thousand hectares with an annual production of 7.8 thousand metric tons (Iqbal et al., 2017). Genetic resources conservation and preservation is an integral part of socioeconomic activities, its sustainable utilization keep balance of natural resources.
       
Although concentrations are concentrating more on molecular and recombination techniques to bring breakthroughs but facing challenges of rapid rate of biodiversity loss. Efforts are devoted towards preserving available genetic resources in gene banks, herbariums and breeders’ collections. It is imperative to have accurate and better understanding of breeding material to develop desirable or target traits to achieve crop improvement objectives (Singh, 1997). It is reported that in the last few decades, Indonesia and China have lost 1,500 local rice and 651 soybean varieties. Cowpea in Asian countries have a very long domestication history that also poses similar threats (Fang et al., 2007). Therefore, protection, research, development and utilization are extremely important for cowpea breeding.
       
Additionally, major crops like chickpea, maize, rice, soybean and wheat have been genetically eroded due to introduction of new material but crops like cowpea due to negligence (underutilized crops) still have great opportunities and genes of novel significance. Multivariate by means of Cluster analysis approach is an appealing technique for germplasm discrimination. Characters like quality characters, nutritional, mineral, protein or even DNA based distributed efficiently in to various groups. It enhances the utility of different groups on priority or performance basis. Present study aims to unveil genetic variability in current collections and to plan a future methodology for catering genetic diversity benefits.

Cowpea 71 landraces collected from 31 districts were planted at Department of Botany, University of Gujrat, Gujrat, Pakistan. Sowing was accomplished in first week of July in an augmented design while in March 2017, ten single plants were sampled at random, tagged, harvested and threshed. Progenies were sown under field conditions during July 2017 and harvested in March 2018. Two rows of 4 meter length for each plant progeny were planted with 15 cm and 1 m intra and inter-row spacing with a local check repeated after 10 rows. Data were recorded following IPGRI descriptors for Vigna unguiculata (IBPGR, 1983). Selected qualitative traits (plant color, leaf shape, plant type, twining tendency, anthocynin pigment and fodder types) and agronomic traits (plant height, branches, pods, grain yield, biomass and harvest index) were recorded.
       
Protein banding pattern was obtained by following Laemmli (1970) with some modifications. Similarity index was calculated for all possible pairs of protein types (+/-). Variance based on inter and intra variability with average agronomic performance for individual landraces were analysed (Steel and Torrie, 1980). Phylogenetic trees were constructed with the help of computer software “Statistica” version 6.0.

Cowpea phenotypic diversity
 
Qualitative traits showed high variation as landraces were collected across the country ranging 100 to 2200 masl. Kurlovich (1998) emphasized phenotypic traits as plant description and consumers preference. Plant color was categorized into pale green in 17 landraces and dark green in 54. Leaf shape was observed at physiological maturity divided into four descriptors (Table 1). 47 landraces were ovate, 8 lanceolate narrow, 10 lanceolate broad whereas 6 lines were as rhombic. Plants with ovate leaf shape were found to be best for moisture absorption and synthesis of food proteins while lanceolate broad and narrow leaf shape is also a distinguishing trait in cowpea which could have the ability to resist drought and hence suggested for future breeding (Iqbal et al., 2017). Al-Saady et al., (2018) reported significant variation in seed color, seed length, seed width and 100 seed weight in collections of Sultanat of Oman.
 

       
All plants were found to be glabrous i.e. without hairs (100%) which helps in harvest and trample. In current studies erect plant type was recorded in 50 landraces was dominant trait as compared to prostrate (14), pronounced (2) and spreading (5). Erect plant type is preferred for fodder use and during humid conditions to avail more sunlight while prostrate type is preferred for planting under rainfed conditions as it facilitates moisture uptake and its efficient use. Genetic variation was recorded low in twinning tendency and anthocynin pigment, further evaluation with known material might help to improve this one. 33 landraces were selected as fodder types and 38 as non-fodder. Fodder types can be utilized as dual purpose (Fatokun et al., 2018).
 
Diversity in agronomic traits
 
Highest plant height mean±SD 177.50+0.57 cm was recorded in Pk-27011 collected from Rawalpindi, with CV as 0.32%. Whereas lowest plant height was recorded in Pk-27096 (Chitral) as 76.33±2.05 cm with CV 2.69%, this trait showed continuous variation (Table 2). Highest number of branches were recorded in Pk-27036 (Rawalpindi) with mean±SD 12.03±1.13 whereas smallest number of branches were recorded in Pk-27078 (Kasur) with mean±SD 3.90±0.16. This trait also couples with low values for pods/plants, biomass, grain yield and harvest index which showed that number of branches played significant role in biological as well as grain yield of the crop. Pk-27107 Bahawalpur showed best performance for pods per plant (50 pods) followed by Pk-27083 (42 pods) Gujranwala, Pk-27047 Vehari (39 pods) and Pk-27018 Rawalpindi (38 pods), prominently. Harvest index is used as best selection criterion for legumes. Lowest harvest index 3.87% was calculated in Pk-27170 collected from Dir, followed by Pk-27064 (Islamabad) with 4.63% for best one performance for future improvement are suggested (Ajayi and Adesoye, 2013). Doumbia et al., (2013) emphasized selection of discriminating traits between and within local and regional gene banks.
 

       
Four genotypes showed best performance; Pk-27154 (Jhang), Pk-27047 (Vehari), Pk-27029 (Sialkot) and Pk-27107 (Bahawalpur) as in Table 3. Pk-27047 was chooses for four traits plant height, pods, biomass and grain yield per plant whereas Pk-27029 for three traits branches, pods and grain yield. Five landraces were selected on the bases of high yield out of 71, viz., Pk-27029 Sialkot 19.11g, Pk-27040 Rawalpindi 18.98 g, Pk-27038 Rawalpindi 18.98 g, Pk-27107 Bahawalpur 18.93 g, Pk-27083 Gujranwala 18.53 g, are hence recommended for crop improvement. Accessions understudies might also test and trialed at various locations to unveil the hidden potential is therefore suggested (Akhtar et al., 2019; Karigiotidok et al., 2019).
 

 
Multivariate analysis based on qualitative traits
 
Nine clusters were formed, cluster 7 was the largest with 31 landraces that represented Punjab (20), KPK (10) and GB (1) showed dissimilarities out of clusters (Fig 1). Cluster 9 consisted of 14 landraces including 11 Punjab, 2 KPK and 1 from Baluchistan. Whereas in the cluster 1, 1 Okara (Punjab) and in cluster 5, 1 landrace from Loralai (Baluchistan), showed variability for qualitative traits. Pigeon peas and cowpeas are known to grown at low elevations while common beans, are adapted to higher elevations, indicates that genetic material is introduced from foothills of KPK to high mountains of KPK and GB (Iqbal et al., 2003b). In another study Xiong et al., (2016) also reported no clustering based on origin of country, however, it seems difficult to separate ecogeographic factors in plants with respect to genetic diversity.
 

 
Multivariate analysis based on quantitative traits
 
Out of 8 clusters largest was C-5, 34 including 1 Islamabad, 16 Punjab, 15 KPK, 1 Baluchistan and 1 from GB showed similarities and harmony (Fig 2). This cluster represented almost all provinces except Sindh and AJK. C-4, 2 landraces both from Kasur (Punjab) with unique characteristics. C-8 was the interesting one, have one landrace of KPK whereas others all were from Punjab, rich cultivated areas hub of the most of the primitive land races of different species. Though cluster analysis grouped together landraces with greater morphological similarity, the cluster did not necessarily include all of the landraces from the same or nearby sites. It is evident that diverse landraces could tolerate stresses as compared to landraces with narrow genetic base  (Kouam et al., 2018). Cowpea, locally known as dal lobia and safaid lobia is nutritious, local germplasm has ability to cater nutritious requirements of the increasing population therefore, it is suggested to explore more remote areas across country.
 

 
Multivariate analysis of protein peptides
 
Eleven clusters were formed based on eleven protein peptide bands ranging from 0.6 Kd to 66.0 Kd, cluster 8 was the largest consisted of 13 landraces, in which check (Narowal) present along with 1 landrace from Islamabad, Chitral (2), Dir (3), Rawalpindi (3) and one each from other districts (Fig 3). Low level of variability in banding pattern might be due to lesser representation of material from Islamabad, Gilgit Baltistan and Baluchistan (Alghamdi et al., (2019). SDS-PAGE showed a low level of diversity in our accessions, but higher than in other legumes and it can be increased by the addition of more extensively sourced germplasm (Iqbal et al., 2003a). Pk-27082 (Kasur) was present in separate clusters in both total seed protein peptide banding pattern and quantitative traits basis. Favoring current findings, Fatokun et al., (2018) also elaborated that IITA’s cowpea collection did not necessarily grouped in the same cluster according to countries of origin, geographically with agronomic and botanical descriptors (Xiong et al., 2016; Manjesh et al., 2019).
 

In conclusion four accessions Pk-27029, Pk-27047, Pk-27107 and Pk-27154 showed best performance are therefore recommended for crop improvement.