Genetic variation and genetic advance in cowpea based on yield and yield-related traits

ABSTRACT

The objective of this study was to determine phenotypic variation for yield and yield-related traits among cowpea genotypes and select best candidate genotypes for breeding. Hundred cowpea genotypes were evaluated across two environments using an alpha lattice design with two replications. The pooled data were subjected to analysis of variance, correlation and principal component analyses (PCA). Significant differences were observed among cowpea genotypes for assessed traits. PCA revealed three principal components contributing to 77.75% of the total variation. Grain yield was significantly correlated with most of the traits. The genotypic coefficient of variation was relatively higher, whereas the phenotypic coefficient of variation values were moderate for branch number, seed number per pod, and high for seed number per plant and pod weight per plant. Heritability and genetic advance values respectively ranged from 37.27% to 97.2% and 73.3% to 2242.6% for the studied traits. High direct path coefficient value of 0.71 for pod weight per plant with highly significant correlation with grain yield was observed. The study identified cowpea genotypes such as Glenda RV 465, RV 574, RV 115, RV 28, RV 419, RV 28, RV 419, RV 213, RV 550, RV 470, RV 111, RV 315 and RV 550 with better responses for yield and yield-related traits.

KEYWORDS:

• Cowpea
• genetic advance
• heritability
• grain yield
• PCA

Introduction

Cowpea [Vigna unguiculata (L.) Walp., Fabaceae (2n = 2× = 22)] is an important legume crop in the tropics and sub-tropical regions of the world (Ehlers and Hall 1997; Freire-Filho et al. 2005), and it is often the primary source of protein and minerals for low-income populations in such regions (Gerrano, van Rensburg, Kutu 2019). Most of the world’s cowpea production is produced in sub-Saharan African (SSA) countries such as Nigeria, Burkina Faso, United Republic of Tanzania, Cameroon, Niger, Mali and Kenya (FAOSTAT 2018). In South Africa, the crop is predominantly grown by small-scale farmers under dryland conditions due to its inherent drought tolerance and atmospheric nitrogen fixation (i.e. rhizobium bacteria) which allows it to grow and improve poor soil fertility an essential component in many cropping systems (Fall et al. 2003). Several plant parts including its grains, green pods and leaves are consumed as leafy vegetable, whereas the vines/haulms are used as animal feed (Oyekan 1977). As a result, genetic improvement of the crop through development of agronomically superior genotypes possessing farmer-preferred traits and enhanced nutrition is useful for improving food security in SSA (Gerrano, van Rensburg, Kutu 2019; Gerrano, van Rensburg, Venter, et al. 2019). Development of highly superior and high-performing genotypes can be achieved through exploitation of cowpea genetic resources to identify and select suitable genotypes with desirable phenotypic traits for use in improvement programmes (Desalegne et al. 2016; Lazaridi et al. 2017; Xiong et al. 2018).

Phenotypic diversity analysis using morphological traits has been extensively used to establish genetic relatedness between and within species and for studying variability and correlated traits in cowpea (Ghalmi et al. 2010; Xiong et al. 2018; Gore et al. 2019). For example, Molosiwa et al. (2016) assessed phenotypic variation of cowpea germplasm collected from Botswana and reported phenotypic differences for agronomic traits such as peduncle length, seed width, seed thickness, pods per peduncle, and hundred seed weight. High genetic distance and diversity among cowpea genotypes were assessed by Adewale et al. (2011) using agronomic traits. Abiola and Adesoye (2013) assessed variability in cowpea using phenotypic descriptors and reported large variation among the studied cowpea genotypes for number of seeds per pod, pod length, plant height, peduncle length, seed weight and number of pods per plant. Understanding the level of genetic diversity is essential for effective conservation and utilisation of germplasm resources (Huaqiang et al. 2012). However, morphological traits are influenced by environmental factors, which may result in varying relationship patterns (Smith and Smith 1989; Selvi et al. 2003). Nevertheless, they are highly effective for selection (Magloire 2005). The estimates of heritability help in the selection of elite genotypes from diverse genetic population. Similarly, genetic advance measures the amount of progress that could be expected with selection of a character. High heritability, coupled with high genetic advance, indicates improvement could be made for a character by simple selection (Tefera et al. 2017).

Although, previous studies assessed phenotypic diversity among cowpea genetic resource from sub-Saharan Africa, very few studies have been conducted in South Africa (Gerrano et al. 2015; Gerrano, van Rensburg, Kutu 2019), which motivates the identification, selection and recommendation of few genotypes for direct cultivation and use in improvement programmes (Gerrano et al. 2015; Gerrano, van Rensburg, Kutu 2019). Furthermore, cowpea is currently regarded as underutilised and neglected legume crop in the country mostly cultivated by resource-poor farmers using unimproved landrace varieties. In the country, the grain yield is estimated between 0.25 and 1.0 ton/ha which is much lower than the grain yield potential of 3.0 ton/ha (Asiwe 2009). Hence, an effort to improve cowpea production through breeding of high-yielding genotypes is urgently required to ensure food and nutrition security, especially among poor-households who rely on the cultivation of cowpea and other crops for sustenance. As a result, the available cowpea genetic pool in the country requires further assessment to identify genotypes with desirable agronomic attributes for breeding. In view of the above discussion, the objective of the current study was to estimate variability, heritability and genetic advance for yield and yield-related traits among cowpea genotypes under different environmental conditions of South Africa.

Materials and methods

Plant material

Hundred cowpea genotypes maintained in the gene bank of the Agricultural Research Council-Grain Crops (ARC-GC) in Potchefstroom, North West Province of South Africa, were planted in Potchefstroom (26.7145° S, 27.0970° E) and Roodeplaat in Pretoria (25.6450° S, 28.3579° E), South Africa during 201/18 growing season. The Potchefstroom site is composed of hutton soil, whereas the Roodeplaat site is composed of sandy loam soil. Potchefstroom is located at an altitude of 1344 m above sea level, and the average minimum and maximum temperature are 9.61°C and 25.48°C, respectively, with an average annual total rainfall of 618.88 mm. Roodeplaat is located in Pretoria at an altitude of 1159 m above sea level with an annual rainfall of 573 mm and a mean temperature of 17.3°C. Information of accession names and sources of origin is presented in Table 1.

Table 1. List of cowpea accessions used in the study.

Serial Number	Accession name	Origin/source	Serial Number	Accession name	Origin/source
1	RV 361	South Africa	51	RV 104	South Africa
2	RV 465	South Africa	52	RV 352	South Africa
3	RV 438	South Africa	53	RV 113	South Africa
4	RV 202	South Africa	54	RV 471	South Africa
5	TVU 9596	IITA	55	RV 207	South Africa
6	RV 551	South Africa	56	RV 440	South Africa
7	98D-1399	IITA	57	RV 43	South Africa
8	RV 568	South Africa	58	RV 173	South Africa
9	RV 499	South Africa	59	RV 501	South Africa
10	RV 343	South Africa	60	RV 320	South Africa
11	RV 553	South Africa	61	RV 555	South Africa
12	TVU 14190	IITA	62	RV 84	South Africa
13	RV 470	South Africa	63	RV 342	South Africa
14	98K-503-1	IITA	64	90K-284-2	IITA
15	RV 416	South Africa	65	TVU 3416	IITA
16	83s-911	IITA	66	RV 504	South Africa
17	RV 121	South Africa	67	95K-589-2	IITA
18	RV 16	South Africa	68	RV 204	South Africa
19	RV 487	South Africa	69	RV 28	South Africa
20	RV 558	South Africa	70	RV 351	South Africa
21	RV 382	South Africa	71	RV 441	South Africa
22	RV 157	South Africa	72	RV 500	South Africa
23	RV 329	South Africa	73	ITOOK-1263	IITA
24	RV 457	South Africa	74	RV 446	South Africa
25	97K-499-35	IITA	75	RV 189	South Africa
26	99K-494-6	IITA	76	Dr. Saunders	South Africa
27	RV 419	South Africa	77	RV 403	South Africa
28	CH-47	IITA	78	RV 498	South Africa
29	RV 503	South Africa	79	TVU 12746	IITA
30	RV 10	South Africa	80	PAN 311	South Africa
31	RV 315	South Africa	81	GLENDA	South Africa
32	RV 439	South Africa	82	RV 550	South Africa
33	RV 502	South Africa	83	TVU 12637	IITA
34	RV 409	South Africa	84	RV 443	South Africa
35	Bechuana White	South Africa	85	TVU 9443	IITA
36	RV 341	South Africa	86	RV 554	South Africa
37	RV 497	South Africa	87	RV 411	South Africa
38	RV 344	South Africa	88	RV 161	South Africa
39	RV 414	South Africa	89	RV 321	South Africa
40	RV 111	South Africa	90	RV 115	South Africa
41	RV 386	South Africa	91	RV 213	South Africa
42	RV 505	South Africa	92	RV 417	South Africa
43	TVU 9620	IITA	93	RV 477	South Africa
44	TVU 2095	IITA	94	86D-1010	IITA
45	RV 464	South Africa	95	TVU 13998	IITA
46	RV 353	South Africa	96	RV 194	South Africa
47	RV 205	South Africa	97	TVU 13778	IITA
48	RV 442	South Africa	98	TVU 11986	IITA
49	98K-476-8	IITA	99	RV 165	South Africa
50	RV 574	South Africa	100	RV 126	South Africa

Note: IITA, International Institute of Tropical Agriculture.

Experimental layout and crop management

Genotypes were evaluated under field conditions using an alpha lattice design with two replications. The intra- and inter-row spacing of 100 and 30 cm was maintained, respectively. Each plot consisted of two rows of 4 m length. Pre-emergent herbicide was applied immediately after planting at a rate of 800 ml per 300 L of water. NPK fertiliser (2:3:2) was top dressed just before the flowering stage. Plants were irrigated using sprinklers three times in a week and weeds were removed manually as necessary.

Data collection and analyses

Data on days to fifty percent flowering (DFW), number of branches (BRN), pod number per plant (PNP), pod weight per plant (PWP), hundred seed weight (HSW), seed number per pod (SPD), seed number per plant (SNP), and grain yield (GRY) were recorded on plot basis. Quantitative data were collected using a standard descriptor list for cowpea (IBPGR 1993). Data were subjected to analysis of variance (ANOVA) using GenStat 18th Edition. The replications and blocks were treated as fixed factors whereas genotypes, environment and their interactions were treated as random factors. The linear model used for data analysis was: $Yijkl=\mu +Gi+Ej+Rk\left(j\right)+Bl\left(jk\right)+GEij+\epsilon ijkl$

where Yijkl is the observed value of genotype i in block l and replication k of environment j, μ is the grand mean, Gi is the effect of genotype i, Ej is the environment or location effect, Rk(j) is the effect of replication k in environment j, Bl(jk) is the effect of block l in environment j and replication k, GEij is the interaction effect of genotype i with environment j, and εijkl is the error (residual) effect of genotype i in block l and replication k of environment j.

Correlation and path coefficient analyses

Correlation analysis was performed to describe the pattern of association among the studied agronomic traits using RStudio version 3.2.3. Path coefficient analysis was conducted using a genotypic correlation matrix set up as A = B × C for grain yield. In this matrix, vector ‘A’ represents the genotypic correlation coefficients of grain yield vs. other agronomic traits. In the same vector ‘B’ is the value of genotypic correlation for all possible combinations among the traits and vector ‘C’, the path coefficients. The inverse of matrix B was calculated using the Matrix Inverse function (MINVERSE) of Microsoft Excel 2010. The path coefficients were calculated as the product of vector A and each row of B⁻¹ using the matrix multiplication (MMULT) function of Microsoft Excel. Direct and indirect path coefficients were calculated according to Dewey and Lu (1959). Grain yield was considered as the response variable, whereas other traits as causal variables.

Principal component analysis

Principal component analysis (PCA) based on the correlation matrix was performed in GenStat 18th Edition. The eigenvectors derived from the PCA were used to identify the variables that tend to have a strong relationship with a particular PC. Biplots of the principal component analysis were constructed to identify genotypes with suitable agronomic traits for effective selection and breeding.

Genetic parameters

The genetic parameters (phenotypic, genotypic and error variance) including phenotypic and genotypic coefficient of variation were computed using mean squares of alpha lattice ANOVA model as below. The variance components were estimated according to Burton (1951).$\mathrm{G}\mathrm{e}\mathrm{n}\mathrm{o}\mathrm{t}\mathrm{y}\mathrm{p}\mathrm{i}\mathrm{c}\mathrm{v}\mathrm{a}\mathrm{r}\mathrm{i}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e}\left({\delta }^{2}g\right)=\frac{\mathrm{M}\mathrm{S}\mathrm{g}-\mathrm{M}\mathrm{S}\mathrm{e}}{r}$where MSg is the mean square due to genotypes, MSe is the error mean square and r is the number of replications.$\mathrm{E}\mathrm{n}\mathrm{v}\mathrm{i}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{e}\mathrm{m}\mathrm{n}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{v}\mathrm{a}\mathrm{r}\mathrm{i}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e}\left({\delta }^{2}e\right)=\mathrm{E}\mathrm{r}\mathrm{r}\mathrm{o}\mathrm{r}\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}\mathrm{s}\mathrm{q}\mathrm{u}\mathrm{a}\mathrm{r}\mathrm{e}=MSe$ $\mathrm{P}\mathrm{h}\mathrm{e}\mathrm{n}\mathrm{o}\mathrm{t}\mathrm{y}\mathrm{p}\mathrm{i}\mathrm{c}\mathrm{v}\mathrm{a}\mathrm{r}\mathrm{i}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e}\left({\delta }^{2}p\right)={\delta }^{2}g+{\delta }^{2}e$ $\mathrm{P}\mathrm{h}\mathrm{e}\mathrm{n}\mathrm{o}\mathrm{t}\mathrm{y}\mathrm{p}\mathrm{i}\mathrm{c}\mathrm{c}\mathrm{o}\mathrm{e}\mathrm{f}\mathrm{f}\mathrm{c}\mathrm{i}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{o}\mathrm{f}\mathrm{v}\mathrm{a}\mathrm{r}\mathrm{i}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\left(\mathrm{P}\mathrm{C}\mathrm{V}\right)=\frac{\surd {\delta }^{2}p}{\mathrm{M}\mathrm{e}\mathrm{a}\mathrm{n}}\times 100$ $\mathrm{G}\mathrm{e}\mathrm{n}\mathrm{o}\mathrm{t}\mathrm{y}\mathrm{p}\mathrm{i}\mathrm{c}\mathrm{C}\mathrm{o}\mathrm{e}\mathrm{f}\mathrm{f}\mathrm{i}\mathrm{c}\mathrm{i}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{o}\mathrm{f}\mathrm{v}\mathrm{a}\mathrm{r}\mathrm{i}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\left(\mathrm{G}\mathrm{C}\mathrm{V}\right)=\frac{\sqrt{{\delta }^{2}g}}{\mathrm{M}\mathrm{e}\mathrm{a}\mathrm{n}}\times 100$

where Mean is the grand mean of a trait.

According to Deshmukh et al. (1986), PCV and GCV values greater than 20% are regarded as high, whereas values less than 10% are considered low and values between 10% and 20% considered medium.

Heritability (H²): heritability in broad sense for all traits was computed using the formula given by Falconer and Mackay (1996).$\mathrm{H}\mathrm{e}\mathrm{r}\mathrm{i}\mathrm{t}\mathrm{a}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}\left(H^2\right)=\frac{\delta {}^{2}g}{{\delta }^{2}p}\times 100$where H² is the heritability in broad sense, ${\delta }^{2}p$ is the phenotypic variance and ${\delta }^{2}g$ is the genotypic variance.

Expected genetic advance (GA) in absolute unit (GA) percent of the mean (GAM) assuming selection was estimated at 5% selection (Johnson et al. 1955):$\mathrm{G}\mathrm{A}=K{\delta }_P^2{H}^2$where K is the standardised selection differential at 5% selection intensity (k = 2.06), δ² _p is the phenotypic standard deviation on mean basis and H² is the heritability in broad sense.

Genetic advance as a percentage of the mean was calculated to compare the extent of the predicted advance of different traits under selection using the following formula:$\mathrm{G}\mathrm{A}\mathrm{M}=\frac{\mathrm{G}\mathrm{A}}{\mathrm{M}\mathrm{e}\mathrm{a}\mathrm{n}}\times 100$where GAM is the genetic advance as a percentage of the mean, GA is the genetic advance under selection and Mean is the mean of the population in which selection employed.

Results and discussion

Effect of genotype, environment and genotype-by-environment interaction on agronomic traits.

Analysis of variance revealed significant (p < 0.001) differences among genotypes for studied traits such as number of branches per plant, number of pods per plant, pod weight per plant, hundred seed weight, seed number per pod, and grain yield (Table 2), indicating genetic variability in the characters. Similarly, the environment showed significant (p < 0.05) differences for all the traits except for seed number per plant. Significant (p < 0.05) genotype-by-environment (G × E) interaction effect was observed for number of branches per plant, pod number per plant, days to flowering which implies diverse nature of the genotypes under diverse environments. Pod weight per plant, hundred seed weight, seed number per pod, seed number per plant and grain yield recorded non-significant.

Table 2. Analysis of variance showing mean square values and significance test of agronomic traits among cowpea genotypes in different sites during 2017/18 season.

Source of variation	df	DFW	BRN	PNP	PWP	HSW	SPD	SNP	GRY
Genotype (G)	97	383.3*	75.68***	855.5***	1066.9***	63.55***	14.289***	26415***	865.1***
Location (L)	1	320,429.4***	37,825.89***	65,675.6***	123,967.1***	747.27***	0.000*	22,982ns	7397.1***
G x L	90	223.3ns	87.85***	614.0***	732.4ns	20.88ns	1.62ns	14,564ns	617.4ns
Residual	271	243.4	33.24	771.1	666.5	20.64	8.302	17,162	601.5

Notes: HSW, hundred seed weight; BRN, number of branches; DTF, days to flowering; PNP, pod number per plant; PWP, pod weight per plant; SPD, seed number per pod; SNP, seed number per plant; GY, grain yield; df, degrees of freedom; ** p < 0.01; *** p < 0.001; ns, non-significant.

Similarly, previous studies on cowpea genotypes indicated significant G × E on several agronomic traits such as number of branches, number of pods per plant, number of seeds per pod, pod diameter, plant height, pod length, hundred seed weight, grain yield, days to flowering maturity, plant height, grain yield, and number of seeds per pod in cowpea (Martos-Fuentes et al. 2017; Gerrano, van Rensburg, Kutu 2019). Ddamulira et al. (2017) reported significant G × E on grain yield among cowpea genotypes. The non-significant G × E observed for yield and yield-related traits in the current study suggests that the studied genotypes were fairly stable across test environments. Stability analysis across divergent environments in South Africa is useful to identify germplasm sources with suitable traits for recommendation.

Pairwise correlation coefficient analysis of agronomic traits

Agronomic traits highly correlated with grain yield are useful for selection and can be targeted in improvement programmes to increase yield gains (Kamara et al. 2011). A correlation matrix showing Pearson’s correlation coefficients depicting the level of association among agronomic traits is presented in Figure 1. Highly significant (p < 0.001) negative association was observed between days to flowering with number of branches (r = −0.26**). In contrary, Gerrano et al. (2015) reported non-significant correlation between such traits in cowpea. Pod number per plant was significantly and positively correlated with pod weight (r = 0.55**), seed number per pod (r = 0.34**), seed number per plant (r = 0.84**) and grain yield (r = 0.71**). Gerrano et al. (2015) reported negative correlation between number of pods per plant with number of seeds per pod with grain yield in cowpea, contradicting present findings. Lazaridi et al. (2017) reported highly significant and positive correlations between number of pods per plant with number of seeds per plant and grain yield concurring with present findings. Hence, the current study suggested that selection of cowpea genotypes with high pod number will result in grain yield improvement.

Figure 1. Correlation coefficient matrix showing associations of agronomic traits among cowpea genotypes evaluated across two environments. DTF, days to flowering; BRN, number of branches; PNP, pod number per plant; PWP, pod weight per plant; HSW, hundred seed weight; SPD, seed number per pod; SNP, seed number per plant; GY, grain yield.

Cowpea genotypes such as 90K-284-2, 98K-503-1, RV 111, RV 115, RV 213, RV 470, RV 501, RV 550 and TVU 2095 recorded pod number per plant ≥ 60 and higher grain yield per plant (≥45 g). Pod weight was positively associated with seed number per pod (r = 0.33**), hundred seed weight (r = 0.24*), seed number per plant (r = 0.75**) and grain yield (r = 0.76**). Some cowpea genotypes including 86D-1010, 90K-284-2, 98K-503-1, CH-47, GLENDA, ITOOK-1263, RV 115, RV 121, RV 202, RV 207, RV 213, RV 419, RV 465, RV 470, RV 477, RV 499, RV 501 and RV 550 recorded high pod weight which resulted in high grain yield. Seed number per pod was highly correlated with seed number per plant (r = 0.39**) and grain yield (r = 0.30**); whereas seed number per plant and hundred seed weight were positively and significantly correlated with grain yield (r = 0.73** and r = 0.33**), respectively. Similar to the current study, Lazaridi et al. (2017) and Gerrano, van Rensburg, Kutu (2019) reported significant correlation between hundred seed weight and grain yield in cowpea. Contrastingly, Martos-Fuentes et al. (2017) reported non-significant correlation between grain yield and kernel weight in cowpea. Genotypes 86D-1010, 90K-284-2, 98K-503-1, CH-47, GLENDA, ITOOK-1263, RV 121, RV 344, RV 419, RV 440, RV 477 and RV 550 recorded higher thousand grain weight (≥ 20 g) and high grain yield. Hundred seed weight of the genotypes in the current study was higher than the hundred seed weights reported by Belko et al. (2014). The result indicates that the grain weight can be improved through strategic recombination of the identified genotypes.

Path coefficient analysis of agronomic traits

Path analysis is useful in improvement programmes to identify and select traits with direct influence on traits of economic importance such as grain yield. Direct and indirect path coefficient values of studied agronomic traits are presented in Table 3. A high direct path coefficients value of 0.71 with significant correlation with grain yield (r_g = 0.76; p < 0.01) was observed between pod weight per plant and grain yield. These suggested direct selections of high pod weight can improve grain yield potential in cowpea. Cowpea genotypes such as RV 115, RV 121, RV 213, RV 28, RV 470, RV 477, RV 501, RV 550 and TVU 2095 recorded higher pod weights of 50, 89, 126, 55, 58, 106, 53 and 85 g respectively. Further, most of these genotypes recorded higher grain yield than most of the tested genotypes (Data not shown). Additionally, a moderate direct path coefficient value of 0.2 with highly significant (p < 0.01; r_g = 0.73) correlation to grain yield was observed for number of seeds per pod indicating direct selection targeting this trait can improve grain yield potential in cowpea. Genotypes such as 83s-911, 86D-1010, 98D-1399, 99K-494-6, RV 111, RV 115, RV 409, RV 414, RV 470, RV 500, RV 550 and TVU 2095 recorded higher number of seed per pod (≥13), thus serving as useful germplasm for improvement of this trait.

Table 3. Direct (diagonal and bold font) and indirect (off-diagonal) path coefficient values of agronomic traits in cowpea.

Traits	DFW	BRN	PNP	PWP	SPD	HSW	SNP	GRYrg
DFW	0 . 03	−0.11	0.00	0.19	0.00	0.00	0.03	0.14ns
BRN	−0.01	0.42	0.00	−0.14	0.00	0.02	0.02	0.30**
PNP	0.00	0.19	−0.01	0.39	0.00	0.01	0.13	0.71**
PWP	0.01	−0.09	−0.01	0.71	0.00	0.02	0.11	0.76**
SPD	0.00	0.01	0.00	0.24	−0.01	0.01	0.06	0.30**
HSW	0.00	0.08	0.00	0.17	0.00	0.08	0.00	0.33**
SNP	0.01	0.05	−0.01	0.54	0.00	0.00	0.15	0.73**

Notes: Days to flowering; BRN, number of branches; PNP, pod number per plant; PWP, pod weight per plant; SPD, seed number per pod; HSW, hundred seed weight; SNP, seed number per plant; GRYrg = genotypic correlation with grain yield,** significant at P <0.001 significance level; ns, non-significant.

Principal component analysis of studied agronomic traits

Principal component analysis (PCA) is useful to identify influential traits for effective selection and use in cultivar development. In the current study, PCA revealed three most important PCs (i.e. PC1, PC2 and PC3) among the studied agronomic traits, which contributed 44.0%, 19.0% and 12.70% variation, respectively, with a total variation of 75.2% (Table 4). Agronomic traits that contributed the most to variation in PC1 were days to flowering, pod number per plant, pod weight per plant, seed number per pod and grain yield, whereas, PC2 was positively correlated with number of branches and negatively correlated with hundred seed weight. Further, PC3 was positively correlated with seed number per pod Gerrano, van Rensburg, Kutu (2019) also reported that influential traits in cowpea include number of pods per plant, number of seeds per pod, hundred seed weight and grain yield.

Table 4. Principal component analysis showing eigenvectors, percent and cumulative variation of agronomic traits among 100 cowpea genotypes.

	Principal components
Traits	PC1	PC2	PC3
HSW	0.21	−0.66	0.01
BRN	0.24	0.87	0.13
DFW	0.89	0.25	0.07
PNP	0.78	−0.43	0.28
PWP	0.52	−0.06	−0.04
SPD	0.05	0.08	0.97
SNP	0.95	−0.07	−0.09
GRY	0.84	0.04	0.36
Eigenvectors	3.5	1.5	1.0
Percentage variation (%)	43.8	18.6	12.7
Cumulative variation (%)	43.8	62.5	75.2

Notes: HSW, hundred seed weight; BRN, number of branches; DTF, days to flowering; PNP, pod number per plant; PWP, pod weight per plant; SPD, seed number per pod; SNP, seed number per plant; GY, grain yield.

PCA bi-plots are very effective to identify genotypes possessing multiple traits to design an appropriate breeding strategy. PCA bi-plots (i.e. PC1 vs. PC2, PC1 vs. PC3 and PC2 vs. PC3) based on combined data across test environments showing groupings of cowpea genotypes and agronomic traits is shown in Figure 2. Cowpea genotypes Bechuana white, RV 161, RV 16, RV 194, RV 126, RV 207, RV 499, RV 121 and RV 447, 98K-503-1 and TVU 2095 were grouped together based on longer days to flowering and high pod weight per plant (Figure 2(a)). Similarly, Glenda, RV 465, RV 574, RV 115, RV 28, RV 419, RV 28, RV 419, RV 213, RV 550, RV 470, RV 111, RV 315 and RV 550 were grouped together based on high seed number per pod, seed number per plant, pod number per pod and grain yield (Figure 2(a,b)) useful for breeding for enhanced agronomic performance. Strategic recombination of these genotypes may improve gains for these traits in the progeny (Olajide and Ilori 2016). Genotypes ITOOK-1263, RV 344, RV 440, 86-D-1010, CH-47, 90K-284-2, RV-419 and RV-121 were grouped based on high values for thousand seed weight (Figures 2(b,c)) useful for breeding for increased grain weight. RV 205, Dr. Saunders and TVU 3416 were grouped based on low values for pod weight per plant, seed number per pod, hundred seed weight and grain yield (Figure 2(b)). These findings suggest the use of such genotypes as parents may result in low genetic gains targeting improvement of these traits.

Figure 2. Bi-plots of principal components showing % variation accounted for PC1 vs. PC2, PC1 vs. PC3 and PC2 vs. PC3 and relationships among cowpea genotypes and agronomic traits based on combined data across test environments. DFW, days to flowering; BRN, number of branches; PNP, pod number per plant; PWP, pod weight per plant; SPD, seed number per pod; HSW, hundred seed weight; SNP, seed number per plant; GRY, grain yield.

Variance components, heritability, genotypic and phenotypic coefficients of variation and genetic gain among agronomic traits

Estimated variance components, phenotypic coefficient of variation (PCV) and genotypic coefficient of variation (GCV) and other parameters, studied agronomic traits based on combined data across the environments are presented in Table 5. Further, parameters such as genotypic (σ² G), environmental (σ² E) and phenotypic (σ² P) variances, broad-sense heritability (H²) and genetic gain were estimated (Table 5). Phenotypic, genotypic, environmental and genotype by location variances were high for seed number per plant, pod weight, pod number per plant, days to flowering and grain yield. In the current study, all the studied traits recorded high GCV, whereas medium PCV values were observed for number of branches per plant (29.79%), seed per pod (22.74%) and days to flowering (26.13%), whereas the remainder of the traits recorded high values (≥44%). High values of PCV and GCV observed for studied traits in the present study suggest effective selection for these traits. Further, this also indicated the existence of substantial phenotypic variability among the tested cowpea genetic resource. Similar reports of high PCV and GCV for agronomic traits such as green pod yield per plant (Vidya et al. 2002; Narayanankutty et al. 2003); number of pods per plant (Vidya et al. 2002; Narayanankutty et al. 2003; Pal et al. 2003; Girish et al. 2006); ten pod weight (Xiong et al. 2002; Narayanankutty et al. 2003; Vavilapall et al. 2013); hundred fresh seed weight (Kumawat et al. 2005; Manggoel et al. 2012) and number of primary branches per plant (Nehru and Manjunath 2001) were reported to be highly variable in cowpea.

Table 5. Estimates of mean, variance, broad sense heritability, genotypic and phenotypic coefficients of variation and genetic advance of agronomic traits in cowpea.

Trait	Mean ± SE	MSG	MSE	MSgl	δ^2gl	δ^2g	δ^2p	δ^2e	PCV	GCV	H^2	GA	GG%
DFW	48.41 ± 10.58	383.30	246.40	223.30	100.10	160.00	381.60	246.40	28.08	26.13	91.93	260.57	538.29
BRN	11.71 ± 4.003	87.85	33.24	75.68	59.06	12.17	32.65	33.24	16.70	29.79	67.27	71.86	613.72
PNP	35.19 ± 14.21	855.50	398.00	614.00	415.00	241.50	582.50	398.00	40.69	44.16	91.46	320.13	909.79
PWP	29.06 ± 19.09	1066.90	666.50	732.40	399.15	334.50	835.63	666.50	53.62	62.93	80.03	376.76	1296.42
HSW	10.89 ± 2.73	63.55	20.64	20.88	10.56	42.67	90.50	20.64	28.82	59.97	97.15	134.56	1235.31
SPD	15.65 ± 3.257	14.29	8.30	1.62	−2.53	12.67	27.41	8.30	13.23	22.74	96.21	173.31	468.40
SNP	199.14 ± 83.02	26415.00	13549.00	14564.00	7789.50	11851.00	27089.25	13549.00	116.63	54.67	93.75	2242.56	1126.11
GRY	33.58 ± 17.49	865.10	601. 50	617.40	316.65	247.70	645.78	601.50	43.85	46.87	38.36	324.21	965.56

Notes: DFW, days to flowering; BRN, number of branches; PNP, pod number per plant; PWP, pod weight per plant; SPD, seed number per pod; HSW, hundred seed weight; SNP, seed number per plant; GRY, with grain yield; MSG, Mean square genotype; MSE, mean square error; MSgl, mean square genotype x location; δ²gl, Genotype x location variance; δ²g, genotypic variance; δ²p, phenotypic variance; δ²e, environmental variance; PCV, phenotypic coefficient of variance; GCV, genotypic coefficient of variance; H², heritability; GA, genetic advance; GG%, percent of genetic mean.

Genetic advance (GA) values ranged from 71.86 to 2242.6% with seed number per pod recording the maximum GA value of 2242.56%, followed by pod weight (376.76), grain yield (324.21), pod number per plant (320.13), and days to flowering (260.57). High values of expected genetic advance in vegetable cowpea were also observed for green pod yield per plant (Kumawat et al. 2005; Vavilapall et al. 2013) and pod weight Narayanankutty et al. (2003) in cowpea. However, low GA was recorded for number branches per plant (71.86) and seed number per pod (96.21), and are in agreement with Khanpara et al. (2016).

Heritability is classified as low (<30%), medium (30–60%) and high (≥60%), according to Johnson et al. (1955). In the current study, broad-sense heritability ranged from 38.36 (grain yield) to 97.2% (hundred seed weight). All agronomic traits under study recorded high heritability values suggesting additive gene effects control the expression of the traits. Similar results were reported by Omoigui et al. (2006), Adeyanju and Ishiyaku (2007), and Manggoel et al. (2012) in cowpea. Additionally, Khanpara et al. (2016) also reported high heritability values for traits such as days to 50% flowering, number of primary branches per plant, number of seeds per pod, pod weight, number of pods per plant and green pod yield per plant. Further, this suggested that the phenotypes were the true representative of their genotypes for these traits and that selection based on phenotypic value could be reliable.

Genetic advance expressed as percentage (GG) was high for pod weight (1296.2%) followed by hundred seed weight (1235.3%), seed number per plant (1126.1%), grain yield (965.56%) and pod number per plant (909.76%). Current findings concur with similar findings by Hermes et al. (2018) who reported higher GG (36.06) for number of pods per plant, and 84.11 for 100 GG values for seed weight. Traits such as pod weight, number of pods per plant and for characters such as days to 50% flowering, and number of seeds per pod (Khanpara et al. 2016; Hermes et al. 2018). Srinivas et al. (2017) also reported that number of branches per plant, plant height, pod yield per plot, total number of pods per plant and number of seeds per pod exhibit high GG values in cowpea. Furthermore, high heritability and genetic advance were recorded for pod weight (80.03%; 1296.42), days to flowering (91.93%; 538.29), pod number per plant (91.46%; 909.79), seed number per plant (93.75%; 1126.11), hundred seed weight (97.15%; 1235.31), and seed number per pod (96.21%; 468.4). High values for heritability coupled with high genetic advance for studied agronomic traits, which indicates additive gene action control the expression inheritance of these traits in cowpea (Johnson et al. 1955; Ubi et al. 2001). Planned crosses among complementary cowpea genotypes possessing these traits may result in improved genetic gains and enhanced performance of progenies.

In conclusion, this study indicated that significant variability exists among cowpea genotypes for agronomic traits. Thus, there exist opportunities for selection of superior cowpea genotypes for use in improvement programmes. Cowpea genotypes namely: Glenda, RV 465, RV 574, RV 115, RV 28, RV 419, RV 28, RV 419, RV 213, RV 550, RV 470, RV 111, RV 315 and RV 550 were selected for breeding with suitable yield and yield-related traits.

Acknowledgements

The first author would like to thank the Department of Science and Technology for funding. The first author gratefully appreciates the technical assistance of Dinah Scott and staff of the Agricultural Research Council-Grain Crops in Potchefstroom.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes on contributors

Maletsema Alina Mofokeng is a Researcher in Plant Breeding at the Agricultural Research Council-Grain Crops, Potchefstroom, South Africa.

Jacob Mashilo is a Crop Scientist at the Limpopo Department of Agriculture and Rural Development, Limpopo Province, South Africa.

Paul Rantso is a Senior Technician in Plant Breeding at the Agricultural Research Council-Grain Crops, Potchefstroom, South Africa.

Hussein Shimelis is a Plant Breeder and the Deputy Director for African Centre for Crop Improvement (ACCI) at the University of KwaZulu-Natal, South Africa.

ORCID

Jacob Mashilo http://orcid.org/0000-0001-9015-043X

Additional information

Funding

This work was supported by the Department of Science and Technology and the Agricultural Research Council.