Cowpea (Vigna unguiculata (L.) Walp) is a crop with wide global distribution, especially in tropical regions. It is an important grain legume crop in sub-Saharan Africa, with significant production in Nigeria, Burkina Faso, Niger, Cameroon and United Republic of Tanzania (Fatokun et al., 2012; FAOSTAT, 2017). The crop is produced under rain-fed conditions in the drier regions of the world where drought is common due to low and erratic rains (Agbicodo et al., 2009). Cowpea is a high valued crop due to  its  high  quality  protein,  adaptation  to  heat  and drought, and ability to fix nitrogen. These characters make it a major crop in the context of climate change and food security (Carvalho et al., 2017).

Cowpea is an important cash crop and the main grain legume in Botswana, most production is mainly from small scale farmers under rain-fed conditions. Despite cowpea widespread cultivation, crop yield is low, with an average of 300 kg/ha, against a potential yield of 2500 kg/ha. Generally, low cowpea productivity and production is attributed to several factors such as  biotic  and  abiotic stress, low yielding varieties, poor soils and poor crop husbandry (plant density) (Olasupo et al., 2016; Mwale et al., 2017). Additional potential constraints of cowpea production could be worsened by climate changes, drastic changes in rainfall patterns, and rise in temperature which could lead to unfavorable growing conditions thereby modifying growing seasons (Ajetomobi and Abiodun, 2010). Therefore, development and adoption of well adapted drought tolerant cowpea varieties that will cope with changing climate conditions is a priority.

International Institute of Tropical Agriculture (IITA) is integral in the development and protection of cowpea and keeps a significant amount of accessions for mining to develop superior performing cultivars (Boukar et al., 2016). The Ministry of Agricultural Development and Food Security, through cowpea breeding program has benefited significantly from this initiative as several cowpea genotypes have been evaluated in Botswana. Varieties ER7 and IT95K6352 sourced from IITA have been evaluated and released for their extra early maturing and early maturing characters, respectively. The other released varieties in the country include Inia 30, Inia 70 and Inia 71 originally from Mozambique (Chiulele, 2010). They have medium maturing characteristics and high yield potential, other varieties released in Botswana include Tswana and Botswana Blackeye. However, several efforts have been made to characterize more than 400 cowpea germplasm accessions collected from many agricultural districts of the country (Molosiwa et al., 2016) with the aim to include more genetic material into the cowpea breeding program. Therefore, the objective of this study was to compare the performance of introduced and local cowpea genotypes under the semi-arid environment of Botswana, and identify well adapted varieties with potential for release.

Plant material

Twenty-four cowpea genotypes with early and medium maturing characteristics and high yield potential were selected for use in the study. Twelve are local lines sourced from the cowpea breeding program but originally from the National Plant Genetic Resource Centre (Botswana), while the other 12 were introduced genotypes from various countries and institutions. The characteristics of the genotypes are described in Table 1.

Study site

Field experiments were conducted for three consecutive seasons at Sebele Agricultural Research Station, 12 km north of Gaborone, in Southern Botswana (24°35’S; 25°56’E). The area has a unimodal rainfall type that falls between November and April each year. The area receives an average 30-year annual rainfall of 500 mm/year, with a semi-arid climatic condition. The rainfall received during the growing season of 2012-13 was 325.3 mm with an average temperature  of   22.8°C,   in   2013-14,   327.3 mm  of  rainfall  was received with a mean temperature of 21.9°C, while in 2014-15, 180 mm rainfall was received with a mean temperature of 23.8°C. The soils in Sebele are shallow tropical soils consisting largely medium coarse texture and sandy loam with low water holding capacity.

Experimental design and crop management

The soil was ploughed and harrowed with a tractor for land preparation to set up the experiment. A three replicated trial with a randomized complete block design was conducted in a total area of 899 m². Each plot length was 4 m with a row spacing of 0.75 m and within plant spacing of 0.2 m, with a plant population estimates of 67000 plant per hectare. Two seeds were planted per hole and one seedling was left after thinning 21 days after planting. Crop management practice of weeding was adhered to, for the control of pests and disease in the experimental plots.

Agro-morphological characteristics measured

Morphological and agronomic characters selected were chosen from the International Board for Plant Genetic Resources, IBPGR (1983). Thirteen quantitative traits were recorded which include, days to 50% flowering (DFF), plant height (PH), pods per plant (PodpPlant), seeds per plant (SeedpPlant), seeds per pod (SeedpPod), pod length (PodL), pod width (PodW), 100 seed weight (100SW), rainfall use efficiency (RUE), seed yield per hectare (YDHA), Spad chlorophyll meter reading (SCMR), shoot dry weight (SDW), days to maturity (DM). At harvest, data were collected within the two middle rows. Shoot dry weight was oven dried for 48 h at 70°C and weighed. Rainfall use efficiency was estimated based on the formula, RUE = Y/R which defines RUE as dry matter accumulated/precipitation in Kg ha^-1 mm^-1, a formula adopted from Wang et al. (2016). Yield (Y) is the dry matter accumulated for the cowpeas, while rainfall (R) is the accumulated rainfall for the growing season. Estimates of RUE for each genotype was computed based on the amount of rainfall received per growing season and the seed yield amount. The SCMR measurements were taken at 45 days after sowing using SPAD-502 meter (Minolta, Japan).

Statistical analysis

Data were subjected to analysis of variance and multivariate statistical analysis was performed for Principal Component and Cluster Analysis to reveal the relationship between the selected cowpea genotypes using Genstat software version 18.0

Genetic variation of the 24 cowpea genotypes

A combined analysis of variance indicated a substantial amount of variation for all the 13 quantitative characters among the cowpea genotypes since most traits were highly significant (P <0.001) (Table 2). The higher genotypic variability is a great opportunity to identify best performing cultivars. The results concurred with those of Osho and Olasanmi (2018), Araméndiz-Tatis et al. (2018) and Gerrano et al. (2015), who reported statistical differences between cowpea genotypes studied. Most of the    characters    were   not    affected   by   genotype   x environment interaction with the exception of days to 50% flowering, plant height, and seeds per plant (Table 2), which imply that the performance of the genotypes was consistent across  the  three  seasons,  even  though there were seasonal differences among most of the traits. The highest coefficient of variation was observed in rainfall use efficiency (59.4%), seed per plant and seed yield  per   hectare   both   at  58%.  Similar  findings  with highest coefficient of variations among cowpeas were observed on seed yield per hectare (46.9%) and seed per plant (44.4%) by Aliyu and Makinde (2016), which reveals greater variation among these characters. Considerable genetic variability that exists among genotypes was evidenced by a highest range of seed yield (318 - 866 kg/ha), pod length (117 - 160 mm) and seed per plant (83 - 211) (Table 3). Relatively higher genotype ranges were also reported for seed yield varying from 263.76 to 2624.8 kg/ha in South Africa by Nkoana et al. (2019), while seed yield per hectare ranged 522.00 to 2807 kg/ha, pod length from 10.74 to 22.80 and seed per plant from 64.2 to 360 were reported by Aliyu et al. (2019) in Nigeria. These greater variations indicate a potential for improvement of this crop. The overall average seed yield of 617.0 kg/ha (Table 2) is much higher than the national average cowpea production, around 139 kg/ha (Statistics Botswana, 2013). However, this average is about the half crop yield of 1270.68 kg/ha recorded for cowpea in Nigeria (Aliyu et al., 2019), and much lower than 2043 kg/ha for cowpea in Zimbabwe (Matova and Gasura, 2018). Generally, there was a lower seed yield performance among the selected cowpea genotypes as indicated by the grand mean (Table 3). This can be attributed to the lower amount of rainfall than the normal amount, that is, 500 mm for a 30-year annual. However, this was the best time to conduct this research in order to compare the performance of introduced and local cowpea genotypes and identify the best performing varieties, selecting for drought adaptation (Hall, 2002).  

Comparison of the performance of local and introduced cowpea genotypes

A summary  of  the  performance  of  local and introduced cowpea genotypes is shown on Table 3. Comparatively, on average the local genotypes matured later (78 days), are taller (33 cm), have higher shoot biomass (56 g), but with slightly less 100 seed weight (14.6 g) and seeds per pod (10), and much less seed yield per hectare (595 kg ha^-1) compared to the introduced improved cowpea genotypes, except genotypes B137B (Table 4). Traits of late maturing plants with high biomass among local genotypes could have been selected by farmers who are mainly interested in the leaf vegetable of the cowpea crop, while the early maturing among introduced lines could be influenced by continued selection for earliness by breeders to reduce the drought effect. A few of introduced genotypes underperformed in some traits, such as lower seed yield (318 kg/ha) on IT95K20722 and lower pod length (117 mm) by ER7. This could be explained by a lack of adaptation to the arid conditions of Botswana. To improve the pod size of the higher yielding ER7 it would be ideal to hybridize it with those genotypes with larger pods such as B137B and Inia 71 (Table 3).

Genetic diversity and relatedness of the selected 24 cowpea genotypes

Principal Component Analysis was used to identify traits that account for most of the variation among the 24 genotypes. The eigenvalues >1 were selected and used to define the 13 agro-morphological traits (Table 5). The first four PCA explained a total variation of 76.16%, with the first two PCs accounting for 51% of variation. These results further support the variation observed in the analysis of variance in Table 2. Characters that contributed mostly to the variation in PC1 (32.5%) in descending order are: RUE, day  to  maturity,  seed  yield per hectare, days to 50% flowering and 100 seed weight; in PC2, accounting for 18.54% variation it was mainly pod width, seed per pod, shoot dry weight and pod length; PC3 contributing to15.64% variation  was mainly influenced by pods per plant, pod width, 100 seed weight, and pod length, while seed per plant and plant height were the main contributor in the last PC4 (9.56%) (Table 5). The findings are in accordance to those by Nkoana et al. (2019) who reported a total variation of 78.20% among the first 4 PCs, where the first two PCs accounted for 50.2% of variation. They also identified influential traits such as number of seeds per pod, 100 seed weight, pod length and grain yield.

A hierarchical cluster analysis dendrogram for the 24 cowpea genotypes based on the 13 traits shows an overall similarity of 1 to 0.8 (Figure 1). The dendrogram clustered the genotypes into two main groups at 0.825 Euclidean distances. The first cluster consists of 16 genotypes ranging from genotype B097 to B342, and this include equal number of both introduced and local genotypes within the cluster excluding the B342 which is an outlier within the first group. This reveals high phenotypic similarity between the local and introduced material, and the low genetic diversity within groups. This could be influenced by the fact that cowpea is predominantly a self-pollinating crop and is expected to a certain degree of heterozygosity (Badiane et al., 2012). Cluster 2 consists eight genotypes (five introduced) which consists all the three genotypes from Mozambique and the three locals (B137B, B138, SCAM151), and these are generally high performers. The clusters have been delineated mainly based on traits with higher loadings as defined in Table 5, such as seed RUE, day to maturity and seed yield per hectare. The cluster analysis revealed some similarities between local and introduced cowpeas, such as among (B137B and INIA71), and also (B097 and IT95K107257) as shown in Figure 1.  For further genetic improvement it would be ideal to conduct some hybridization between the clusters to improve cowpea genotypes in Botswana. Genotype B342 which is an outlier is more diverse and should be prioritized in selection. It was the last one to reach 50% flowering, had the lowest plant height, lowest number of seeds per pod and lowest pod width, but was among the top 5 best seed yield producers, with 778 kg/ha (Table 4). Therefore, selection of desirable traits in this genotype could assist to improve it. Farmer variety is genetically similar to other local cowpea genotypes such as SCAM123 and B549 (Figure 1), and it yield relatively low at 329 kg/ha, and flowered late, longer than 50 days for 50% flowering (Table 4). Similarly, Aliyu et al. (2019) observed that farmer varieties are well adapted to low input conditions, poor yielders, indeterminate, mature late and are susceptible to biotic and abiotic conditions. Joseph (2014) argued that farmers continue growing their low yielding varieties of cowpea because they prefer specific traits of interest such as grain quality, taste or leafy vegetables. These are some of the traits that plant breeders will have to incorporate in their breeding program to develop farmer preferred varieties.

In this research, based on thirteen selected traits the best performing varieties were identified. Even though the introduced cowpeas have a relatively better performance, two local genotypes, B137B and B342, were among the top five in terms of seed yield per hectare. Overall, the study revealed that the local cowpea genotypes compared well with some introduced lines. Introduced genotypes with larger seeds such as IT95K20722 will be useful sources of genes for other cowpeas and this could be useful for market acceptability This results could also have implications in selection in the local cowpea breeding program based on the performance of the best varieties such as, B137B and B342 which are not yet released to be evaluated for on-farm under participatory variety selection.

The authors have not declared any conflict of interests.