Association Mapping for Drought Tolerance and Yield-Related Traits in Cowpea Accessions

The objective of this study were to conduct association mapping for drought tolerance at the seedling stage and yield-related traits. 60 cowpea accessions were used in the study. Singlenucleotide polymorphisms (SNPs) discovered through genotyping by sequencing (GBS) were used for genotyping. Association mapping was conducted using single-marker regression (SMR) in Q Gene, and general linear model (GLM) and mixed linear model (MLM) built in TASSEL. The population of the cowpea accessions were analysed using STRUCTURE 2.3.4 and the peak of delta K in the greenhouse showed seven population types, whereas the peak of delta K in the glasshouse indicated the presence of six population types. One SNP marker, 14083649|F|0-9 was associated with NP with a p value <0.001. Fifty SNP markers were associated with PWT at p <0.001. Four SNP markers, 14074781|F|0-16, 100047392|F|0-36, 14083801|F|0-28 and 100051488|F|0-49 were associated with AVSPD at p <0.001. SNP markers, 14074781|F|0-16, 14083801|F|0-28 and 100051488|F|0-49 were associated with PL at P <0.001. Five SNP markers, 100047392|F|0-36, 14083801|F|0-28, 100072738|F|0-34, 14076881|F|0-49 and 14076881|F|0-49 were associated with PWDTH at p <0.001. The 65 SNP markers identified can be used in cowpea molecular breeding to select for AVSPD, NP, PL, PWDTH, PWT, and RR through marker assisted selection (MAS).


Introduction
Cowpea [Vigna unguiculata (L.) Walp.] is a food legume of the family Fabaceae/Papilionaceae (1). According to (2), all cultivated cowpeas are grouped under the species Vigna unguiculata, which is subdivided into four cultivar groups: Unguiculata (common cowpea for food and fodder), Biflora (catjang), Sesquipedalis (yard long or asparagus bean used as a vegetable), and Textilis (used for fibres). The crop is of major importance to many smallholder farmers in Africa and the developing world, as it serves as food, cash crop, animal feed, and manure (3).
The major aims of cowpea breeding are high yield, early maturity for grain production, long vegetative period for vegetable production, high leaf and grain nutrient contents, high cooking quality, and high emergence rate. In order to provide farmers with quality seeds of improved cultivars, breeding programmes and seed systems should be based on information on the genetic diversity available in the germplasm (4). According to (5), assessments of phenotypic or genotypic diversity in cultivated plants provide useful information for the improvement of germplasm collections, which provide material for genetic improvement and breeding.
Studies of association mapping for drought tolerance in cowpea using DArTSeq genotyping data are very limited. Drought is a major production constraint in the smallholder farming sector in Zimbabwe; thus, there is a need to develop drought-tolerant varieties, which in turn requires the identification of genotypes that carry genes associated with drought tolerance.
Association mapping was used to investigate the associations among 76 SSR markers and six drought-related traits on a set of 107 barley accessions evaluated under well-watered and drought-stress conditions (6). The results showed that there were 36 significant marker-trait associations for drought-related traits. (7), used single nucleotide polymorphisms (SNP) associated with drought tolerance indices in 328 wheat lines using a genome-wide association study (GWAS) under fully irrigated and rain-fed conditions. Results showed that most associations were located on chromosome 4A, and that this chromosome is very important in drought tolerance and should be used in wheat improvement programmes.
In a study of correlation coefficient and path analysis in the cowpea germplasm line, (2), observed significant and positive correlations between the growth characters and seed yield of cowpea. Using a path analysis study, the experiment further concluded that seed yield in cowpea can be improved by focusing on the traits of biological yield per plant, harvest index, number of pods per plant, and plant height. (8), studied phenotypic and genotypic divergence for yield and related quantitative traits among 30 cowpea landraces in Cameroon. The study revealed strong correlations between seed length and grain yield, 100-seed weight and grain yield, 100-seed weight and seed length, number of seeds per pod and pod length, number of branches per plant and plant biomass, and grain yield and leaf width. Thus, characters such as seed length or 100-seed weight are very useful in early selection when improving yield. (9), evaluated the genetic variability among 20 wild cowpea accessions and observed high morphological variability among the accessions. The high variability observed among the wild cowpea accessions in terms of their agro-morphological and yield parameters provided useful traits in the crop that can be exploited for its improvement. Results obtained from (10), on population structure analysis and association mapping of the seed antioxidant content in the 369-accession USDA cowpea [Vigna unguiculata (L.) Walp.] core collection using SNPs show that there were significant correlations between the seed antioxidant content and black seed colour. It was further observed and concluded that cowpea accessions with red and black seed coat colours were useful as parents in cowpea breeding programmes to provide new cowpea cultivars with high seed antioxidant contents. (11), analysed the genomic regions, cellular constituents and genes controlling pod length variation in cowpeas. The research observed that cell proliferation was the major reason for extended pods as against cell elongation or enlargement. A total of 116 and 155 cowpea accessions during emergence and seedling stages, respectively were analysed for salt tolerance index with 1,049 single nucleotide polymorphisms (SNPs) that were used for association analysis (12). A total of three SNPs, Scaffold 87490_622, Scaffold87490_630, and C35017374_128, were highly associated with salt tolerance during germination stage while seven SNPs, Scaffold93827_270, Scaffold68489_600, Scaffold87490_633, Scaffold87490_640, Scaffold82042_3387, C35069468_1916, and Scaffold93942_1089, were associated to salt tolerance at the seedling stage. Thus, these SNP markers could be used as a tool to select salt-tolerant lines for breeding improved salt-tolerant cowpea cultivars.
The objective of this study were to conduct association mapping for drought tolerance at the seedling stage and yield-related traits in cowpea.

Phenotype data
A total of 60 cowpea accessions collected from three geographic origins were used in this study (Table 1). Of these, 33 accessions were from the International Institute of Tropical Agriculture (IITA) in Nigeria, 19 were from the Agricultural Research Council -Grain Crops in South Africa, and eight were from smallholder farmers in Buhera District in Zimbabwe.
The seeds used were grown under favourable conditions in two screen houses (glasshouse and greenhouse). All of the populations phenotyped were grown in greenhouse and glasshouse trials. The cowpea accessions were planted in pots in topsoil mixed with compost (3:1) at the Agriculture Research Council -Grain Crops, Potchefstroom, South Africa in January 2019 for the greenhouse trial and February 2019 for the glasshouse trial. A triplicated 6× 10 alpha lattice design was used for the experiment. In all greenhouse and glasshouse trials, mature pods were harvested and dried for storage (<15% moisture) after screening for drought tolerance. Seeds were subsequently cleaned from the pods, counted, and weighed.

Population structure analysis
The population structure of the cowpea accessions evaluated for growth traits was inferred using STRUCTURE 2.3.4 (14). Population structure (K) was analysed with an admixture model with a correlated allele frequency model, which was independent for each run. The identification of the delta K values and optimal K, based on the formula devised by (15). The formula allowed a reliable screening of appropriate K values using Structure Harvester (http://taylor0.biology.ucla.edu/structureHarvester/; 16). A Q-matrix and K vectors were established shortly after the optimal K was computed. The Q-matrix was used for association analysis studies in TASSEL (Trait Analysis by Association Evolution and Linkage) (17).

Association analysis
SNP genotype data generated by genotype by sequencing (GBS) was first filtered to remove the monomorphic SNP sites. Report_DCpe18-2608_SNP_singlerow_2.csv was converted to vcf format using matk. The vcf file was filtered using vcf tools. The vcf format was then converted to plink files using vcf tools. Marker-trait association analysis was evaluated using using plink. Analysis in R software was done using the following packages; vcfR, poppr, ape and qqman. Significantly associated SNP markers with traits were identified at p<0.001 (17).

Results
The population of the cowpea accessions were analysed using STRUCTURE 2.

Association Analysis
SNP markers were identified for number of pods, recovery rate, pod weight, average seeds per pod, pod length and pod width. Two SNP markers, 14083649|F|0-9 and 100100635|F|0-53 were associated with number of pods (NP) with a p value <0.001 ( Table 2). The significant markers occurred on chromosome 10. SNP marker 100100635|F|0-53 contributed 43% of the phenotypic variation.SNP marker 100084158|F|0-6 was associated with recovery rate (RR) at p <0.001 and was positioned at chromosome 10 while R 2 was at 10%. (Table 2).
Fifty SNP markers were associated with pod weight (PWT) at p <0.001 ( Table 2). Out of these,  Association mapping was performed using rrBLUP to identify loci linked to the evaluated traits. Significant SNPs were compared to those that passed a significance threshold of log10 (p)>5] in TASSEL 5.0 analysis. Figure 2 highlights the association in glasshouse experiment on the number of pods (NP), recovery rate (RR), and pod weight (PWT).
The average seeds per pod (AVSPD), pod length (PL), pod width (PWDTH) and pod weight (PWT) had were significantly associated in the greenhouse experiment ( Figure 3).  (19). Thus the PWT trait was more pronounced in the glasshouse with more chromosomes exhibiting the trait than in the greenhouse. The study pointed out that these yield related traits are some of the important parameters to be used when selecting cowpea accessions for drought tolerance at seedling stage in screen houses. However, (20) observed that stem greenness, survival and recovery dry weights in greenhouse were the useful traits to screen cowpea genotypes for their ability to withstand drought stress at the seedling stage.
Under this study it was also observed that there was a strong co-location of SNP markers especially on chromosomes 1, 3, 5, 6, 7, 8, 9, 10 and 11. SNP markers 100051488|F|0-49 and 14083801|F|0-28 on chromosome 3 were associated with both AVSPD and PL. SNP marker 14083801|F|0-28 was also associated with PWDTH on chromosome 3. This suggests that drought tolerance traits are complex and these determines accurate measurements. (21), investigated candidate genes for seedling drought stress-induced premature senescence and observed seven markers co-located with peaks of previously identified QTL using restriction site polymorphisms. The co-location of these markers suggested that these markers were derived from genes which were involved in cowpea response to drought stress-induced premature senescence. (22), observed that when there is a smaller p value, then that SNP marker is very ideal and should be validated for marker assisted selection (MAS).
Most of the trait-associated markers were different under the two screen houses, indicating the environmental effects in these associations (23). These results showed that different genes might contribute to the same trait in several environments (24) or there could be a change within the expression level of the same gene between two environments (25). Associated markers repeatedly detected in two or more different environments are considered more reliable than those present in just one environment (26). In this study, 2 markers showed stable association with different traits under both screen house conditions, notably markers 100051488|F|0-49 and 14083801|F|0-28. The detection of genomic regions associated with multiple traits across variable environments is essential in breeding crops for wide adaptation and yield stability (27).
Previous research has shown that plants with good drought tolerance at early vegetative growth were also able to withstand drought stress at a later stage of plant development (28). Drought tolerance is a complex phenomenon as it is controlled by many genes thus the use of more efficient tools like genomic selection (GS) for accelerated trait improvement is ideal (29). This is because during QTL analysis, (21), observed that the tolerant genotypes also contributed alleles that negatively influenced drought tolerance, and that the susceptible parent contributed alleles that enhanced drought tolerance. The use of SNP markers at seedling stage for drought tolerance can be a fast and cheaper way than the use of conventional breeding methods in a field environment.

Conclusion
The screening of cowpea accessions in a controlled environment is a fast way of evaluation, especially where temperature regulation is needed. Some variability in drought tolerancerelated traits among cowpea genotypes was observed in this study in both greenhouse and glasshouse experiments. The population structure analysis revealed that under that two screen houses there were seven subgroups although this was more pronounced in the greenhouse experiment. Drought tolerance in cowpea is controlled by multiple traits in cowpeas as was observed with SNPs100051488|F|0-49 and14083801|F|0-28. It is thus necessary to have accurate measurements of intended traits. In terms of drought tolerance at the seedling stage, various temperature regimes can be controlled; this can give desired results much quicker than field selection. The 65 SNP markers identified may be used in cowpea molecular breeding to select for AVSPD, NP, PL, PWDTH, PWT, and RR through MAS.