Cowpea [Vigna unguiculata (L.) Walp.] is a predominantly self-pollinated crop and a diploid (2n = 2x = 22) with genome size of 640.6 Mbp (Lonardi et al., 2019). Cowpea is one of the most important legume crops cultivated and consumed in dry areas of the tropics and subtropics (Boukar et al., 2016). The crop plays vital roles as a nutritional and food security crop and generates income for farmers and food vendors. On average, cowpea grains contain 25% protein, 53.2 mg/kg iron, 38.1 mg/kg zinc, 826 mg/kg calcium, 1,915 mg/kg magnesium, 14,890 mg/kg potassium, and 5,055 mg/kg phosphorus (Boukar et al., 2011). The crop is also an important source of nutritious fodder for ruminants mainly in the Sahel regions of West and Central Africa (Boukar et al., 2016, 2019). Global annual cowpea production is estimated at 6.5 million metric tons on about 14.5 million hectares (Boukar et al., 2016; Kebede and Bekeko, 2020). Of the total world production, about 83% is obtained in Africa, with West Africa accounting for 80% of African production, and Nigeria taking the lead in world production and consumption. Despite the importance of cowpea, its productivity is still below expectation in most African countries due to attack by multiple insect pests, diseases and parasitic weeds, as well as drought, heat, and low soil fertility (Carsky et al., 2001; Agbicodo et al., 2009; Boukar et al., 2016, 2018; Togola et al., 2017, 2020; Ongom et al., 2021). These constraints, if not mitigated by crop improvement through plant breeding, will continually dwindle the future of cowpea, denying farmers and consumers the opportunities to reap the potential benefits of agricultural productivity and food and nutrition security from this crop. Among multiple mitigation measures that have been advanced to curve the highlighted challenges in cowpea production, the deployment of cultivars that display resistance and/or tolerance to the stresses in farmers’ fields has been the focus of breeding programs in many research institutions.

Conventionally, the development of an improved cultivar involves an initial step in hybridization among selected parental lines having attributes of interest such as higher yield, resistance to any of the stress causing factors. This is then followed by advancing the progenies through repeated self-pollination until the lines are homozygous and stable before extensive testing are conducted to identify improved lines (Khanna, 2018; Osei et al., 2020). This process is lengthy and costly, yet the eventual creation of the intended variation through hybridization and the ability to further study the genetics of cowpea cultivar development requires validity of the initial cross. First, the parents of a cross should not be genetically similar in order to maximize the diversity for selection and improvement (Mafakheri et al., 2017; Nkhoma et al., 2020). It is therefore advantageous to fingerprint parental lines and assess relatedness prior to hybridization. Secondly, the progenies being advanced must be a product of successful cross pollination so as to capture the intended inheritance of trait combinations while maintaining an optimal level of diversity in the breeding program. Cowpea, being a highly self-pollinated crop, is inherently and botanically designed to avoid outcrossing. Consequently, hybridization requires special techniques to remove the anthers just before selfing occurs (Myers, 1996). Therefore, inaccurate timing of flower emasculation may result in self- pollinated plants and not true F₁ hybrids as the breeders intended. This quickly becomes a quality concern in breeding programs that often produce thousands of crosses. A routine best practice in modern breeding programs incorporates quality control (QC) steps by leveraging molecular breeding tools at stages in the breeding workflow that optimize the cost benefit of delivering high quality cultivars. It has been observed that crossing in cowpea by hand emasculation is relatively easy compared to some other legumes given that its flowers are large and easy to manipulate, the keel is straight, beaked and not twisted. However, the success rate is reportedly as low as 10–20% (Myers, 1996). This low success rate is attributable to high flower abscission due to mechanical handling during emasculation (Myers, 1996). Interestingly, out of the perceived successful crosses, some end up as self-pollinated plants having escaped the manual cross-pollination process. Without any quick detection technique, these escapes would not be easily detectable, constituting in potential wasted time and resources that would be carried forward in the breeding program. Efficient and optimized breeding programs include routine QC to eliminate the false F₁ plants early enough to make the most impact with the resources available (PASSeL, 2019). With the advancements in genetics, molecular biology and tissue culture, plant breeding is now increasingly being carried out using molecular genetic tools (Tester and Langridge, 2010; Boukar et al., 2018, 2020; Khanna, 2018). Parental fingerprinting and hybridity tests, which are among the key aspects of QC in the breeding process can now be conducted using molecular markers (Weerasinghe et al., 2004; Wu et al., 2010; PASSeL, 2019; Osei et al., 2020). The past few decades have witnessed notable progresses in the evolution of molecular marker systems and the detection platforms (Mammadov et al., 2012). Of all the marker systems, single nucleotide polymorphism (SNP) has become a standard marker type of choice in molecular breeding and genetics due to its specificity and abundance in the genome coupled with ease of automation and high-throughput detection system (Mammadov et al., 2012; Jiang, 2013). The wide-scale use and application of SNP markers in molecular breeding has been enabled by the development of user-friendly tools and platforms (Jiang, 2013). For greater precision and effectiveness in marker assisted selection (MAS), multiplex detection system for SNPs has been developed (Jiang, 2013; Bernardo et al., 2015; Nadeem et al., 2018). KASP, developed by Laboratory of the Government Chemist (LGC) Genomics (Braae et al., 2014) is an example of a recently developed multiplexed SNP marker detection system that combines several markers in a single assay (Nadeem et al., 2018). It is a standardized technology whereby genotyping is based on fluorescence. The technique rests on an allele-specific oligo extension and fluorescence resonance energy transfer that are used for signal detection (Braae et al., 2014; Nadeem et al., 2018). The KASP assay, together with other types of markers, have been applied successfully for germplasm diversity analysis or fingerprinting and hybrid purity tests in several crops including tomato (Osei et al., 2020), melon (Kishor et al., 2020), lettuce (Patella et al., 2019a), rice (Yashitola et al., 2002; Sundaram et al., 2008), maize (Patella et al., 2019a), and cotton (Selvakumar et al., 2010).

Recently in cowpea, there has been tremendous efforts to develop genomic resources to facilitate molecular-aided breeding. In this regard, Muchero et al. (2009a) developed an Illumina 1,536-SNP GoldenGate genotyping array for cowpea derived from 183,118 expressed sequence tags (ESTs). Out of these, 928 quality SNPs were incorporated into a consensus genetic map of cowpea, developed based on 741 recombinant inbred lines derived from 6 mapping populations. This linkage map spans 680 cm with 11 linkage groups and an average marker distance of 0.73 cm. This linkage map provides a framework for molecular marker deployment in cowpea breeding. In addition, a whole-genome shotgun (WGS) assembly and a bacterial artificial chromosome (BAC) physical map were developed from the IITA cowpea variety IT97K−499−35 (Muñoz-Amatriaín et al., 2017). These genomic resources allowed the development of 51,128 Illumina Cowpea iSelect Consortium Array (Muñoz-Amatriaín et al., 2017). Several of these SNPs were utilized to develop KASP markers and are available either as trait-linked markers for marker-assisted breeding or for QA/QC system (Huynh et al., 2015; Boukar et al., 2016, 2018). A database of available KASP markers for QC in cowpea is obtainable from Excellence in Breeding platform at https://excellenceinbreeding.org/module3/kasp. However, the effectiveness of these QC markers for large scale detection of F₁ hybridity in cowpea and the rate of success in obtaining true F₁s in highly productive breeding programs are not well-documented. The objective of the present study was to evaluate the genetic relationships among cowpea parental lines and determine the hybridity of resultant putative F₁ progenies for further advancement in the program using KASP-based SNP assay. The study highlights the significance of molecular marker deployment for QC in cowpea breeding and corroborates the efforts toward optimization of the breeding program for increased genetic gain.

The demands of mankind for food, feed and various agricultural products have been increasing. To develop products that meet this increasing demand, breeders are challenged to accelerate genetic gain continuously, at ever higher rates, while they bridge the gaps that remain between the yield potential in breeders demonstration trials and the actual yield in farmers’ fields (Xu et al., 2017). The use of molecular markers has been demonstrated to play key roles in breeding optimization to attain enhanced genetic gain, mainly, through improved precision and reduction in time through more rapid and efficient breeding cycles (Perez-de-Castro et al., 2012; Sussumu Sakiyama et al., 2014). The past 15 years has seen key advances in development of cowpea genetics and genomics resources (Singh, 2007; Muchero et al., 2009a; Boukar et al., 2016, 2018; Munoz-Amatriain et al., 2016; Muñoz-Amatriaín et al., 2017; Fatokun et al., 2018; Huynh et al., 2018; Lonardi et al., 2019). This has facilitated the discovery of molecular markers for traits discovery and deployment in the breeding programs. For example, a number of studies have identified quantitative traits loci (QTLs) associated with resistance to biotic and abiotic stresses in cowpea (Muchero et al., 2009b; Agbicodo et al., 2010; Lucas et al., 2013; Huynh et al., 2015; Benoit et al., 2016; Essem et al., 2019; Miesho et al., 2019). However, although the application of molecular markers in hybridity testing and hybrid purity assessment have been demonstrated in several other crops (Patella et al., 2019a; Kishor et al., 2020; Osei et al., 2020), there is no documentation of its deployment in cowpea. Our work, therefore, is intended to integrate SNP markers as a QC mechanism against the unintended self-pollination or cross contamination following hybridization. This study has demonstrated that the strategy is also effective in cowpea. This is of paramount importance considering the significance of cowpea in food and nutrition (Kebede and Bekeko, 2020) and the need for constant and rapid development of new varieties. Released varieties of cowpea are often pure lines given its autogamous nature. Manual pollination between homozygous parents carrying agronomic traits of interest is, therefore, required to introduce genetic variability required for improved variety development. Despite the efficiency of hand emasculation method developed in the past (Myers, 1996), it remains difficult to rapidly differentiate true F₁ seeds from self-pollinated ones in cowpea and yet this is a crucial step in the breeding process. The use of molecular assays to quickly and accurately screen progeny makes it possible to overcome some of the potentially frustrating occurrences in conventional breeding. Our KASP-based SNP markers have facilitated routine use of low-density fingerprinting of cowpea parental lines and allowed accurate detection of hybrids amongst their progenies.

This study began by assessing the informativeness of 17 QC cowpea SNP markers based on allele frequencies and PIC. The minor allele frequencies refer to the proportion of the second-most-common alleles at a genetic locus in a population and play a key role in determining allele diversity and heritability in the population (Dussault and Boulding, 2018; Hernandez et al., 2019). Dussault and Boulding (2018) reported that markers with high minor allele frequencies have high resolving power in distinguishing between individuals. PIC on the other hand, measures the ability of a marker to detect polymorphisms, hence, it determines the quality of a molecular marker for genetic analysis (Serrote et al., 2020). Results from our study revealed that 16 out of the 17 (94%) markers had minor allele frequencies of >0.05 and PIC above 0.2. Average level of heterozygosity was 0.29 when both parents and F₁ progenies were considered together but this measure dropped drastically to a mean of 0.02 when parents were considered alone, which suggested that the SNPs had the power to distinguish heterozygotes from homozygotes. Overall, the estimates of these marker properties implied that the 17 SNP marker panel used were informative, that is, they were capable of distinguishing between parents, henceforth, identify true hybrids. These outcomes were supported by the estimates of marker efficiency which revealed only one marker (snpVU00007) with the lowest marker efficiency of about 3.4%.

We continued to assemble SNP fingerprints of parental lines and used them to dissect the genetic relationships among parents and the F₁ progenies to discern which crosses benefited from maximum diversity between the parents. Analysis of genetic relatedness among plant germplasm plays a key role in plant breeding as it allows effective characterization of parental lines for hybridization (Govindaraj et al., 2015; Mafakheri et al., 2017; Khanna, 2018; Nkhoma et al., 2020). In the present study, genetic similarity analysis placed the parental lines into three groups, with a modest amount of variation still detectable within the groups. The landraces were grouped together in Cluster I, however, when considered alone, a considerable diversity was detectable among them and they were clustered into three sub-groups. The clustering of the landraces together when the all lines were considered is likely due to the fact, they are quite unrelated to the other parental lines, and as such the cladogram placed them in the same sub-group. Interestingly, a number of the parental lines shared the same cluster with the landraces, and therefore, they potentially carry some beneficial adaptation traits. As expected, the two male parents IT99K-573-1-1 and IT99K-573-2-1, showed 100% similarity and were grouped together with several other parents in the same cluster. Given that the rest of the parental lines were females, hybridizations between all closely related females and the two sister male lines is likely not beneficial. The parental similarity analysis, therefore, allowed us to identify progenies from crosses that involved closely related parents. Most of the female parents were selected from IET, PYT, and AYT based on their yield potentials and quickly recycled into the hybridization program. However, it was expected that some of these parents would be closely related since they shared common pedigrees. The fact that crosses were made prior to genetic similarity analysis, results from the present study provided a QC tool for early discarding of crosses with closely related parents. Representative crosses can be selected based on parental clustering and advanced in the breeding program. While comparing the F₁ and their parents, the first two principal components revealed a considerable overlap between the F₁ and parents. However, the F₁ progenies appeared to be more closely related to each other while the parents were highly divergent. This outcome is expected given that all the F₁ progenies were technically half-sibs, having received a common set of alleles from the two sister lines (IT99K-573-1-1 and IT99K-573-2-1). The expected variation for effective selection at later generations, therefore, comes solely from the diversity among the female parents. It is important therefore, to sort the crosses to be advanced based on relationships among parents.

A final focus was made on the molecular assessment of hybridity of the F₁ plants derived from multi-parental crosses. Molecular determination of hybridity has been successfully demonstrated in various crops; tomato (Osei et al., 2020), maize (Salgado et al., 2006), lettuce (Patella et al., 2019a), Capsicum (Mongkolporn et al., 2004), rice (Yashitola et al., 2002), cotton (Dongre and Parkhi, 2005), cassava (Vincent et al., 2014), and peanut (Gomez et al., 2008). Apart from that on tomato, these studies deployed microsatellite markers for hybridity testing as a QC to determine true F₁ for advancement. The present study, utilizing 17 KASP-based SNP markers, determined the homozygosity of parental lines and successfully identified true F₁s. We found that 6% of the parents were unstable (heterozygous) at 2–8 loci, however, their inclusion as parents in hybridization was based on their yield potential and several other farmer-preferred attributes. Most of the parents detected as heterozygous were recycled at IET (F₅ generation), PYT (F₆ generation), and AYT (F₇ generation) stages, while two of them were landraces. Heterozygosity observed among parents could be attributed to residual heterozygosity, outcrossing and heterogeneity (Plaschke et al., 1995; Fatokun and Ng, 2007; Gioi et al., 2012). To distinguish between F₁ individuals resulting from cross-pollination and those resulting from self-pollination, we first examined the informative loci among the parental lines used in the crosses. Only polymorphic loci between pairs of parents were considered informative and used in hybridity assessment. Among the cross combinations, our analysis showed a range of 0–11 informative markers out of the 17 SNPs, 4–6 being the most frequent (70 parental combinations). Overall, 94% of the parental combinations had more than two informative markers, allowing for the authentication of putative hybrids from these cross combinations. Patella et al. (2019a) reported that restricting this kind of analysis to the informative loci brings considerable savings in terms of time and costs. All in all, molecular assessment of hybridity was successful: Over 72% of the crosses yielded 100% true F₁ progenies (100% success rate) while 8% had unsuccessful F₁ offspring (Selfed progeny). In lettuce, a highly self-pollinated crop, 100% success rate was reported in only 30% of the crosses (Patella et al., 2019a). Considering all the 1,436 F₁ plants developed through manual hybridizations and tested in the present study, a success rate of 79% was attained. In peanut, out of 179 putative F₁s, 70% were detected by SSR marker to be true hybrids (Gomez et al., 2008). To our knowledge, the present study is the first report on hybridity testing using molecular makers in cowpea. The crossing success rates of 10–20% reported by Myers (1996) and 28.36% by Ajayi et al. (2020) in cowpea are not necessarily the percentage of true F₁s obtained, but the percentage of hybridizations that survived mechanical handlings and other stresses during manual emasculation. Our study found that of the hybridizations that survived mechanical handling during hand emasculation, 14% were a product of self-pollination. We also found that the hybridity status of about 1% of the progenies could not be determined because all the 17 markers were not polymorphic between their parents. This lack of polymorphism was observed in about 12 parental pairs and this could be attributed to the genetic similarity among parents. In fact, when we checked the pedigrees of lines involved in these parental pairs, we found that 9 out of 12 had 573-1-1, 573-2-1, and IT97K-499-35 as 1 of their parents. Interestingly, all the three lines (IT99K-573-1-1, IT99K-573-2-1, and IT97K-499-35) share a common striga resistant parent IT93K-596-9-12 (Singh et al., 2006). The remaining three lines shared TVu450 genome, a landrace with unknown pedigree, as one of their parents. These observations provide a plausible reason for lack of polymorphism between these lines and 573-1-1/573-2-1 (the two sisters lines used in this study). The limitation of detecting differences among closely related lines could be minimized by increasing the number of markers for hybridity tests and using genetically diverse parents. Usually, 20–30 SNP markers that are carefully chosen to have good genome coverage is sufficient (PASSeL, 2019). However, different marker densities, some with lower numbers than what is employed in the present study have been used to successfully dissect hybridity and fingerprints in crop plants (Nandakumar et al., 2004; Romdhane et al., 2018; Josia et al., 2021). According to Josia et al. (2021), the common assumption that the use of a large number of markers results in higher accuracy should be considered carefully due to the fact that in most sequence-based marker systems, the levels of missing data can lead to wrong interpretation. Therefore, the selection of fewer markers with high and repeatable representation across samples is desired and is cost-effective. Chen et al. (2016) also noted that fewer markers with high expected heterozygosity, missing value of <20%, and observed heterozygosity of <6% are ideal markers for accurate QC genotyping. The present study used 17 informative SNPs and we included as control, known sister lines and landraces which are quite diverse from some of the elite breeding lines. The fact that these markers were able to consistently diagnose these differences and further delineate the F₁s from the homozygous parents demonstrated the effectiveness of the 17 QC marker set for fingerprinting and hybridity testing of cowpea.

The results of this study contribute to the body of knowledge about the advantages of integrating molecular markers as a QC step in a breeding program. We used KASP-based SNP markers in the first breeding phase which involves parental selection and identification of true hybrids. We demonstrated, firstly, the efficiency of SNP markers not only in detecting the best parental lines for crossing based on their observed homozygosity and genetic relatedness, but also in screening the resulting F₁ progeny to distinguish between the offspring resulting from cross-pollination and those resulting from self-pollination. The markers identified true hybridity with 100% success in 72% of the populations. Overall, we detected 79% of the putative F₁ plants to be true hybrids while 14% were selfed plants. Furthermore, given the ability of the SNP marker panel to detect heterozygosity or admixture in the parental lines, these markers can be utilized downstream the breeding scheme to assess the uniformity of potential candidate or pre-commercial cowpea lines. The fact that these sets of SNP markers are reproducible through the KASP assays, the genotype data from this study constitute a molecular fingerprint of the parents that would be utilized for future identity and purity assessment. The markers could also be used by seed companies to assess newly developed varieties for distinctiveness, uniformity and stability (DUS), which are the three key necessities for registering plant materials (Patella et al., 2019b).

The original contributions presented in the study are publicly available. This data can be found here: European Nucleotide Archive (https://www.ebi.ac.uk/ena/browser/home) database under accession number: PRJEB46458.

PO wrote the manuscript, organized samples for genotyping, and conducted the statistical analysis. CF, OB, AT, and SS revised the manuscript. OO, MA, IJ, and GB organized and coordinated pollination and tissue sampling for DNA analysis. OB and CF developed the populations and designed the study. All authors reviewed and approved the manuscript.

SS is employed by the company Bayer Research and Development Services LLC. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.