Potential of genomics for the improvement of underutilized legumes in sub‐Saharan Africa

Underutilized, or orphan legumes, are widely distributed across farming landscapes in sub‐Saharan Africa (SSA) but often have low yields and do not fulfill their potential due to very limited research, breeding, development, marketing, and awareness of their benefits. These advantages include nutritional quality and climate resilience. In this review, we focus on Bambara groundnut, African yam bean, and Kersting's groundnut. Knowledge of the challenges and rewards of exploiting them will provide opportunities for concerted approaches to their revival and contribution to future global food systems, especially in the context of climate change. This review identifies the institutional and noninstitutional challenges, the constraints, the prospects, and the rewards that can be derived from exploiting orphan legumes in SSA. The genetic resources center (GRC) of the International Institute of Tropical Agriculture (IITA) conserves a diverse collection of about 2500 accessions of these crops with the majority from Africa. In this review, we focus on the ex situ conservation of the genetic resources of these indigenous African legume crops, their characterization and evaluation, prospects for the development of improved cultivars, and the role they could play, particularly with respect to nutrition and adaptation to climate change. We emphasize progress made in recent years concerning the assembly of information required for application of genomics tools to these crops and how this will underpin the development of improved varieties.

given in Table S1. Substantial numbers of the collections of Bambara groundnut (19%), African yam bean (97.4%), and Kersting's groundnut (100%) were collected from Nigeria. GRC, with partners, is studying morphological and yield related traits and climate adaptive traits in Bambara groundnut and African yam bean. Molecular characterization is also in progress for genetic diversity studies and quantitative trait loci (QTL) discovery for drought and other useful traits including nutrition in these two crops.

| AFRICAN YAM BEAN
African yam bean (Sphenostylis stenocarpa Hochst Ex A. Rich Harms) is an underutilized indigenous legume of sub-Saharan Africa. It is an annually prostrate or climbing vine that produces both nutritious seeds and tubers ( Figure 1). Okigbo (1973) and Duke (1981)  Its seed, tubers, and leaves are nutritionally rich. It is mainly grown West Africa but also in East and Central Africa for its tubers.
The fully matured pod houses 20-30 seeds depending on genotype.
Seeds of African yam bean are usually brown, cream, and orange brown or mottled and ovoid in shape. African yam bean is interplanted with yams (Dioscorea spp.) and some other vegetables in village settings (Ezueh, 1984). The number of days to flowering in African yam bean ranges between 80 and 130 days after planting whereas physiological seed maturity ranges between 150 and 300 days, depending on the genotype. Seed yield varies per location, as an example, from 63 lines evaluated at the International Institute of Tropical Agriculture (IITA), Ibadan, Western Nigeria, the best line yielded 1860 kg per ha whereas at Nsukka, Eastern Nigeria, a seed yield of about 2000 kg per ha has been recorded. The spindle shaped tubers usually occur in small quantities and range from 5 to 7.5 cm length, weighing between 50 and 300 g each (NAS, 1979).
A number of reports have confirmed that the tuberous roots of African yam bean are a good source of carbohydrates in West Africa (Ezueh, 1984;Okigbo, 1973;Potter & Doyle, 1992). The underutilization of African yam bean may be due in part to some production and utilization constraints, such as long duration of cooking, low yields, antinutritional factors, and long maturity period (Nnamani et al., 2017;Ojuederie, Balogun, & Abberton, 2016).
GRC conserved 456 accessions of African yam bean, of which 444 were collected from Nigeria. This collection needs to be expanded to become more comprehensive particularly to include accessions from other African countries, especially Ethiopia, which was the center of origin with broad genetic diversity. GRC has assessed the variation in 127 of 456 African yam bean accessions. Balanced nutrition, feed for animals, N fixation, and can grow in marginal soils. (Potter & Doyle, 1992) Bambara groundnut Vigna subterranea (L.) Verdc. Passport data for these accessions is shown in Table S2. Initial characterization was carried out according to the established crop descriptors at GRC, IITA (Table 2).
Bambara groundnut is a crop that survives harsh weather. It originated in West Africa and has a growing cycle and harvest time ranging from 4 to 6 months depending on genotype and end use ( Figure 2). It is considered to be the third most important legume in sub-Saharan Africa after cowpea and groundnut (Linnemann & Azam-Ali, 1993).
Bambara groundnut has low commercial value, but its nutritious green and matured pods are sought after by farmers for household consumption. There is variation between landraces in the growing degree days to maturity and many other physiological traits.
Bambara groundnut seed is regarded as a very well balanced food because of its nutrient composition. Its protein content ranges between 18% to 24% , and it is rich in eight of the nine essential amino acids, including lysine, methionine, isoleucine, leucine, threonine, phenylalanine, and valine, whereas tryptophan is the limiting amino acid (Yao et al., 2015). Carbohydrate content ranges between 51% and 70% (Halimi, Barkla, Mayes, & King, 2019).
The gross energy value of its seed is higher than any other legume crop (Anchirinah, Bennet-Lartey, & Yiridoe, 2001;Feldman, Ho, Massawe, & Mayes, 2019;Rowland, 1993 & Linnemann, 2018). Nigeria is the largest producer of Bambara groundnut in Africa with an average of 0.1 million tonnes. Bambara groundnut yield (t ha −1 ) in Africa varies between land races and locations (0.5-3 t ha −1 ) with yield potential of above 3 t ha −1 (Massawe, Roberts, Azam-Ali, & Davey, 2003). The crop has an average yield of 0.85 t h −1 , which is comparable to some other legume (Begemann, 1988). It is a source of protein and fiber and nutritionally complements cereal crops (Massawe, Mwale, Azam-Ali, & Roberts, 2005). Adu-Dapaah, Berchie, Amoah, Addo, and Akuamoah- Boateng (2016) reported that Bambara groundnut milk has 15%-16% more protein than soymilk. In recent years, interest has begun to increase in Bambara cultivation and consumption, possibly due to its potential as a food crop that thrives in dry areas, although Bambara still lacks adequate seed systems and best agronomic practices (Mubaiwa et al., 2018). The effort required to prepare Bambara groundnut seeds for meals in many countries is increased by the long cooking time required, which translates into increased cost of cooking fuel. Consequently, this is one of the major bottlenecks identified for the consumption of Bambara groundnut (Adzawla et al., 2016). The "hard-to-cook" phenomenon can also be influenced by seed storage conditions, particularly heat and humidity. Several other theories have been put forward to explain components of this storage trait (Mubaiwa, Fogliano, Chidewe, & Linnemann, 2017

Kersting's groundnut (Macrotyloma geocarpum [Harms] Marechal and
Baudet) belongs to the Fabaceae family of the Phaseoleae tribe (Amuti, 1980), and it is grown as an annual herbaceous and geocarpic legume crop ( Figure 3). It is also known as Hausa groundnut or ground bean. It is predominantly cultivated in the arid and semi-arid regions of West Africa (Aremu, Olaofe, & Akintayo, 2006). Kersting's groundnut has a very high adaptation to dry areas as it can thrive with rainfall as low as 500-600 mm, well distributed over 4-5 months (Mergeai, 1993). Dako and Vodouhe (2006) reported that Kersting's groundnut originated from either Northern Togo or central Benin.
Cultivation spans countries in West Africa including Burkina Faso, Nigeria, Ghana, Mali, Togo, and Benin. The crop is also cultivated outside West Africa in Tanzania, Mauritius, and Fiji (Dako & Vodouhe, 2006). The morphological diversity of the seed is reflected in seed color variation (Pasquet, Mergeai, & Baudoin, 2002). Kersting's groundnut cultivation is being carried out on a small scale for home consumption by few farmers despite its potential importance as a protein source, with better adaptation to local climatic conditions and lower production costs than cowpea. Despite the minimal cultivation of Kersting's groundnut, there is a long history of cultivation by local farmers (Mergeai, 1993), but information about its agronomic and genetic potential is very scarce ( (Mergeai, 1993). There are no available statistics on the global harvested area, yield, and production of the crop.
The major factors affecting cultivation of underutilized crops in general also have a negative impact on Kersting's groundnut, including low seed yield, small grain size, labor-intensive cultivation and harvesting, low market value, and lack of improved cultivars. There has been a reduction in cultivation of this crop in many growing areas (Amujoyegbe, Obisesan, Ajayi, & Aderanti, 2010;Tamini, 1995).
Gradual extinction of this crop from production systems was observed in Ghana (Bampuori, 2007), Togo (Mergeai, 1993), Burkina Faso (Tamini, 1995), and more recently Nigeria (Amujoyegbe et al., 2010) and Benin (Assogba et al., 2015). It is becoming necessary to consider defining and including the current status of this legume species in the International Union for Conservation of Nature (IUCN) red list. The traditional utilization of this food crop and its inclusion in various ceremonies by some ethnic groups could be an advantage that will help its conservation and prevent its total neglect in such areas.
A total of 22 accessions of Kersting's groundnut were collected from Nigeria and conserved at GRC. It shows a geographical diversity gap and need to collect additional material from other African countries including Northern Togo or Central Benin to provide adequate representation of the geographical distribution of genetic diversity and to prevent genetic erosion.

| CHALLENGES OF EXPLOITING ORPHAN LEGUMES FOR SSA
Up to now, around 240 whole plant genomes have been sequenced with varying quality (Chen et al., 2018). Single genome sequencing is not adequate to capture complete genetic diversity within a species It will speedup trait mapping in orphan crops to mine essential agronomical genes and incorporate important genes to develop improved lines for harsh and extreme climate conditions.
In many parts of SSA, climate change has led to severe consequences with respect to food security and income generation for smallholder farmers with little or no mitigation measures (Edgerton, 2009;Fisheries, 2014). In general, SSA is considered to be vulnerable without a strong capacity to manage the impacts of extreme weather conditions (Chauvin, Mulangu, & Porto, 2012;Edgerton, 2009).
According to Wezel et al. (2009)  Genetic diversity in underutilized crops has been evaluated using different types of molecular markers, for example, isozymes (Howell, 1990;Pasquet, Schwedes, & Gepts, 1999) Resequencing of underutilized crops is also in progress (Prasad et al., 2019). The Bambara groundnut genome was sequenced using high-density Illumina short-read data, which means that complete pseudochromosomes are not yet available. There are also plans to sequence the Bambara groundnut genome using long-read sequence data (Gregory et al., 2019). The total genome size of Bambara groundnut is 550 Mb as compared to 640.6 Mb of the cowpea genome (Lonardi et al., 2019). The number of protein-coding genes identified was 31,707, which is higher than mung bean (22,427) and lower than adzuki bean (34,183; Chang et al., 2018b). In total, 98.0% of the Bambara groundnut genome was functionally annotated (Chang et al., 2018a). A total of 605 gene families were identified in Bambara groundnut and its paralogs linked mainly with glyoxylate and dicarboxylate metabolism, zeathin biosynthesis, and carbon fixation (Chang et al., 2018a(Chang et al., , 2018b. Gene ontology (GO) analysis revealed that the paralogous genes of Bambara groundnut were enriched in ion binding (Chang et al., 2018a). Africa and one in South-East Africa, using both SSR markers (Molosiwa, Aliyu, & Stadler, 2015;Somta et al., 2011) and DArT markers (Stadler, 2009 (Paliwal et al., 2019). Three distinct population groups from West Africa, Central Africa, and East Africa were found using principal component analysis (Paliwal et al., 2019). The majority of wild relatives were grouped with West African accessions (Paliwal et al., 2019), confirming that it is an indigenous West African crop. A genetic relationship study was carried out for 56 Indonesian Bambara Groundnut with 114 diverse accessions from Africa to investigate their origin with SSR and SNP markers by Redjeki et al. (2020). They suggested East Java Bambara groundnut lines could be introduced from West Java materials. The current Indonesian accessions, most probably introduced from South Africa, showed a small fraction of the genetic variability within the species (Redjeki et al., 2020). The narrow genetic base of Indonesian lines could be extended by introducing lines from other geographical regions like West Africa and East Africa.
They reported three major distinct subpopulations, which coincided with their geographical regions (Central and Western Africa, Southern and Eastern Africa, and Indonesian collection). The genetic diversity study of Bambara groundnut indicates geographical-specific selection and founder effect, which could play a significant role in influencing its genetic diversity (Redjeki et al., 2020). The extensive degrees of allelic diversity of different geographical population could be potentially used for developing improved cultivars. With this available diverse regional genetic collection from across Africa, and its wild relatives, GRC has an excellent resource for future crop improvement.   and SSR (Shitta et al., 2015) markers. The accessions used in these studies were grouped in three to four clusters in both genetic diversity Nnamani et al., 2019;Shitta et al., 2015) and population structure (Ojuederie et al., 2014;Nnamani et al., 2019;Ojuederie et al., 2014). These studies showed wide variation is available in the African yam bean collection and can be useful for future breeding research. A genetic diversity study of over 250 accessions of African yam bean was carried out using DArTseq SNP markers (Paliwal et al., 2019) at GRC, IITA. The proportion of heterozygosity was higher in the less grown landraces than in the most popular ones (Paliwal et al., 2019). All accessions were grouped into six different clusters (Paliwal et al., 2019) which revealed the existing genetic diversity of the collection and can be use by breeders to select diverse lines.
Genetic diversity research on Kersting's groundnut has been reported with isozyme markers that were used for the characterization of eighteen domesticated and two wild accessions from Togo and Burkina Faso by Pasquet et al. (2002). They found a narrow genetic base, which is not a favorable sign to increase genetic gain in breeding programs (Spillane & Gepts, 2001). The transferability of 12 SSR markers of cowpea to Kersting's groundnut was reported, with nine SSR monomorphic (Mohammed et al., 2018) A total of 10.6% of these SNPs were aligned on the reference genomes of adzuki bean and mung bean, indicating an evolutionary relationship of Kersting's groundnut with adzuki bean and mung bean (Akohoue et al., 2020). The population of Kersting's groundnut was grouped into four distinct clusters based on seed coat color (white-seed, red-seed, black-seed, and white with black eye seed color) using the unrooted neighbor-joining tree method. In contrast, it was grouped only into two populations (K = 2) in admixture-model based clustering method (Akohoue et al., 2020). The expected heterozygosity (He) ranged from 0.01 to 0.09 within the clusters, indicating narrow genetic diversity between within clusters (Akohoue et al., 2020). DArTseq GBS-SNP genotyping has also been initiated for Kersting's groundnut at GRC and will be used for genetic diversity and population structure analysis.
With these results, parental lines can be selected for crossing to develop transgressive segregation lines for desired traits and for the development of linkage maps and QTL discovery.
Genetic linkage mapping and QTL discovery for important traits are valuable for any genomic assisted breeding program (Collard et al., 2005). Biparental mapping populations have been used for linkage mapping, which is a classical method for discovery of QTL discovery (Collard & Mackill, 2008). Many QTL discovery studies on different traits using this method have been reported, including on legumes (Emebiri et al., 2017;Hong et al., 2010;Saxena et al., 2012).
However, little research has been carried out for linkage mapping and QTL discovery in underutilized crops. No linkage maps or other QTL discovery research has become available in African yam bean or Kersting's groundnut, and in Bambara groundnut development of linkage mapping and QTL discovery research was only recently initiated (Ahmad et al., 2016;Ho et al., 2017). In the first report, Ahmad LGs with 293 markers and covered 1376.7 cm genetic distance (Ho et al., 2017). Both linkage maps showed synteny with common bean, adzuki bean, and mung bean (Ho et al., 2017). A total of 36 QTL have been discovered in Bambara groundnut for different traits, including internal node, days to emergence, growth habit, seed weight, pod length, and width (Ahmad et al., 2016). The phenotypic variation of these 36 QTLs ranged from 11.6% to 49.9% (Ahmad et al., 2016). There were only two stable QTL discovered for growth habit (Ahmad et al., 2016) and internode length (Ahmad et al., 2016;Ho et al., 2017). The development of biparental mapping populations of Bambara groundnut, African yam bean, and Kersting's groundnut is in progress for QTL discovery of different traits in GRC. However, the lack of ability to easily cross between Bambara groundnut genotypes remains a constraint to both breeding and linkage mapping.
With the advancement in NGS and computational technology, genome-wide association studies (GWAS) have been utilized as a robust tool for QTL discovery in different crops (Buckler et al., 2009;Schlappi et al., 2017;Yan, Warburton, & Crouch, 2011;Zhou et al., 2015). GWAS can access marker-trait association for QTL discovery in thousands of genotypes using millions of genome-wide SNP markers that provide high genetic resolution in the kilobase range.
The accessibility of large SNP marker datasets of reference or nonreference genome crops has opened another way for QTL discovery using natural populations with the GWAS approach. Biparental mapping needs time and considerable expense. High-density SNP data have been generated in underutilized crops populations in GRC, and its use for GWAS is in progress. By combining phenotypic and highdensity genotypic data, GWAS can overcome several of the limitations of linkage mapping and provide a powerful complementary strategy to discover QTL for complex traits (Schlappi et al., 2017;Yan et al., 2011). In Kersting's groundnut, a GWAS study reported a significant marker-trait association of 10 SNP markers with six different agronomical traits (grain yield/plant, 100 seed weight, days to 50% flowering, days to maturity, number of seeds/plant, and number of pods/plant). The marker M1 was significantly associated in two locations with 100 seed weight and contributed to over 24% of the phenotypic variation (Akohoue et al., 2020). Identification of QTL for phenotypic traits of underutilized crops can play an important role in increasing the speed and efficiency of developing improved climateresilient cultivars of underutilized crops using MAS. A population of 420 accessions of Bambara groundnut was developed from indigenous farmers, breeder seed, and genebanks by crop for future (CFF) and is being characterized for both phenotypically and genotypically to use for GWAS analysis of important agronomical traits (Muhammad et al., 2020). Bambara groundnut and African yam bean at GRC have been characterized and evaluated for morphophysiological traits, yield, nitrogen fixation, the nutritional quality of tuber and seed, and drought tolerance. GWAS of these traits in Bambara groundnut and African yam bean are in progress.
Knowledge of genetic changes that occurred during both biotic and abiotic stresses in underutilized crops at the RNA level is limited.
With the advancement of NGS, transcriptome analysis using RNA-sequencing can characterize genes for complex traits, including differentially expressed genes (Wang, Gerstein, & Snyder, 2009). For functional genomics studies, gene expression analysis under specific stress conditions can be used, especially in underutilized crops, with limited information available for the genomes (Afzal et al., 2020). SSR markers developed from expressed sequence tags (EST) known as genic SSR markers are not as abundant and as polymorphic as genomic SSR. Genic SSRs are highly conserved, which provides a high degree of transferability across related species (Mathi Thumilan et al., 2016;Xiao et al., 2016). Genic-SSR markers may provide a higher probability of marker-trait association with functional candidate gene (Ukoskit et al., 2019) than other SSRs. In cowpea, the total RNA of stems, roots, and leaves of five seedlings was sequenced to identify differentially expressed genes and develop SSR markers (Chen et al., 2017). Thus, RNA sequencing can establish genic-SSR markers which could be useful for genetic diversity and QTL identification. By using high density SNP markers, a significant QTL (QRk-vu9.1) was identified for root-knot nematode resistance in cowpea by Santos, Ndeve, Huynh, Matthews, and Roberts (2018). The transcription factors "MYC, WRKY protein, and DREB" which have a role in water-deficit response in crops (Bartels & Sunkar, 2005;Chen et al., 2002) were absent in the Bambara groundnut dataset (Stadler, 2009).
Genome-editing offers to look beyond traditional and molecular plant breeding and can play a significant role in mitigating the adverse effects of yield constraints of crops, especially under climate change scenarios.
Genomic selection (GS) is a genomic-assisted breeding method in which a large dataset of SNP markers is used. The use of molecular markers in different genomic-assisted breeding approaches to select the desired traits has been a popular area of crop research in past decades, but their utility in the genetic dissection of complex quantitative traits has been limited (Bernardo, 2008;Harris et al., 2015). In the marker-assisted selection (MAS) method, introgression of few major QTL is straightforward. However, as the number of QTL increases for introgression, the deployment of MAS in crop improvement becomes complicated and costly, which limits the expected achievement of MAS. An advantage of GS is that it enables selection based on a large marker dataset that can densely cover the whole genome to assure that all relevant genes are expected to be in LD (linkage disequilibrium) with at least a few of the markers (Bhat et al., 2016;Heslot, Jannink, & Sorrells, 2015). GS uses a prediction model to minimize the effects of biased markers (Heffner, Sorrells, & Jannink, 2009) by using all associated markers to estimate the breeding values for a trait. The majority of complex traits are quantitative and governed by minor QTLs (Desta & Ortiz, 2014), which govern the majority of phenotypic variation of a trait, including epistatic effects (Deshmukh et al., 2014).
Because GS can capture both small effects of QTL and epistatic interaction effects, it could play an important role in enhancing genetic gain. This approach has been used for improving different complex traits like forage yield data , quality traits (Biazzi et al., 2017) in alfalfa, and cyst nematode resistance (Bao et al., 2014), grain yield (Jarquin et al., 2014), and seed weight (Zhang, Song, Cregan, & Jiang, 2016)   Orphan legumes possess significant attributes in contributing to the generation of income for smallholder farmers and supporting food security. If utilized, they can support low input agricultural systems of farming and respond adequately to biotic and abiotic shocks (Vanlauwe et al., 2014). Orphan legumes are being intercropped in some areas as they have been identified to play special roles in vulnerable regions in SSA (Snapp, Jones, Minja, Rusike, & Silim, 2003).
An element of strategies for technology delivery and increasing support to small holder farmers is increasing the availability and quality of seeds (Shiferaw, Kebede, & You, 2008). Several countries in SSA have developed policies that ensure the safe distribution and use of only approved seeds for sales to farmers. Notwithstanding, as in the case of Tanzania, the community has also been involved in the sales and distribution of seeds to reduce the challenge of acute shortages (Abate et al., 2012).
Overall, the public and private sectors must join together to ensure adequate crop improvement plans, while also emphasizing management and distribution of plant genetic materials to end users (Minot et al., 2007).

| CONCLUSIONS
In this review, we identify challenges and rewards associated with the exploitation of orphan legumes for SSA. The potential of these underutilized indigenous legumes is still largely untapped but could become a valuable element of enhanced food and nutrition security in SSA.
The available diverse genetic resources for these legumes provide an opportunity for research to explore their potential for food security, employment and income generation, particularly for smallholder farmers of SSA. Sustained research investments, nutritional awareness, and genetic characterization of these underutilized indigenous legumes are prerequisites for exploring their full potential to ameliorate climate impacts.
Some of these legumes still require an expansion of available genetic resources to avoid genetic erosion of their diversity. For example, the majority of African yam bean and Kersting's groundnut in the collection come from Nigeria, which indicates gaps in the GRC collections of these two legumes. It will be necessary to broaden genetic diversity and geographical representation of these collections.
Awareness of these crops and traditional knowledge about their use and cultivation still lie in aged farmers' hands in most rural areas where the crops are grown.
Advanced breeding research on these underutilized crops is far behind other legumes crops, such as soybean, groundnut, chickpea, cowpea, and pigeon pea, but the breeding efforts for important and complex traits of these legume crops can be hastened by applying cutting edge genomics and phenotyping tools. The use of genomic-assisted breeding can help unravel the genetic potential of these legumes for their improvement and open up further research opportunities.

CONFLICT OF INTEREST
The authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS
Writing-original draft and review and editing: MA, RP, and TTA.
Writing-review and editing: BF and OO. All authors read and approved the final manuscript.

FUNDING INFORMATION
The work of GRC is funded by Global Crop Diversity Trust and the CGIAR. The funders had no association in the present study design, data collection, and publication decision.