Kersting's groundnut is a small legume species with a maximum spread of 50 cm and a height of up to 40 cm. Unlike other geocarpic legumes such as peanuts, KG branches are coiled and interspersed in a spiral form lying above the ground. On the other hand, the leaves are erected on the branches, giving the plant a bushy growth habit. Kersting's groundnut accessions were characterized and evaluated for various agromorphological traits (Bayorbor et al., 2010; Assogba et al., 2015; AVRDC, 2015; Akohoue et al., 2019) (Table 2). Major descriptors included growth habit, flowering and maturity times, plant height, spread diameter, grain yield, and yield components. All morphological markers were reported to be significantly affected by environmental factors, except growth habit traits (Adu-Gyamfi et al., 2012; Assogba et al., 2015; Akohoue et al., 2019). Adu-Gyamfi et al. (2012) reported significant variation among the White mottled with black eye (White), Black, and Brown (Mottled) landraces of Ghana, based on their agro-morphological performance. Similarly, in Benin, the agromorphological evaluation showed that the White, Black, and Red landraces were significantly different for agronomic performance (Assogba et al., 2015). Still, in Benin, Akohoue et al. (2019) analyzed the diversity in four landraces of KG (White, Black, White mottled with black eye, and Red) using morphological markers and found four clusters based on genotypes performance. The first three clusters were mainly composed of the White landrace while the fourth cluster included the other coloured-coat ones. In terms of performance, individuals in clusters 2 and 4 exhibited higher performance and were intermediate and early maturing genotypes, respectively. Although there is a relatively increasing genetic and phenotypic data on traits, in-depth phenotypic characterization through multi-trait and multi-environmental trials should be conducted using the available germplasm, as a whole, to shed light on the trait variations in the characterized germplasm as well as the performance of landraces grown by farmers. In addition, investigating the response of these landraces under biotic and abiotic stress conditions could be relevant for the improvement of KG. Low yields were obtained in KG by farmers, as well as by researchers, hence, breeding activities must focus on improving yield and tolerance to biotic and abiotic stress factors. However, genotype by environment interactions (GEI) affects yield, making it challenging to select genotypes with wide adaptation, resulting in delayed cultivar release (Abady et al., 2019). Crop breeding strategies for higher yield and disease tolerance can be accelerated through the use of high throughput phenotyping (Shakoor et al., 2017). This technique was successfully used in phenotyping groundnut for the total oil and high oleic acid contents (Sundaram et al., 2010; Awada et al., 2018). Although the high-throughput phenotyping technique is an emerging approach, and its application in crop breeding is still very limited, its utilization in KG breeding could be explored. In 2002, Pasquet used isoenzymes to assess the diversity within and between cultivated KG and its wild form. He found a low variability within each group and high genetic divergence between the cultivated and the wild types. Mohammed et al. (2018) assessed the transferability of cowpea-derived Simple Sequence Repeat markers (SSRs) to KG and revealed genetic variability among the landraces studied. More recently, Akohoue et al. (2020) and Kafoutchoni et al. (2021a) applied SNP markers to assess KG genetic diversity and population structure and found low variation within landraces and relatively high genetic distance between landraces. Furthermore, Akohoue et al. (2020) analyzed marker-trait association and genomic prediction accuracy for main agronomic traits of KG. They found markers related to plant morphological traits, flowering time, maturity, yield, yield components, and seed characteristics. Their results also showed low prediction accuracies for yield and related traits and high prediction accuracies for flowering time, maturity, and 100 seeds weight traits. The findings of these different researches showed the existence of genetic variability in KG and provided the first insight into the relationships of phenotype-to-genotype in KG.

Quantifying the biochemical properties in KG can serve as a guide to exploit its potential and benefits for human and animal nutrition. Proximate compositions of KG (Supplementary Table 1) showed that crude protein content varied from 12.90 to 22.95%, while total fiber ranged between 2.01 and 10.90%, and crude carbohydrate of 57.87–81.00%. Results also indicated a low crude fat in KG with a proportion of 1.00–5.29%. The proteins of KG exhibit interesting essential amino-acid profiles (32.7–44.1%) that make the crop attractive for smallholder farmers. It has a higher arginine proportion compared to many other legumes such as Bambara groundnut (0.064–5.48%, Aremu et al., 2017; Oyeyinka et al., 2017), cowpea (3.5–8.52%, Khattab et al., 2009; Eashwarage et al., 2017), common bean (Phaseolus vulgaris L.) (1.17–7.59%, Junkanti et al., 2012; Bouchenak et al., 2013). These findings showed high variability in the levels of nutrients and anti-nutrients in KG, which is potentially due to the biochemical analysis techniques, seed quality, environments, as well as landraces used. Akpavi et al. (2008) compared the proteins and antinutrient contents of two landraces of KG (White and Black) and found a difference between the landraces. Moreover, Badii et al. (2011) suggested that the higher tannins content in the Black and Brown landraces compared to the White ones conferred them more resistance to pulse beetles. Based on these results, we can hypothesize that the varying content of these compounds among landraces is genetically determined. Such genetic variations in KG seed composition offer possibilities for the improvement of related traits through intraspecific crosses. Therefore, accurate information about the proximate and antinutrient compositions of each landrace has become essential for the development of cultivars with high-quality nutrient content.

The development of breeding populations through intraspecific hybridization is required to efficiently address the absence of improved cultivars in KG (Ayenan and Ezin, 2016; Akohoué et al., 2018; Coulibaly et al., 2020; Kafoutchoni et al., 2021a). For a successful hybridization in the crop, the flower biology and structure, as well as the pollination patterns have to be well-understood. Hence, floral anatomy and physiology, floral and fruiting phenology, and reproductive biology were at the heart of the rising research interest in KG. This is timely, as limited knowledge of reproductive biology is a hindrance to the improvement of most orphan crops (Cullis et al., 2019). Such knowledge has represented breakthroughs in the improvement of many plants, thus contributing to the Green Revolution (Singh et al., 2010; Whitford et al., 2013; Jaiswal et al., 2016). For instance, the proper knowledge of the flowering stage at which emasculation should be performed and the adequate time to pollinate is the basis in achieving a high rate of successful hybridization in peanut (Chu et al., 2016), Capsicum annuum L. (Kivadasannavar et al., 2013; García-Tierrablanca et al., 2015).

In the efforts to improve grain yield in KG, plant breeders and geneticists encounter several challenges including the high rate of flower abscission, which represents a major source of continuous failure in attempts to cross KG and low productivity of the crop (Obasi, 1989). As a response to that, three main studies investigated the reproduction systems in KG. Amuti (1980) gave the first description of KG floral biology and deduced from their observations that the crop is a self-pollinated plant with white or purple flowers. Obasi and Ezedinma (1994) focused on floral biology while Kafoutchoni et al. (2021b) went further to study the floral and fruiting phenology, stigma receptivity, pollen viability, and germinability, to devise insights to designing hybridization protocol that guarantees maximum success in the development of breeding lines. Similar research efforts in pigeon pea, which previously had a low success of artificial hybridization (Kalve and Tadege, 2017), has led to the development of hybrid varieties with 25–69% yield superiority over the local cultivars in India (Saxena, 2015).

Allen and Allen (1981) reported that KG has a cleistogamous flower type while Pasquet et al. (2002) classified the crop as chasmogamous. Lord (1981) described the “pre-anthesis cleistogamous” phenomenon as when bud pollination occurs before anthesis, which contributes to increasing the selfing rate. Such a floral structure is known to promote spontaneous selfing (Freitas and Sazima, 2009; Kumari and Sharma, 2017). On the other hand, bud pollination occurs after anthesis in chasmogamous species, which may allow a relatively low rate of allogamy. Based on the changes observed in flower color and size, Kafoutchoni et al. (2021b) described six floral phenological stages followed by six fruiting stages for KG, viz initiated flower (S1), young bud (S2), developed bud (S3), mature bud (S4), opened flower (S5), and wilted flower (S6) for flower development and beginning peg (F1), beginning pod (F2), full pod (F3), beginning seed (F4), full seed (F5), and mature seed (F6) for pod development. Although studies of reproductive biology on KG have provided useful insights into flower and reproductive description and physiology, further investigations are needed to deepen the understanding of cross-pollination processes. Particularly, the information about the species flower category, whether KG is chasmogamous or cleistogamous must be established. A clear knowledge of the pollination of KG would be very useful in determining the handling procedures and the possible strategy for artificial hybridization to increase hybrids' production efficiency.

The small size of flowers, and climate conditions are potential barriers to successful hybridization in KG. These barriers make it necessary to proper timing, skilfully and delicately operating while emasculating and pollinating flowers of KG. Kumar and Singh (2005), reported also the relatively low efficiency of hand emasculation in species with smaller flowers such as sorghum and rice compared to species with larger flowers such as cotton and okra. Moreover, Tamini (1997) showed that variations in weather conditions influence KG flowering cycle, by delaying or accelerating the flowering time. In Bambara groundnut, for instance, temperatures of 33–36°C, adversely affect pollen viability and germination Dhanaraj (2018). The evaluation of thermo-tolerance of pollen is hence recommended before any hybridization activity. However, the physiological effects of environmental parameters on pollen viability and stigma receptivity in KG are not yet known.

In self-pollinating crops, various crossing methodologies such as mechanical emasculation, genetic male sterility, the use of chemical hybridizing agents (CHAs), and genetic transformation have been used for the development of breeding lines and hybrid seeds (Veerappan et al., 2014). However, hand emasculation combined with hand pollination which remains the most widely used technique is tedious, time-consuming, labour-intensive, and costly (Fu, 2014). Moreover, it requires proper skills and delicate operations, especially when flowers are of small size or present several physical and physiological barriers. Therefore, emasculation techniques such as hot-water treatment, anther aspiration (McDonald, 1994), plastic-bag method (Schertz and Clark, 1967), alcohol emasculation, and cold treatment using CHAs, genetic emasculation (Salgare, 2004; Sleper and Poehlman, 2006; Mohammed et al., 2019) need to be potentially explored to cope with pollination issues in KG.

Understanding the appropriate karyotype is important for characterizing genomes of a species and for identifying closely related species (Saensouk and Saensouk, 2018; Senavongse et al., 2018). Kersting's groundnut is a diploid species for which, different numbers of chromosomes were reported for the wild and cultivated types (Pasquet et al., 2002). Miège (1954) evaluated the karyotype of the cultivated KG and found a chromosome number of 2n = 22. However, According to Hepper, (1963), the chromosome number of KG wild type was 2n = 20. More recently, Odo and Akaneme (2021) used six accessions and found a chromosome number of 2n = 20 in the cultivated KG. The clarification in the chromosomal number will certainly be fueling for further investigations particularly relevant to understand the compatibility phenomenon of reproductive organs in intraspecific cross-pollination toward accelerating KG breeding.

Kersting's groundnut is naturally a highly self-pollinated crop (Allen and Allen, 1981) with a narrow genetic basis (Pasquet et al., 2002), suitable for pure-line cultivars development (Figure 3). Improved varieties can be developed mainly through mass selection, pedigree breeding, and backcross methods, using the available landraces. Landraces are a valuable source of genetic diversity and possess useful traits for breeding (Lopes et al., 2015). For instance, the average yield in the research stations ranged from 77 to 1,548 kg.ha⁻¹ (Assogba et al., 2015) and from 126.89 to 1444.29 kg.ha⁻¹ (Akohoue et al., 2019). Hence, there is evidence that crop landraces can potentially produce high yields. Moreover, the genetic differentiation among KG landraces was found to be moderately high, suggesting a possible low rate of outcrossing. In that case, for the development of pure lines, landraces should be purified through subsequent selfings and selections for a minimum of five generations (Acquaah, 2007; Singh et al., 2015) to reduce unwanted alleles (Ahmar et al., 2020). The evaluation of selected genotypes in different environments/agroecologies is required to fix the homogeneity, stability, and adaptability of new lines. However, pure line cultivars may not respond effectively to producers and consumers desires for many reasons: (1) pure lines have very low adaptability due to their narrow genetic base, the selection is powerless to bring changes in hereditary factors i.e., to develop new genotype (Acquaah, 2015); (2) pure lines are poor candidates for multiple trait selection because of the difficulty of finding all the desired traits in a single genotype (Priyadarshan, 2019), (3) Pureline selection requires more time, space, and expensive yield trials (Acquaah, 2015). These lines can often be used as parents in the production of other types of cultivars or breeding populations such as BILs, ILs, RILs, MAGIC, or mutants. Developing such populations in KG would create an avenue to unravel the genetic potential in the crop.

Mutation breeding is an alternate method to conventional plant breeding for increasing genetic variability and conferring specific improvement without influencing the crop phenotype expression (Kulthe and Kothekar, 2011). Whether chemical or physical, the use of mutagenic agents in the creation of genetic variability is becoming increasingly important in plant breeding (Reha-Krantz, 2013). This technique was very successful in the genetic improvement of several leguminous species such as common bean, groundnut, pigeon pea, soybean, pea (Pisum sativum L.), cowpea, mungbean, Bambara groundnut, for which the improved traits were different and consist in: disease and pest resistance, earlier or later flowering, higher yield, higher protein content, or less toxic compounds (Adu-Dapaah and Sangwan, 2004). Given the success of this technique in the legumes listed above, the application of mutation breeding may be an alternative to improve the traits of interest in KG. However, mutation induction is a random process and does not always guarantee an ideotype variety ready for commercialization; it just provides a large population of mutants each with specific characteristics (Micke, 1993). It is often difficult and rare to obtain after mutation induction, a mutant possessing all the characteristics of interest. Thankfully, the possibility to combine the chemical mutagenesis with the Targeted Induced Local Lesions in Genomes (TILLING) tools is important progress for accelerating the mutation breeding and enhancing selection accuracy of mutant desired products. In addition, hybridization can be used as a complementary method to mutation breeding insofar as it can support the transfer of the genes in a traditional way between the mutants or lines used as parents (Solanki et al., 2011).

The genetic characterization of KG revealed that there exists a moderate genetic distance between the White landrace and the other landraces which exhibited higher performances. However, landraces with White and White with black eye seed coat color are the most preferred and widely grown by farmers for consumption (Coulibaly et al., 2020). The opportunity for intraspecific hybridization can be exploited to transfer genes between KG landraces, particularly from coloured landraces to the White landrace. Moreover, the interspecific crosses for enhancing genetic variability and introgressing useful genes into KG from closely related species could be explored. Interspecific hybridization involves two different species and is widely used in breeding programs (Da Motta et al., 2020; Pratap et al., 2021). Although this technique is widely used, it remains challenging in the case of KG for several reasons. It seems that the wild form of KG is still unknown and is not found in any genebank (Ayenan and Ezin, 2016). However, Pasquet et al. (2002) had reported using two wild accessions from Cameroon in genetic diversity study. No studies on the agronomic potential and reproductive biology of any wild species have been published to date. Furthermore, possibilities of hybridization of KG with Horsegram can be considered to create early maturing hybrids resistant to drought and pest attacks (Chahota et al., 2013; Amal et al., 2020). In this case, a study of crossability barriers must be carried out to understand the factors of success or failure of a crossing between the two species (Akkerman and Bakker, 2011; Martins et al., 2019; Ferreira et al., 2021).