Review of empirical and emerging breeding methods and tools for yam (
 Dioscorea
 spp.) improvement: Status and prospects

Yam is a multi-species monocotyledonous crop extensively disseminated in Africa, Asia, Oceania and South America. The genus Dioscorea to which yam belongs encompasses about 600 species (Burkill, 1960); however, a few are cultivated for food and income. Dioscorea alata, D. cayenensis and D. rotundata are by far the major cultivated species worldwide, while D. bulbifera, D. esculenta, D. opposita, D. japonica, D. nummularia, D. pentaphylla, D. transversa, D. trifida and D. dumetorum are also economically important (Lebot, 2009; Sonibare, Asiedu, & Albach, 2010). Many wild yam species serve as sources of food in West Africa especially in times of food scarcity (Bahuchet, McKey, & Garine, 1991; Sato, 2001). Dioscorea rotundata is indigenous to West Africa and represents the most important species in terms of volume of production while D. alata, which was introduced to Africa from Asia, is the most widely cultivated species globally (Abang, Winter, Mignouna, Green, & Asiedu, 2003). Yam is an important staple source of carbohydrates (starch, sugars and fibres), proteins, minerals, vitamins and small amounts of lipids in the diets of millions of people in the tropics and subtropics. Yam is not only cultivated for consumption and as a source of income, but it is a highly esteemed food crop integrated into the social, cultural, economic and religious aspects of life of West Africans (Zannou et al., 2004). The rituals, ceremonies and superstitions Received: 27 December 2018 | Revised: 10 October 2019 | Accepted: 26 October 2019 DOI: 10.1111/pbr.12783

A number of constraints including high cost of planting materials, high labour costs, poor soil fertility, low yield potential of local varieties, pests and diseases (yam anthracnose, virus and nematodes), and shortage of quality seed yam of popular landraces and released varieties have been identified as the major constraints of yam production in Africa (Aidoo et al., 2011;Baimey, Coyne, & Labuschagne, 2006; Lopez-Montes, Bhattacharjee, & Tessema, 2012). The presence of virus in yam tubers not only causes yield loss but also hinders the international exchange of yam planting materials and germplasm (Asala et al., 2012;Kumar, 2015). The increase in yam production in Africa over the years is mainly due to the rapid increase in the area of yam fields planted into marginal lands and into non-traditional yam-growing areas (Mignouna, Abang, & Asiedu, 2008) rather than resulting from increased yield per unit land area. The current trend of area expansion with reduced fallow period will soon reach the limit. The need to provide farmers with improved varieties that combine high and stable yields with pest and disease resistance and acceptable tuber quality for processed and fresh tuber consumption markets can therefore not be over-emphasized.
A number of improved yam varieties having multiple pest and disease resistance, wide adaptability and good organoleptic attributes have been developed with empirical breeding methods in Africa and Asia by collaborative efforts of the international institutes and national yam breeding programmes. This progress with selection based on phenotypic expression of the crop for the traits of interest has however been challenging and slow due to inherent biological constraints that impede the elucidation of the genetics of important traits in yam (Mignouna et al., 2008). Although empirical breeding approaches have led to crucial achievements, contemporary tools and resources must be developed and applied to create the paradigm shift necessary to contribute significantly to feeding the rapidly increasing global population amidst climate change, dwindling resources and dynamics of consumer preferences.
The decreasing cost and rapid advancement in next-generation sequencing procedures, together with high-performance computational methodologies, have led to extensive discovery of advanced genomic resources in numerous model and non-model plants.
Significant genetic gains have been recorded in several crops such as rice, soya bean, maize and wheat, and more recently cassava through the application of DNA-driven breeding techniques such as genomic selection and marker-assisted breeding. The wealth of knowledge emanating from genomes, metabolomes, transcriptomes, phenomes and gene expression profiles will immensely contribute to our understanding of underlying gene regulatory networks to enable systematic improvement of crop breeding. The deployment of these contemporary methods will therefore expedite yam breeding efforts. However, much information is needed before full deployment of molecular breeding in yam as most of the information currently available for the crop is on diversity studies and few QTL detections with less emphasis on the validation and development of informative markers for direct use in breeding.
Earlier efforts in genomics research in yam were mainly focused on the development of polymorphic molecular markers and evaluation of their potential applications for investigating genetic diversity, demarcation and identification of relationships among the various species. Quantitative trait loci (QTLs) controlling the resistance of yam to anthracnose and yam mosaic virus disease have also been identified Mignouna et al., 2002b;Petro, Onyeka, Etienne, & Rubens, 2011;Saski, Bhattacharjee, Scheffler, & Asiedu, 2015). Significant efforts have been made recently to develop more genetic tools and genomic resources to transform yam breeding. This paper reviews the advancements achieved in empirical yam breeding endeavours and the development, status and application of emerging breeding tools and methodologies for the genetic improvement of yam.

| G ERMPL A S M RE SOURCE S FOR YAM IMPROVEMENT
Availability and access to smart germplasm resource is vital for the development of improved crop varieties that possess superior agronomic and food quality traits. Yam germplasm collections conserved in the field and in vitro genebanks constitute a huge gene pool for yam improvement activities in order to achieve its optimum poten- in Vanuatu, South Pacific islands (Papua New Guinea, Fiji, and New Caledonia islands), and Ethiopian Biodiversity Institute and National Research System in Ethiopia maintain a substantial number of Dioscorea species under ex situ collections (Arnau et al., 2017;Chi, Hue, & Trinh, 2007;Lebot, 2009;Mekbib & Deressa, 2015). IITA has increased tremendously its yam The yam germplasm collections are rich in rare alleles for target traits from which breeders source genetic materials to expedite genetic gain in a sustainable way. However, the systematic use of genebank accessions has not progressed very far in yam breeding programmes. The development of core collections representing the diversity of the entire germplasm (Frankel & Brown, 1984) facilitates easy access and enhances the utilization of the main collection for fast evaluation in crop improvement programmes. The first core collection (391 accessions) from the IITA yam germplasm represented 75% of the morphological diversity of the entire collection using data on 77 morphological descriptors and country of origin (Mahalakshmi et al., 2007). This core collection represented 13% of the entire collection. With the inclusion of new germplasm and availability of information, the yam core collection at IITA was revisited and updated using 56 morphological traits (Girma et al., 2018). The updated core collection comprised 843 accessions and represented about 20% of the whole collection. Delineation of core collections and diversity studies in general has often been based on morphological traits rather than molecular markers. The high phenotypic plasticity of yam is a limitation to morphological characterization, hence, the need to enhance "omic" resources.
Phenotypic characterization and assessment of germplasm for morphological and agronomic traits, tuber quality, ploidy status, and flowering ability have resulted in the identification of accessions as parents with attributes pertinent to the objectives of the breeding programme (Lopez-Montes et al., 2012). To explore the use of aerial tubers of D. alata as alternative planting materials owing to the high cost and supply shortage of seed yam propagules, over 800 accessions from the IITA germplasm collection were evaluated for aerial tuber production and a set of accessions bearing aerial tubers were identified (Girma, Gedil, & Spillane, 2015). A number of studies have successfully assessed the IITA's and other institution's yam diversity represented by different genebank accessions, landraces and breeding lines for host-plant resistance and quality traits and have shown a rich base of germplasm resource that can inform breeding strategies for resistance to major yam pest such as nematodes, anthracnose and virus diseases, and genetic enhancement for quality traits including various secondary metabolites, tuber carotenoids and other food quality traits (Mohandas, Ramakrishnan, & Sheela, 1996;Plowright & Kwoseh, 2000;Mignouna, Abang, Green, & Asiedu, 2001;Mignouna, Njukeng, Abang, & Asiedu, 2001;Abang et al., 2003;Onyeka, Petro, Ano, Etienne, & Rubens, 2006;Arnau, Maledon, & Nemorin, 2007;Egesi, Odu, Ogunyemi, Asiedu, & Hughes, 2007;Kwosch, Plowright, Bridge, & Asiedu, 2007;Kwon et al., 2015;Price, Wilkin, Sarasan, & Fraser, 2016;Lebot et al., 2018;Price, Bhattacharjee, Lopez-Montes, & Fraser, 2018). There is also an ongoing initiative at IITA to sequence all the genebank yam collections to expedite its use for breeding. This effort shall be intensified by systematically arranging the trait-specific core sets to identify and utilize useful genetic variability that exists in wild and cultivated genepools for long-term use in yam breeding. Extensive phenotyping of large set of genetically diverse yam germplasm for different traits and developing trait-specific mini-core sets such as biotic stress resistance, quality and abiotic stress tolerance could provide an opportunity to mine novel alleles from underutilized germplasm resources for breeding next generation of yam varieties.

| Yam breeding targets and challenges
The breeding programmes in many countries focus on a few Dioscorea species, most specifically the dominantly cultivated yams such as D. rotundata and D. alata. The primary focus of yam breeding is the development and deployment of robust varieties with unique combinations of preferred traits required for production and consumption. The programmes generally target traits related to increasing intrinsic tuber yield potential as well as increasing tolerance for, and resistance to, yield-limiting and quality-reducing factors. Still, the specific breeding targets varied according to the region and species involved (Mignouna, Abang, & Asiedu, 2007). The breeding targets have, however, evolved over the years to meet the changing needs and preferences of farmers and other end-users.
From the inception of formal yam breeding since the 1970s, the breeding targets have been gradually adding new traits along with the primary focus on high and stable tuber yield, higher dry matter, resistance to economically significant diseases (e.g. anthracnose, viruses, tuber rots) and pests (e.g. nematodes), tuber characteristics cherished by consumers (e.g. size, shape, and culinary quality), and plant architecture (e.g. dwarf genes) that reduces the need for staking (Asiedu, Ng, Bai, Ekanayake, & Wanyera, 1998;Mignouna et al., 2007). Additionally, traits such as colour of tuber flesh, tuber oxidation, starch properties and sugar contents are now routinely measured in breeding programmes as they influence the acceptability of newly developed yam varieties (Arnau, Maledon, Nudol, & Marie-Claire, 2016;De Koeyer et al., 2017). However, high-level resistance genes to viruses, anthracnose and nematodes; shrub-like or dwarf plant architecture with stiff or stout vine base and early branching; early maturity or tuber bulking; tubers less susceptible to deformation in the soil; tolerance for low soil fertility, drought and heat; long shelf life of fresh tubers or processed food products; and high levels of culinary attributes suited to consumer needs for fresh and processed yam are among the missing traits in current advanced breeding lines and released cultivars.
The breeding efforts over the last five decades have resulted in the identification of trait progenitors and commercial release of improved cultivars. However, the local farmers' varieties are still the leading and dominant cultivars in the yam cultivation and consumption systems (Alene, Abdoulaye, Rusike, Manyong, & Walker, 2015).
While this low market penetration of improved varieties could be attributed in part to the limited dissemination efforts, limited exposure of the customers to the released varieties or lack of preferred attributes in the varieties meeting the needs and demands of farmers and market, the setting of breeding goals or specifications for new varieties was refocused and restructured to raise the market penetration potential of future releases with correct product profile.
The breeding targets have been transformed from undifferentiated product portfolio to a differentiated product concept where customer needs are profiled and translated to product specifications.
Clients/customer needs are now the main focus of the variety development plan of yam breeding programmes in Africa. Accordingly, three key breeding product profiles have been carefully framed for targeting (Table S1) by IITA and its national partner breeding programmes in West Africa. The product concept where the right features or traits for the varieties that the clients (growers, processors, retailers and consumers) require or demand is identified and prioritized with a clear roadmap to achieve the target product in a specified timeframe: five years to replace the current market-leading varieties with improved features to drive adoption in short term (current product profile) and ten years to develop parental lines feeding the variety development pipeline in medium term (future product profile). The yam breeding programme, therefore, envisions implementing a product concept that can drive rapid and successful uptake of new varieties in the production and consumption system.
Making breeding progress in yam has been challenging. Many factors such as unpredictable or no flowering, non-synchronous flowering of elite genotypes, lengthy growth cycle, low multiplication ratio, polyploidy and high heterozygosity negatively influence the genetic improvement of yam . Additionally, the dioecious nature of yam, cross-incompatibility of inter-and intra-species hybridization, and the enormous phenotypic plasticity make the evaluation and stacking of several interdependent characters a problematic process (Price, Bhattacharjee, Lopez-Montes, & Fraser, 2017). Limited understanding of the genetics of main and component traits contributing to enhanced productivity under current production challenges and limited breeding enabling technologies that can contribute to its accelerated improvement have also thwarted yam breeding efforts. The influence of biochemical composition on organoleptic properties of yam is not well understood and, therefore a significant hindrance to detecting genetic/biochemical trait markers effectively translating the genetic diversity to enduser preferred genetic gain (Price et al., 2017).
Furthermore, the existing yam germplasm has not been well characterized, limiting the efficient utilization of available diversity in genetic improvement for the agronomic and food quality traits.
Transfer of genes of interest from the secondary gene pool of wild relatives to the cultivated primary gene pool is still challenging in several crops, including yam (Lebot, Abraham, Kaoh, Rogers, & Molisalé, 2019;Spillane & Gepts, 2001). Nevertheless, the wild relatives of yam could harbour essential genes and genetic variability useful in breeding endeavours to enhance the performance of yam for various economically important traits. Yam germplasm conservation has predominantly focused on the main cultivated species with little emphasis on wild relatives; hence, future studies can be limited if conservation efforts are not increased. Basal species with diverse traits are also lacking (Price et al., 2016), although they could be especially useful for genome editing. Several wild relatives of yam recently appeared on the International Union for Conservation of Nature (IUCN) red data list and have been declared threatened and may further compound the problem of the availability of genetic diversity for utilization in the breeding of this valuable staple crop.

| Conventional hybridization and selection
Traditionally, yam breeding is a two-step process that utilizes both sexual and asexual reproduction. Different breeding schemes are planting as in D. alata, which has no seed dormancy (Abraham, 1992).
The botanical seeds are sown in nursery beds or seedling trays filled with suitable growing media such as carbonized rice husk and coco peat. Seed germination starts in ten days after sowing and continues for one month. The seedlings are then transplanted to the pots under screenhouse or nursery beds in the field for single plant selections at seedling and tuber progeny nurseries and the subsequent annual advancement of the clonal derivatives through early clonal generation evaluation nursery, preliminary performance trial, advanced multi-location and multi-season performance trial to on-farm variety validation trial for an official release and commercial deployment  Table 1). Crossing and selection have been used as a prime tool to develop improved varieties in yams. Table 2 (Table 2). Lebot, Abraham, et al. (2019) reported variation in pollination efficiency, seed set and seedling survival in crosses involving diploid × diploid, diploid × tetraploid, and the same pollen parent in different cross-combinations.
Sparse to non-flowering, flower asynchrony, pollen viability, the receptiveness of stigma, pollen-stigma compatibility, pollinators' efficiency in loading sufficient pollen on the stigma, embryo abortion, low seed germination, low seedling survival rate, and differential cross-fertility of parents are severely impacting the success rates of yam cross-pollination efforts. Improved techniques with flowering induction and pollen storage would increase the success rates in crosses. Moreover, ploidy determination, screening accessions for specific or general cross-compatibility for fruit and seed set, seed germination and seedling survival rate and identification of compatibility groups, development of functional molecular markers for flower sex, flowering intensity and cross-compatibility prediction, and confirmation of hybridity in progeny will contribute to overcome the low success rate and improve the efficiency of hybridization in yam breeding.
The choice of the right parents for use in the crossing scheme is a stepping stone towards success in yam breeding effort and entails the use of per se phenotypic performance for the trait of interest, estimation of breeding values, and dissection of genetic relatedness using pedigree and molecular marker information.
Per se performance has been widely utilized but not always effective in producing heterotic progenies. Joining per se performance along with their combining ability is an important tool for the selection of parents for hybridization. The use of such classical breeding technique alone is slow to achieve the anticipated breeding gain. Interspecific hybridization has been less exploited due to cross-incompatibility and non-synchronization of flowering. Successful interspecific hybridization was reported between D. japonica and D. opposita under natural open pollination (Araki, Harada, & Yakuwa, 1983) and between D. alata and D. nummularia under artificial hand pollination (Lebot, Abraham, et al., 2019).
Dioscorea rotundata has been successfully crossed with D. cayenensis at IITA but hybridizing either of the two to D. alata has not been successful (Lopez-Montes et al., 2012). The unsuccessful interspecific hybridization between D. alata and other yam species was also reported by Rao, Bammi, and Randhawa (1973). The  However, this effort has not progressed well due to the natural cross barrier among the different species. The failure/discontinuation of interspecific hybridization of D. rotundata and D. cayenensis was mainly because D. cayenensis was wrongly believed to have a higher level of carotenoids (Price et al., 2018).
Phenotypic selection of the variants within natural barriers is a core yam breeding effort by different programmes. Three to four clonal selection and testing cycles (from tuber family evaluation to advanced performance trial) are often attempted to identify superior recombinants for further assessment under a wider range of environments and user preferences (Table 1). At seedling and first clonal generation, selections are based mainly on reaction to a natural infestation of anthracnose in D. alata and viruses in D. rotundata, along with tuber appearance. At subsequent clonal derivate stages, from the second clonal generation onwards, selection is made, among other things, for tuber shape and appearance (uniform, globular to cylindrical or round shape with smooth skin), tuber dry matter (≥25%), tuber enzymatic oxidation (non-oxidizing white or cream flesh after peeling or cutting the tubers for at least 180 minutes), tuber yield and stability (≥5% yield advantage over best check), yam mosaic virus and yam anthracnose tolerance (≤2.5 severity score under field natural infestation) and plant architecture or type for adaptation to no or minimal staking. At advanced clonal stages (advanced performance or multi-environment trials), the genotypes are subjected to series of food quality analysis, most specifically boiled and pounded yam quality test in participatory trial setup engaging end-users. The most promising two to three clones from multi-environment testing then enter the final wide-scale onfarm validation or verification assessment for commercial deployment.
The current breeding scheme for yam is lengthy and slow.
Setting testing and selection trials with the needed replications at scale is a challenge due to the difficulty to obtain sufficient quantity of uniform clonal planting materials. Propagules for planting from different position of the same mother tuber (proximal, central and distal parts) often produce heterogeneous sprouting and cause a high interplant variability within a clone in field trials (Cornet, Sierra, Tournebize, & Ney, 2014). With technological advancements, present-day yam breeders should increasingly be able to select superior recombinants accurately in breeding populations with reduced breeding cycles, manipulate game-changing plant attributes beyond the natural barriers and also produce sufficient quality planting materials to expedite the genetic gain.

| Pattern of inheritance for key traits
Nature of trait heritability is essential for the design of an optimal breeding strategy. Traits can be classified as either qualitative or quantitative based on the pattern of their variation in a population. Variation is discrete for qualitative traits which are often controlled by one or a few genes whereas it is continuous for the quantitative traits which are mostly controlled by several genes with each having a minor effect. Few studies have dissected the nature of trait heritability in yam crop. Single dominant gene in a simple or major recessive gene in duplex condition controls the inheritance for yam mosaic virus (YMV) resistance in D. rotundata (Mignouna, Njukeng, et al., 2001). Genetic inheritance for reaction to anthracnose disease in D. alata, caused by Colletotrichum gloeosporioides, was reported as dominant gene action with quantitative inheritance . Sartie and Asiedu (2014)  for genetic parameters reported was much lower in seedling-derived plants compared to the clonal-derived plants (Akoroda, 1983) making reproducibility for the phenotypic selection at early generation stage a challenging task in yam breeding. Genes controlling essential traits such as tuber yield, tuber quality, resistance to pests, and diseases are quantitatively inherited and less likely linked, making their improvement through empirical breeding a challenging task. The efficiency and effectiveness of breeding for essential traits in yam would significantly improve with the use of genomic and genetic assisted breeding.

| Envirotyping
Accurate examination of complex environmental factors for both target environments and specific genotypes provides a unique avenue for management and optimization of environmental variables for enhanced genetic improvement and efficient crop production in an era of global climate change. All environmental factors that influence plant growth and yield are defined as envirotypes, and the process for determination and measurement of all the environmental factors is envirotyping (Xu, 2015(Xu, , 2016. The concept "envirotyping" as suggested by Xu (2015) is another "typing" technique supporting phenotyping and genotyping to elucidate environmental effects on crops. Through its efficient mechanisms such as integrative phenotyping, genotype-by-environment interaction, genes responsive to environmental signals and biotic and abiotic stresses, envirotyping facilitates crop modelling and phenotype prediction (Xu, 2016).
The partitioning of yam production environments into similar mega-environments will have functional implications for variety testing and performance prediction for final product recommendation thereby enabling a broader impact of breeding products. Alabi et al.

| Enabling seed technologies to expedite genetic gain
The low multiplication ratio of yam has significant effects on yam production, breeding and dissemination of released varieties to farmers. The regular replacement of stocks of seed yams infested by pests and diseases is usually not possible due to short supply of quality seed tubers at affordable prices thereby forcing farmers to recycle poor quality seed yams with the risk of low yields (Aighewi, Asiedu, Maroya, & Balogun, 2015). Farmers produce quality seed yams through traditional approaches such as selection of small whole tubers from crop harvest; stimulating the development of seed tubers by "milking" ware tubers while the leaves of the plant are still green (double harvest system); cutting ware tubers into setts about the same sizes as normal seed tubers; and the "Anambra" system where seed tubers are produced from smaller setts (Aighewi et al., 2015). However, low multiplication ratio, high cost and high risk of exposure of seed tubers to contamination with pests and pathogens often characterize the traditional systems of seed yam production (Nweke, Ugwu, Asadu, & Ay, 1991

| Reference genome sequence
The These data also facilitate the detection of allelic variations and candidate genes modulating key agronomic and quality traits, which is essential for the success of yam breeding.

| Molecular markers and genotyping systems
Molecular markers are relevant tools for applications such as assessing genetic diversity and phylogenetic relationships, cultivar identification, mapping of major effect genes and QTLs, estimating population structure, identification of elite genotypes in crop improvement programmes, and for validation of progenies emanating from genetic hybridizations (Tamiru et al., 2015). Many marker systems have been developed and routinely applied to yam improvement activities ( Table 3). Some of these marker systems applied in yam are listed below.

| Isozyme markers
Isozymes were the first molecular markers to be established which showed Mendelian inheritance, co-dominant expression, complete penetrance and non-existence of pleiotropic and epistatic interactions (Weeden & Wendel, 1989). The potential of isozyme markers for molecular characterization within and between various Dioscorea species has been well established (Table 3).

| Restriction Fragment Length Polymorphism (RFLP)
Terauchi, Chikaleke, Thottappilly, and Hahn (1992) exploited heterologous DNA sequences as a source of RFLP markers and developed the first RFLP markers for yam. Restriction fragment length polymorphism markers were successfully used to study the phylogeny and origin of Guinea yam (Terauchi et al., 1992) and genetic diversity of D. bulbifera accessions originating from Asia and Africa (Terauchi, Terauchi, & Tsunewaki, 1991).

| Simple Sequence Repeats markers (SSRs)
Microsatellites or SSR, due to their co-dominant nature, high level of polymorphism, high abundance and even distribution across the genome were necessary as yam genomics progressed. Terauchi and Konuma (1994) (Table 3).

| Next-generation-based genotyping procedures
The field of genomics has been markedly transformed in recent years due to the advances in high-throughput sequencing technologies, making it feasible to generate large volumes of sequence data quickly and at a considerably lower cost (Elshire et al., 2011). The high-marker density methodologies presented by next-generation-

| Transcriptome sequencing
Analysis of genome-wide differential RNA expression gives researchers a better understanding of biological pathways and molecular mechanisms that control important but complex traits in plants. Narina et al. (2011) (Narina et al., 2011). A total of 104 candidate SNPs were also discovered between TDa95/0310 and TDa95/0328 libraries that were homologous within each genotype and useful for genotyping and developing genetic maps in yam (Narina et al., 2011).
A setback of this study is that some of the SNP markers identified are likely to be associated with broad stress or infection response and not necessarily anthracnose infection alone.  (Wu et al., 2015). With these unigenes, 11,793 SSRs were detected and 6,082 SSR markers were developed.

| Metabolomics
The efficacy of trait characterization is enormously enhanced when genomics is combined with other "-omics" approaches such as transcriptomics, proteomics, metabolomics and phenomics (Fukushima, Kusano, Redestig, Arita, & Saito, 2009). This is very true of metabolomics where, in many instances, desirable characters can be directly linked with metabolite composition (Bino et al., 2004). Metabolomic techniques produce extensive biochemical phenotypes that can be indicative of quality traits.
Association of specific plant traits to metabolite compositions provides quantifiable markers similar to genetic quantitative trait loci, for example metabolite quantitative trait loci (mQTL) which can be recorded through breeding programmes both independent of underlying genetic mechanisms and coupled with typical genomic analysis to improve association of genotype with phenotype, such as metabolite-based genome-wide association analysis (mGWAS) for functional genomics (Adamski & Suhre, 2013;Luo, 2015). An intraspecific genetic linkage map of D. alata was constructed using 523 polymorphic AFLP markers and nine putative QTL(s) identified for anthracnose disease resistance on five different linkage groups (Petro et al., 2011). The phenotypic variance explained by each QTL ranged from 7.0% to 32.9% while all significant QTLs accounted for 26.4 to 73.7% of total phenotypic variance depending on the isolate and the variable considered (Petro et al., 2011). A genetic linkage map of D. alata was developed from 380 EST-SSRs on 20 linkage groups for the identification of QTLs controlling anthracnose disease resistance . Linkage analysis conducted independently on data collected for three years and average scored data consistently found one QTL on linkage group 14.
This QTL observed at a position interval of 71.1-84.8 cM explained 68.5% of the total phenotypic variation in the average score data.
The genetics of sex determination was also elucidated by performing whole-genome resequencing of bulked segregants in an  The "Leaf Doctor" and "ESTIMATE" applications are quantitative tools for precise evaluation of percentage symptomatic tissue area of a diseased leaf based on image analysis (Pethybridge & Nelson, 2015). The applications help to differentiate healthy from diseased leaf tissues based on artificial intelligence and machine learning. A phenotyping system based on DLA, "Leaf Doctor" and "ESTIMATE" software enhances the throughput and precision and expedites the selection of promising lines for further evaluation (Kolade et al., 2018). An essential feature of the tools is the elimination of bias that arises during conventional visual scoring. The optimization of the detached leaf assay and the software applications for yam anthracnose disease evaluation has resulted in the harmonization of the phenotyping process across different countries in addition to the comparison of results and performance of genotypes. The use of these high-throughput phenotyping methods optimized for yam and others such as NIRS and ground-penetrating radar (for non-destructive estimation of root bulking rate) is expected to provide more data that could contribute to efficient product profiling and value addition.

| Genetic engineering and gene editing for yam improvement
Genetic engineering has emerged as an important alternative and complementary methodology to improve crops including yam. The application of transgenic methods to yam improvement is particularly compelling due to the difficulties associated with conventional yam breeding. However, an efficient plant regeneration system is the main prerequisite for the achievement of successful transformation (Nyaboga, Tripathi, Manoharan, & Tripathi, 2014). Due to its ease of accessibility, ability to transfer low copies of DNA fragments carrying the desirable genes at higher efficiencies with minimal cost as well as the transfer of very large DNA fragments with low rearrangement, Agrobacterium-mediated gene delivery system is the most preferred (Gelvin, 2003;Shibata & Liu, 2000). Quain, Egnin, Bey, Thompson, and Bonsi (2011)  Targeted genome alteration approach has become a promising tool for crop breeding. The recently established gene-editing technique, the clustered regularly interspaced short palindromic repeats (CRISPR/Cas9) system, resulting from the adaptive immune system of Streptococcus pyogenes, has been demonstrated to be a potent tool for targeted genome editing in many species (Feng et al., 2018 (Tripathi et al., 2019) clearly indicates that such an approach can be implemented to develop yam varieties resistant to yam mosaic virus, in that, as in the case of banana streak virus, viruses affecting yam have also been found to be integrated into the genome of yam Umber et al., 2014). Genome editing should therefore be incorporated into the yam improvement programme and traits to be targeted should be decided in consultation with breeders.
The ethics and regulation of genetically modified and gene-edited crops should be taken into serious consideration in the application of these technologies. The major challenge of the CRISPR/ Cas9 technology is that it may recognize sequences with up to five mismatched bases suggesting high rates of off-target effects (Roy et al., 2018). Some approaches such as DNA-RNA chimeric guides, Cpf1, a single RNA endonuclease that employs a T-rich PAM on the 5′ side of the guide, and specific point mutations have been developed to mitigate this limitation (Kleinstiver et al., 2016;Zetsche et al., 2016).

| Marker-assisted selection (MAS)
To our knowledge, there is no released yam variety whose devel-  Tamiru et al., 2017), resistance to yam anthracnose disease Mignouna et al., 2002b;Petro et al., 2011) and yam mosaic virus disease (Mignouna, Abang, Onasanya, et al., 2002;Mignouna et al., 2002a) have been identified. Marker discovery for other essential traits such as tuber flesh oxidation, starch property and dry matter content is ongoing at IITA and other research institutions in collaboration with several international partners and national agricultural research programmes across SSA and beyond (https ://afric ayam.org). The next step is the conversion of these QTLs to diagnostic SNP markers. These markers will then go through verification and subsequent deployment in breeding programmes.
The QTLs identified for plant sex determination in D. rotundata by Tamiru et al. (2017) have successfully gone through this process, and SNP markers are available for use. These SNP markers have been successfully deployed in ongoing work at IITA to predict or determine the sex of early generation genotypes in the yam breeding programme. The application of these novel methods will enhance yam breeding efforts and ensure the quick delivery of high yielding, nutrient-dense and climate-resilient varieties to farmers.

| Parentage reconstruction
For many root and tuber crops such as yam, planned crosses are quite challenging due to inherent biological bottlenecks such as cross-incompatibility, non-synchronous and erratic flowering and polyploidy. A viable alternative for genetic improvement in outcrossing species that are constrained in executing planned and controlled crosses is half-sib breeding which entails the random pollination among desirable parents (Norman et al., 2018). The progeny emanating from half-sib breeding must therefore be authenticated in order to ascertain the selection gain in the breeding programme.
Parentage analysis anchored by DNA markers is a reliable method for breeders to improve genetic gain in a breeding programme. This entails DNA profiling of progeny and possible parents and comparing their alleles for determination and validation of existing relationships (Jones, Small, Paczolt, & Ratterman, 2010). It can help to elucidate the identity of half-sib progenies and reconstruct the pedigree in the outcrossing crops (Telfer et al., 2015). Potential application and usefulness of parentage analysis in crop improvement programmes include reliable estimation of paternal breeding values in the half-sib family, reduces pollination and labelling errors, assessment of the level of inbreeding and incompatibility, and estimation of genetic effects including general and specific combining abilities in the breeding programme. Zoundjihekpon, Hamon, Tio-Touré, and Hamon (1994)

| Polyploidy breeding
Yam is represented by multiple species with varying intraspecific and interspecific ploidy levels ranging from diploid to octoploid.
The different ploidy levels in yam constitute a challenge but also an opportunity when it comes to its genetic improvement through breeding. It is a challenge as it hinders the effective transfer of relevant genes among individuals with different ploidy levels due to cross-incompatibility and erratic flowering. Ploidy manipulation through induced polyploid or conventional hybridization, however, in many crops has been reported as a potent breeding tool for improving yield potential and also resistance to biotic and tolerance for abiotic stresses. Association of ploidy variation with yield potential and other plant traits has also been reported in yams. Higher vigour and tuber yield advantage were reported with tetraploid (2n = 80) and triploid (2n = 60) water yam compared to its diploid (2n = 40) counterpart (Arnau et al., 2007;Lebot, 2009).
Kenji, Ohara, and Iwasa (2005) reported a shorter and thicker vine in D. japonica with artificially induced tetraploid. Likewise, Huang, Gao, Chen, and Jiao (2008) reported an increase in the diosgenin content in D. zingiberesis with induced polyploid. However, a preliminary observation on artificially induced auto-tetraploid in D. rotundata accessions at IITA has not shown a clear advantage over their diploid counterparts for tuber yield traits.
The discovery of tetraploid (2n = 80) with a high degree of sexual fertility in D. alata has opened a new perspective in polyploidy breeding using the conventional hybridization (Arnau et al., 2010).
The tetraploid clones were found to be crossable among themselves, and the inter-ploidy crosses between diploid females and tetraploid males were reported successful. By this new avenue of breeding in D. alata, superior polyploid hybrid selections combining several desirable traits, including anthracnose resistance, were identified in Guadeloupe, French West Indies (Arnau et al., 2007).
In Vanuatu (South Pacific), anthracnose-resistant D. alata hybrids were produced for the first time utilizing the tetraploid fertility in addition to combining other desirable traits (Lebot, Abraham, et al., 2019).
Ploidy breeding might hold potential for increasing yield, improving product quality and increasing resistance to biotic and tolerance for abiotic stresses in yam but its application requires accurate knowledge of the level of ploidy of the genotype (Gamiette, Bakry, & Ano, 1999). Heritability and segregation stud-

| Breeding data management
The rapid advancement and application of high-throughput genotyping and phenotyping technologies, coupled with the expansion of breeding activities, has resulted in the generation of large volumes of data. To make these data easily available and accessible in real time to the yam breeding research community scattered across Africa and beyond, a database system called "yambase" has been developed to store and retrieve phenotypic and genotypic data (https ://yamba se.org/) under the AfricaYam project.
Several relevant statistical methods and bioinformatics tools for data quality control, pedigree information visualization, experi-

| CON CLUS I ON AND PROS PEC TS
Though several inherent biological constraints make yam breeding a very arduous and lengthy endeavour, significant milestones have with the integration of clonal propagation innovations such as the SAH system which is extremely valuable to rapidly supply the sufficient quantity of quality planting materials required for accurate phenotyping in genetic study and breeding trials. The benefits of these new breeding tools cannot be over-emphasized; however, much preparation and the initial investments are required for full deployment. Molecular laboratories at the national breeding institutions will need upgrading, and the capacity of staff strengthened to handle the molecular breeding activities better. The IITA and other international partner institutes continuously play crucial roles in re-tooling and capacity development of the national agricultural research systems; more investments are, however, required in this area. Recent advancements and application of other "omics" technologies such as metabolomics, transcriptomics and phenomics will provide a new understanding of trait expression in yam. The synergistic effect of integrating these "omics" approaches into current genetics-led yam breeding programmes would undoubtedly remove some of the constraints encountered for typical breeding approaches and expedite the delivery of improved varieties to farmers. of Ibadan, Nigeria. We also thank anonymous reviewers for their insightful comments and suggestions that substantially improved the manuscript.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R ' S CO NTR I B UTI O N
AA and KD planned the content, wrote the draft and revision. BO and RA contributed in write up and revision. AA, KD and RA did formatting.