Keywords

1 Introduction

Yam is a collective name of tuber crops belonging to the genus Dioscorea. In 2018, the global yam production was around 72.6 million tons (FAOSTAT 2018). The major yam species include Dioscorea rotundata, D. alata, D. trifida, D. polystachya, and D. esculenta (Arnau et al. 2010). White Guinea yam (D. rotundata) is the most important yam worldwide, mainly grown in West and Central Africa, especially Côte d’Ivoire, Ghana, Togo, Benin, Nigeria and Cameroon, the region known as the “yam belt”, which accounts for ∼92.5% of the total world yam production (FAOSTAT 2018). Yam is a staple crop in many tropical countries, and it also plays important roles in society and culture of the people in the major yam-growing regions (Coursey 1972; Obidiegwu and Akpabio 2017; Obidiegwu et al. 2020). However, due to its localized importance, yam has been regarded as an “orphan crop” and received considerably less research attention compared to the major crop species.

The genetic improvement of yam is urgently needed for the food security of yam-growing regions, but it is constrained by various abiotic and biotic factors (Mignouna et al. 2003). For example, the entire genus Dioscorea is characterized by dioecy, with male and female flowers borne on separate individuals, which imposes obligate outcrossing to the species in the genus. Due to its dioecy, farmers clonally propagate yams to maintain its germplasms, and true seeds are rarely used as the starting materials for planting. However, this clonal propagation reduces the genetic diversity, which causes the vulnerability to plant diseases. Also, the clonal propagation causes the difficulty of purging deleterious mutations from the germplasms like in cassava (Ramu et al. 2017). To achieve effective yam improvement by overcoming these constraints, we need to answer key questions in yam genetics and genomics including: (1) what is the genetic relationships between cultivated yams and their wild relatives, and how the domestication of yam happened? (2) how to deploy Dioscorea genetic diversity to improve agronomic traits of cultivated yams? (3) how dioecy of Dioscorea is genetically controlled and how we can manipulate it to make an efficient cross breeding? Thanks to the recent development of genome sequencing technologies, we can now address these questions using population genomics approaches. In this chapter, we review the latest findings of the domestication of yam from population genomics perspectives.

2 The Genus Dioscorea: Its Origin and Botanical Characteristics

The genus Dioscorea, which consists of approximately 630 species, is the largest one in the family Dioscoreaceae of monocotyledons (WCSP 2020). It is widely distributed in the tropical and temperate regions and occurs in diverse environments from forests to grasslands (Wilkin et al. 2005; Maurin et al. 2016; Viruel et al. 2016). Several studies have been conducted on the phylogenetic relationships of species in Dioscorea. Previously, intrageneric taxa have been proposed based on morphological characters (Uline 1898; Knuth 1924; Prain and Burkill 1936, 1939; Burkill 1960; Huber 1998). However, diagnostic keys and delineation of taxa varied according to the authors. Recently, phylogenetic analyses have been conducted based on chloroplast DNA (cpDNA) sequences and nuclear gene sequences (for review, see Noda et al. 2020). Noda et al. (2020) provided a large-scale phylogenetic tree containing 183 species and proposed dividing Dioscorea into two subgenera (Dioscorea and Helmia), with 11 major clades and 27 sections/species groups.

Dioscorea likely originated in the Laurasian Palaearctic between the Late Cretaceous and the Early Eocene (Fig. 1). In the Eocene and Oligocene, Dioscorea expanded to the southern region by long-distance dispersal or migration by land bridges. In the Oligocene and Miocene, main Dioscorea lineages experienced divergence events on a worldwide scale. In the Miocene and Pliocene, some lineages dispersed into new areas. The number of biogeographical speciation events seems to have decreased after the Quaternary period began (Maurin et al. 2016; Viruel et al. 2016; Couto et al. 2018).

Fig. 1
figure 1

Biogeographical origin and distribution of Dioscorea species (Viruel et al. 2016). (a) Dioscorea likely originated in the Laurasian Palaearctic in the Late Cretaceous and the Early Eocene (1) and then dispersed from Asia to South America (2). (b) In the Oligocene and Miocene, Dioscorea mainly expanded to the southern region. (c) Some lineages dispersed into new areas in the Miocene and Pliocene, but speciation events decreased in the Quaternary. (d) Geographical distribution in the present era. (Maps are based on C. R. Scotese’s PALEOMAP project; www.scotese.com)

The majority of Dioscorea species are perennial herbaceous climbers with simple or compound leaves and reproduce sexually and/or clonally (Fig. 2). Flowers in Dioscorea are mostly dioecious with male and female flowers borne on separate individuals, and multiple sex-determination systems were reported in the genus (see Table 2 and Sect. 6). Most species produce winged seeds and capsular, six-seeded fruits, while some species have wingless seeds, samaroid or berry fruits (Caddick et al. 2002; Noda et al. 2020). In addition to sexual reproduction, Dioscorea species propagate clonally by bulbils, rhizomes, or tubers. Bulbils are aerial tubers that are formed in the axils of leaves or bracts of some Dioscorea species (Fig. 2f). They are mainly consumed as food, but also used as folk medicine in many cultures (Ikiriza et al. 2019). Bulbils are generally brown-colored and have small tubercles over their surface, but their shape and size vary in the different species (Murty and Purnima 1983). D. bulbifera (also known as aerial yam) is the major bulbil-producing species and is characterized by considerable bulbil shape diversity (Terauchi et al. 1991). Rhizomes and tubers represent morphologically diverse structures that serve as underground starch storage organs (Fig. 3). Because these storage organs serve as food sources for various wild animals, they have evolved defense traits. For example, D. praehensilis has crown roots with spines to protect tubers from burrowing or digging animals (Fig. 3c). Some species of the African clade have thick corky barks covering the pachycaul structure that may provide protection against fire and herbivores (Scott 1897; Maurin et al. 2016). In addition, Dioscorea species produce diverse secondary metabolites such as saponins, alkaloids, and tannins that serve a variety of functions including defense against herbivores (Coursey 1967). Chemical components of some species have medicinal values (Liu et al. 2008; Dutta 2015).

Fig. 2
figure 2

Morphological diversity of the above-ground parts of Dioscorea species. (a) D. tokoro, (b) D. quinqueloba, (c) D. rotundata, (d) a stem of D. mangenotiana with thorns, (e) flowers of D. japonica, (f) a bulbil of D. bulbifera

Fig. 3
figure 3

Rhizomes and tubers of Dioscorea species. (a) rhizomes of D. tokoro, (b) Tukuneimo group (left top), Ichoimo group (left bottom), Nagaimo group (right) in D. polystachya, (c) D. praehensilis, (d) D. minutiflora, (e) D. rotundata (left), D. cayenensis (right), (f) D. mangenotiana, (g) D. abyssinica

3 Ploidy in Dioscorea

Basic chromosome number of Dioscorea is × = 10 in the section Stenophora. However, in the section Enantiophyllum that includes the major cultivated species, the basic chromosome number is × = 20 (Scarcelli et al. 2005; Arnau et al. 2009). Our survey of the available literature shows that at least 33% of the species in the genus Dioscorea have variable ploidy levels including within the same species (Table 1). As we discuss in the later sections, Dioscorea species have undergone several hybridization events that sometimes led to changes in the ploidy levels. Information on ploidy is important to apply the correct population genomics approaches for the study of Dioscorea. For example, if the sequence reads obtained from a polyploid individual are aligned to a consensus haploid reference genome, variant calling based on the assumption of diploid genome may lead to wrong genotyping results.

Table 1 The various ploidy levels of Dioscorea species

4 Reference Genome Sequences for Dioscorea

A reliable reference genome sequence is indispensable for genome diversity studies of a species. The first chromosome-level reference genome obtained for Dioscorea species was that of D. rotundata (Tamiru et al. 2017). The size of D. rotundata genome was estimated to be ~570-Mbp using flow cytometry and k-mer analyses of genome sequences. The genome contained 26,198 protein coding genes. White Guinea yam genome sequence is distant from those of other monocotyledon species including Poales (Oryza and Brachypodium), Arecales (Elaeis and Phoenix), and Zingiberales (Musa), indicating that Dioscorea lineage was split from other monocotyledons early in the evolution. Publication of the reference genome, which also reported a DNA marker for sex identification in D. rotundata, has served as a catalyst for further studies into genomes of not only D. rotundata (Scarcelli et al. 2019; Bhattacharjee et al. 2020; Zhang et al. 2020; Sugihara et al. 2020) but also of D. alata (Cormier et al. 2019a, b; Sharif et al. 2020) and the pathogens associated with yams such as yam mosaic virus (Silva et al. 2019). D. rotundata scaffolds were generated by combining two types of Illumina short reads: paired-end and mate-pair jump reads. The mate-pair jump reads bridged the contigs assembled using paired-end short reads, with the gaps between the contigs represented with “N” for missing bases. The scaffolds were then ordered to generate a chromosome-level reference genome guided by a linkage map generated by the DNA markers of restriction site associated DNA (RAD)-tags (Baird et al. 2008) using the pseudo-testcross mapping method (Grattapaglia and Sederoff 1994). This reference genome has been shared with the wider scientific community via Ensembl (Howe et al. 2019; ftp://ftp.ensemblgenomes.org/pub/plants/release-48/fasta/dioscorea_rotundata).

Recently, the reference genome of D. rotundata has been updated with long reads generated by Oxford Nanopore Technologies (Sugihara et al. 2020). Long read-based de novo genome assembly resulted in longer contigs with a minimum number of missing bases, unlike those constructed by mate-pair jump reads. The newly generated contigs were also ordered using a linkage map generated by massive number of single nucleotide polymorphism (SNP) markers, instead of RAD-markers, obtained from whole-genome re-sequencing of 156 F1 progeny. This reference genome has been made available via Ensembl (Howe et al. 2019) as D. rotundata reference genome ver. 2 (https://plants.ensembl.org/Dioscorea_rotundata/Info/Index).

Reference genomes of two other Dioscorea species are also available in the public database. The recent chromosome-level reference genome of D. alata is yet to be published, but it is accessible on YamBase (https://yambase.org/). Additionally, the reference genome of D. dumetorum, which was generated by sequencing with Oxford Nanopore Technologies, has been reported (Siadjeu et al. 2020; https://pub.uni-bielefeld.de/record/2941469). Both D. rotundata and D. alata belong to the botanical section Enantiophyllum. The D. dumetorum reference genome represents the first genome sequenced from the section Lasiophyton. These multiple reference genomes will facilitate the comparative genomics and pangenome of Dioscorea species and be especially useful for the analysis of the transition of the sex-determination locus within Dioscorea (Cormier et al. 2019a; see Sect. 6).

5 Origin and Domestication of Dioscorea Species

Yams of different Dioscorea species are believed to be independently domesticated in different continents: D. rotundata and D. cayenensis in West and Central Africa, D. alata in Southeast Asia, and D. trifida in South America. However, our knowledge of their origins has been limited until recently. This is mainly due to the frequent hybridization and polyploidization of many species including D. rotundata (Terauchi et al. 1992; Scarcelli et al. 2006; Chaïr et al. 2010; Girma et al. 2014; Scarcelli et al. 2017; Sugihara et al. 2020) and D. alata (Chaïr et al. 2016; Sharif et al. 2020). The recent population genomics studies have started unveiling the domestication processes of the major species (Scarcelli et al. 2019; Sugihara et al. 2020; Sharif et al. 2020).

5.1 Origin and Domestication of Guinea Yam

White Guinea yam, D. rotundata, is a true cultigen and the wild species D. abyssinica and D. praehensilis have been proposed as its two candidate progenitors (Coursey 1976a, b; Terauchi et al. 1992; Scarcelli et al. 2006; Chaïr et al. 2010; Girma et al. 2014; Scarcelli et al. 2017; Magwé-Tindo et al. 2018). D. abyssinica and D. praehensilis are distributed in savannah and rainforest areas, respectively, of West and Central Africa. By comparing whole-genome sequences of 80 D. rotundata, 29 D. abyssinica, 26 D. praehensilis from West Africa (Western D. praehensilis), and 18 D. praehensilis from Cameroon (Cameroonian D. praehensilis), Scarcelli and colleagues recently proposed that D. rotundata was domesticated from D. praehensilis in the northern part of Benin within the Niger River basin (Scarcelli et al. 2019). This report is a major contribution towards elucidating the origin of Guinea yam. However, after a careful reassessment of the Scarcelli et al. (2019) data and results, we reached at a different conclusion (Sugihara et al. 2020).

Our study included genome sequences of 336 D. rotundata accessions in addition to the D. rotundata and wild species accessions analyzed by Scarcelli et al. (2019). First, we conducted clustering analysis and flow cytometry analysis of our 336 D. rotundata accessions. Based on these results, we classified 308 accessions as diploid and 28 accessions as triploid. Focusing on the diploid D. rotundata accessions, we attempted to elucidate their phylogenetic relationships with the wild relative species; D. abyssinica, Western D. praehensilis, and Cameroonian D. praehensilis as reported by Scarcelli et al. (2019). We reconstructed a rooted neighbor-joining tree of four African yam taxa using D. alata, an Asian species, as an outgroup (Fig. 4a). In our result, D. rotundata was genetically closer to D. abyssinica than to D. praehensilis, which was not consistent to Scarcelli’s hypothesis indicating that D. rotundata was directly domesticated from D. praehensilis. To test whether D. rotundata was derived from D. abyssinica or D. praehensilis, we focused on the allele frequencies on 144 SNPs which were oppositely fixed in D. abyssinica and D. praehensilis (Fig. 4b). If D. rotundata was derived from either of D. abyssinica or D. praehensilis, allele frequencies of the analyzed SNPs should be highly skewed to either of the candidate progenitors. However, the allele frequencies of D. rotundata were intermediate. This observation suggested a hybrid origin of D. rotundata between D. abyssinica and D. praehensilis. To conform this hypothesis, we compared the three evolutionary models by ∂a∂i which assesses evolutionary models with their likelihood based on the site frequency spectrum (Gutenkunst et al. 2009). The first model was that D. rotundata had been derived from D. abyssinica. The second model was that D. rotundata had been derived from D. praehensilis (Scarcelli’s hypothesis). The third model was that D. rotundata had been hybrid-derived between D. abyssinica and D. praehensilis. As a result, ∂a∂i showed the highest likelihood in the third model (the hybrid origin of D. rotundata) out of the three models.

Fig. 4
figure 4

Domestication history of Guinea yam. (a) Neighbor-joining tree of four African yam taxa reconstructed using D. alata as an outgroup (adopted from Sugihara et al. 2020, Fig. 1c). (b) Frequencies of individuals homozygous for D. abyssinica allele (A: indicated by yellow color), homozygous for D. praehensilis allele (P: indicated by blue color), and heterozygous for A and P (indicated by white color) among the 388 D. rotundata sequences as studied for 144 SNPs (adopted from Sugihara et al. 2020, Fig. 2b). (c) Demographic history of D. rotundata as inferred by ∂a∂i (Gutenkunst et al. 2009). N, T, M, and f represent a relative population size from the ancestral population (NAP), divergence time, migration rate, a fraction of genomic contribution, respectively. The ∂a∂i analysis inferred the hybrid origin of D. rotundata with the high migration rates between D. rotundata and the two wild relatives (adopted from Sugihara et al. 2020, Fig. 2c). (d) Haplotype network of the whole plastid genomes of 416 D. rotundata (including the triploid accessions), 68 accessions of wild relative species, and two D. alata accessions used as the outgroup. The number of vertical dashes represents the number of mutations (adopted from Sugihara et al. 2020, Fig. 3a). (e) Reticulated evolutionary history of Guinea yam (Sugihara et al. 2020; Girma et al. 2014). D. rotundata (white Guinea yam) is derived from a homoploid hybridization between D. abyssinica and D. praehensilis. D. cayenensis (yellow Guinea yam) and the majority of triploid D. rotundata are derived from the polyploid hybridization between a female diploid D. rotundata and a male African wild yam. At least four species contributed to the gene pool of Guinea yam

Our finding suggested that D. rotundata is most likely a homoploid hybrid between D. abyssinica and D. praehensilis (Fig. 4b, c; Sugihara et al. 2020). Homoploid hybrid speciation is the formation of a new hybrid species without altering the ploidy levels of the parents (Mallet 2007). The origin of D. rotundata by hybridization seems to be recent when compared with timing of the speciation of D. praehensilis from D. abyssinica (Fig. 4c, d). Genomic contributions from D. abyssinica and D. praehensilis during the hybridization event were estimated to be ~68% and 32%, respectively (Fig. 4c). The relative population size of D. rotundata is much smaller than those of its wild relatives, which indicates that D. rotundata was affected by domestication bottleneck. Chloroplast DNA is predominantly inherited maternally in angiosperms. We extracted chloroplast DNA sequences from the whole-genome sequence reads and studied its polymorphisms in our samples. A chloroplast DNA haplotype network suggested that D. abyssinica and D. praehensilis were the maternal and paternal parents of D. rotundata, respectively (Fig. 4d, e; Sugihara et al. 2020). The hybrid origin of white Guinea yam was initially proposed by Coursey in 1976 based on morphological comparisons (Coursey 1976a). Our results from genome analyses support his hypothesis that spontaneous hybridization between wild yams could have occurred at the artifactual “dump heaps” created by people living in the savannah between the forest and the Sahara (Coursey 1976b).

The most common cases of origin of crops by hybridization accompanies polyploidization (also known as allopolyploidization) as exemplified by bread wheat (Triticum aestivum) (Peng et al. 2011), banana (Musa acuminata) (Heslop-Harrison and Schwarzacher 2007), cotton (Gossypium spp.) (Zhang et al. 2015), and canola (Brassica napus) (Chalhoub et al. 2014). The homoploid hybrid speciation in D. rotundata is unique in that its domestication did not involve polyploidization.

We hypothesize that Guinea yam has been established by the process of “ennoblement” (Dumont and Vernier 2000; Mignouna and Dansi 2003; Scarcelli et al. 2006; Chaïr et al. 2010). “Ennoblement” is a traditional farmers’ practice that involves collecting tubers of wild yams from the bush and forest and planting them in their fields, and it likely contributes to the genetic diversity of yam through hybridization and introgression (Jarvis and Hodgkin 1999; Scarcelli et al. 2006; Chaïr et al. 2010). Similar farmers’ practices of sympatrically planting crop wild relatives and cultivars often occur in the centers of origin of domesticated plants (Jarvis and Hodgkin 1999).

The locus of the SWEETIE gene is an example that the crop wild relatives contributed to the genetic diversity of D. rotundata. The SWEETIE gene is known to be involved in sugar flux (Veyres et al. 2008a, b). In the diploid D. rotundata, this locus showed the signature of extensive introgressions from D. abyssinica and D. praehensilis (Fig. 5). We identified the introgressions by applying f4 statistic (Reich et al. 2009) of population genomics to our diploid D. rotundata accessions and its wild progenitors as analyzed in Scarcelli et al. (2019). The f4 statistic can detect genomic regions showing discordant topologies among different genetic groups of D. rotundata with respect to D. abyssinica and D. praehensilis. We hypothesize that this locus was under selection.

Fig. 5
figure 5

Genetic relationship among D. rotundata and its wild progenitors around the SWEETIE gene. A Neighbor-Net (adopted from Sugihara et al. 2020, Fig. 4c) was constructed by SplitsTree (Huson and Bryant 2006). Based on STRUCTURE analysis, five clusters (cluster 1–5) were defined in Sugihara et al. (2020). In this analysis, the accessions in cluster 1 were removed because they were triploid. Based on this result, the gene of cluster 2 and 5 were genetically close to D. abyssinica, while that of cluster 4 was genetically close to D. praehensilis. This different genetic relationship in the different clusters indicate the introgressions from both wild progenitors

In addition to the majority of diploid accessions, triploid accessions of D. rotundata have been identified by flow cytometry analysis (Girma et al. 2014; Sugihara et al. 2020). These triploid D. rotundata accessions seem to have been derived from a hybridization between the diploid D. rotundata and the wild yam species D. togoensis (Fig. 4e; Girma et al. 2014). Since the chloroplast sequences of the triploid D. rotundata were shared with the diploid D. rotundata, their maternal and paternal parents are likely the diploid D. rotundata and D. togoensis, respectively (Fig. 4d; Sugihara et al. 2020). Successful interspecific crosses between D. rotundata and D. togoensis were also reported (Girma et al. 2014). The triploid D. rotundata formed by autopolyploidization seems to be the minority as compared to those formed by allopolyploidization (Girma et al. 2014). It is not easy to distinguish the triploid D. rotundata accessions from the diploid accessions based on morphology. However, some morphological traits, such as the presence of barky patches, absence of waxiness, and dark green leaf color, have been shown to correlate with ploidy level (Girma et al. 2014). Agricultural importance of the triploid D. rotundata is yet to be studied.

D. cayenensis (yellow Guinea yam) is another species that is likely a triploid hybrid between the diploid D. rotundata and the rainforest-adapted wild species D. burkilliana (Fig. 4e; Terauchi et al. 1992; Girma et al. 2014). Based on chloroplast DNA polymorphisms, it was suggested that diploid D. rotundata is the maternal parent of D. cayenensis (Terauchi et al. 1992). Using nuclear ribosomal DNA polymorphisms (Terauchi et al. 1992) and Genotype-by-Sequencing (GBS) analysis (Girma et al. 2014), it was inferred that D. burkilliana is the paternal parent. D. cayenensis has a woody corm above the fleshy tuber, and this trait is shared with D. burkilliana. Interestingly, D. burkilliana is also subjected to “ennoblement” together with D. abyssinica and D. praehensilis (Mignouna and Dansi 2003), which probably contributed to its hybridization with D. rotundata.

D. mangenotiana (syn. D. baya) is genetically close to Guinea yam (Girma et al. 2014; Magwé-Tindo et al. 2018). Previously, D. mangenotiana and D. baya were considered different species. However, a recent study proposed D. mangenotiana as the adult form of D. baya (Magwé-Tindo et al. 2018). D. mangenotiana is a rainforest-adapted wild species characterized by spiny stems and the production of very big tubers (Figs. 2d and 3f; Dounias 2001). The bases of its fleshy edible tubers are attached to very large woody corms, which probably provide protection from herbivores. In Southern Cameroon, D. mangenotiana has long been subjected to “paracultivation” by the Baka Pygmies (Dounias 2001). “Paracultivation” is the exploitation and maintenance of wild plants in their original/natural environments (Dounias 2001). Intriguingly, D. mangenotiana was reported as a triploid species (Girma et al. 2014) with a large number of heterozygous DNA markers, which probably suggests that it may be an allopolyploid. However, no research has been carried out to date to identify the ancestors of D. mangenotiana.

5.2 Origin and Domestication of D. alata

D. alata (greater yam) is widely cultivated in pantropical regions (Asia, the Pacific, Africa, and the Caribbean) in contrast to D. rotundata that is restricted to Africa and the Caribbean. This worldwide dispersion of D. alata occurred mainly through human migrations (Sharif et al. 2020). As discussed above, the domestication process of D. rotundata seems mainly driven by hybridization and introgression. By contrast, the domestication of D. alata was mainly driven by vegetative propagation and autopolyploidization, which caused erratic or no flowering in this species (Arnau et al. 2010; Sharif et al. 2020). This predominant vegetative propagation in D. alata is in line with its very low nucleotide diversity (π = 0.96 ~ 1.29 × 10−5 in Fig. 6: Sharif et al. 2020) as compared with the higher level of nucleotide diversity in D. rotundata (π = 1.48 × 10−3) (Sugihara et al. 2020).

Fig. 6
figure 6

Global dispersion of D. alata (Sharif et al. 2020) and distributions of other cultivated yam species. Figures in purple color represent nucleotide diversities of the diploid D. alata accessions in different regions. D. alata was originated in Southeast Asia (①). After its domestication in Southeast Asia, it reached Africa via the Indian Peninsula (② and ③), and finally reached the Caribbean (④) from Africa. The introduction to Africa might be through either Madagascar (broken line in ③) or East Africa via the Swahili coast (solid line in ③). The gradual reduction in D. alata nucleotide diversities clearly suggests the founder effects

Sharif et al. (2020) used genome sequence analysis of 643 accessions spanning four continents to study the dispersion route of D. alata, which confirmed the region around tropical Eastern Asia as the geographical origin of D. alata. This finding supports the previous proposal by Burkill (1960) that D. alata was originated in the tropical Eastern Asia. Additionally, Sharif and colleagues hypothesized that D. alata was domesticated independently in the Mainland Southeast Asia and the Pacific (① in Fig. 6). Their hypothesis is based on the early divergence between the two regions estimated for D. alata (Sharif et al. 2020) as well as the estimated date of early human settlement of Sahul (at least 50,000 years ago) (Bird et al. 2019). After the divergence between the subgroups of Mainland Southeast Asia and the Pacific, D. alata reached the Indian subcontinent (② in Fig. 6). According to their demographic inference, there was a continuous migration between the Indian subcontinent and the Pacific. From the Indian subcontinent, D. alata reached Africa (③ in Fig. 6). Since the African and Madagascar D. alata are genetically closer to that of the Indian subcontinent than to that from the Mainland Southeast Asia, this dispersion seems to have occurred via the Indian subcontinent (Sharif et al. 2020). The detailed route of how D. alata reached Africa is still unclear. However, it might be through either Madagascar or East Africa via the Swahili coast (③ in Fig. 6). The relatively low nucleotide diversity of African accessions suggests a founder effect from the Indian subcontinent (Sharif et al. 2020). From Africa, D. alata was introduced into the Caribbean (④ in Fig. 6) most probably during the Colombian exchange, which introduced African crops to the tropical America (Boivin et al. 2012). The lowest nucleotide diversity of the Caribbean accessions suggests a strong founder effect from Africa (Sharif et al. 2020; Fig. 6). Taken together, the decreasing nucleotide diversity of D. alata clearly reflects the route for its dispersion.

The distribution of the clonal and polyploid accessions of D. alata also reflects the route of dispersion of the species (Sharif et al. 2020). Approximately 68% of the inferred clonal clusters (multi-locus lineages in Sharif et al. 2020) had an intercontinental distribution. The Caribbean accessions had the highest number of the intercontinentally shared clonal clusters, while those of Asia had the lowest. Therefore, the distribution of the clonal clusters reflects dispersion of this species through vegetative propagation. The dispersion route of D. alata was also reflected in the distribution of its polyploid accessions. Most of the polyploid accessions are genetically close to either the Asian or the Pacific accessions. This suggests that the polyploidization of D. alata occurred several times before migration of the species from Asia and the Pacific to Africa and the Caribbean. Moreover, most triploid accessions were genetically closer to the Asian lineages rather than to the Pacific lineages. This suggests that most triploid accessions were derived from Asia.

The wild progenitor of D. alata is still unknown. However, some studies revealed that D. alata is phylogenetically close to the wild species D. nummularia, D. transverse, and D. hastifolia (Malapa et al. 2005; Chaïr et al. 2016). Especially, D. nummularia, a species native to Melanesia and Island Southeast Asia, is also subjected to a “paracultivation”-like practice (Dounias 2001; Chaïr et al. 2016). Interestingly, cultivars of natural hybrid between D. alata and D. nummularia have been reported and known as “strong yam” by farmers of Vanuatu (Chaïr et al. 2016). These interspecific hybrids were treated as unidentified taxa or erroneously assigned to D. transverse (Malapa et al. 2005; Chaïr et al. 2016). This mis-assignment is probably related to their potential hybrid status (Chaïr et al. 2016). Recently, researchers have started paying attention to the diversity of D. nummularia (Lebot et al. 2017) and attempts are made to introduce the resistance trait of D. nummularia against the anthracnose disease caused by Colletotrichum pathogens into D. alata by artificial hybridization (Lebot et al. 2019). Although the cross between D. alata and D. nummularia resulted in fertile seeds, D. nummularia is not regarded as a direct ancestor of D. alata (Chaïr et al. 2016). Further population genomics study is needed by including multiple accessions of D. nummularia, D. transverse, and D. hastifolia to clarify the origin of D. alata.

5.3 Origin and Domestication of D. trifida

D. trifida of the section Macrogynodium was domesticated in South America and is commonly referred to as the indigenous “Amerindian” yam. Genetic segregation study and cytogenetics suggested that the cultivated D. trifida is autotetraploid (2n = 4× = 80) (Bousalem et al. 2006), while a wild D. trifida with diploid genome (2n = 2× = 40) was found in French Guyana (Bousalem et al. 2010). A phylogenetic study based on amplified fragment length polymorphism (AFLP) clearly separated the cultivated (4×) and wild (2×) D. trifida (Bousalem et al. 2010). A region where the diploid D. trifida predominate was also found. D. trifida represents a clear case that whole-genome duplication may have played an important role in the domestication of Dioscorea. The autopolyploidization is not a preferred subject in population genomics, but this may help to calibrate the time when the polyploidization happened.

5.4 Origin and Domestication of D. dumetorum

D. dumetorum, known as the “trifoliate yam” because of its trifoliate leaves, belongs to the section Lasiophyton. D. dumetorum is widely consumed in West and Central Africa and occurs in both cultivated and wild forms. Despite its high yielding nature and nutritional richness, severe postharvest hardening of the tubers makes D. dumetorum a minor crop (Sefa-Dedeh and Afoakwa 2002). Genetic diversity study of D. dumetorum revealed that the accessions from Togo and Nigeria had the highest genetic diversity (Sonibare et al. 2010). This indicates that the center of genetic diversity and the possible origin of D. dumetorum might be around Togo and Nigeria. Another study focusing on the accessions from Cameroon identified gene flow and admixture among the accessions, which is probably caused by farmers’ breeding practices (Siadjeu et al. 2018). The same study revealed different ploidy levels in D. dumetorum and showed that diploids and triploids have different geographical distributions in Cameroon. Interestingly, population genomics study showed that the distribution of the triploid D. dumetorum was positively correlated with the region containing a higher level of gene flow (Siadjeu et al. 2018).

6 Evolution of Sex in Dioscorea Species

Most Dioscorea species are dioecious, bearing male and female flowers on separate individuals. This sexual system affects genetics and population genomics of the genera and consequently deserves a special attention. Dioecious plants account for ~5–6% of angiosperm species (Renner 2014). Based on its scattered taxonomic distribution, dioecy is suggested to have evolved recently and independently from hermaphroditic co-sexual ancestors (Renner and Ricklefs 1995; Charlesworth 2002). Many studies have focused on understanding the process of this large-scale convergent evolution from co-sexual to dioecy in plants. To elucidate the process, sex-determination systems have been studied in several taxa (Akagi et al. 2016, 2019; Harkess et al. 2020).

In Dioscorea, multiple sex-determination systems have been reported by cytological observations and molecular analyses (Table 2). As part of our work on the genetics of D. tokoro, a wild species from East Asia, we studied the inheritance and segregation pattern of AFLP markers in an F1 family derived from a cross between male and female D. tokoro plants, which suggested an XY/XX (male/ female) sex-determination system in this species (Terauchi and Kahl 1999). Although most Dioscorea species have male heterogametic sex-determination system, female heterogametic sex-determination system (ZZ/ZW) and extra chromosomes in female (XO/XX) have also been reported (Table 2). Sex change of individuals was observed in D. rotundata having female heterogametic sex-determination system (ZZ/ZW) (Tamiru et al. 2017). Here, we review the sex-determination systems of ZZ/ZW and XY/XX that occur in D. rotundata and D. alata, respectively.

Table 2 Reported sex-determination systems in Dioscorea

Sex in D. rotundata is regulated by a female-specific genomic region that we recently identified by QTL-seq analysis of an F1 progeny segregating for sex (Tamiru et al. 2017). QTL-seq is an NGS-based bulked segregant analysis (BSA) method to identify the genomic regions underlying traits of interest using progeny derived from crosses made between cultivars/lines showing contrasting phenotypes for the traits (Takagi et al. 2013; Itoh et al. 2019). To identify the genomic region associated with sex in D. rotundata, QTL-seq was applied to sequences generated for male and female DNA pools prepared from an F1 progeny derived from a cross between male and female plants. Accordingly, a candidate genomic region was detected on chromosome 11 using SNP markers that were heterozygous in the female parent whereas no candidate regions were detected when SNP markers heterozygous in the male parent were used. The candidate genomic region showed significant structural differences between the male and female sequences. PCR amplification and short read mapping analysis further identified a female-specific region within the candidate genomic region delineated by QTL-seq. These results suggest that the sex-determination system of D. rotundata is ZZ/ZW (male/female), not XO/XX (male/female).

A DNA maker, “sp16,” was developed within the female-specific (W-) genomic region of D. rotundata for prediction of sex of the plant at the seedling stage (Tamiru et al. 2017). The usefulness of the DNA marker has been demonstrated in diverse D. rotundata accessions (Agre et al. 2020; Denadi et al. 2020). However, the marker type was not perfectly associated with sex, suggesting the manifestation of sex is unstable in D. rotundata over a time period (Tamiru et al. 2017). Sex change is widely known in plants and animals (Policansky 1982). Interestingly, this sex change was rarely observed in the male D. rotundata plants, suggesting a gene on the W-region with “sp16” marker seems to suppress maleness, and its effect is unstable.

In our recent study, we also investigated the genomic contribution of D. abyssinica and D. praehensilis to D. rotundata chromosome-wise (Fig. 7; Sugihara et al. 2020). The chromosome 11 of D. rotundata harboring the sex-determination locus was highly skewed towards that of D. abyssinica. As described in the previous Sect. 5.1 above, D. rotundata is likely a hybrid species derived from D. abyssinica and D. praehensilis, and the genetic divergences from both wild progenitors are basically similar across the genome (Sugihara et al. 2020). However, of all the chromosomes, chromosome 11 of D. rotundata had the shortest genetic distance from that of D. abyssinica and the largest genetic distance from that of D. praehensilis (Sugihara et al. 2020). Similar interspecies divergence differences between autosomes and sex chromosome have also been reported in the dioecious plant species of the genus Silene (Hu and Filatov 2016).

Fig. 7
figure 7

Genomic scan of FST values between cultivated yam (D. rotundata) and two wild species D. abyssinica (top) and D. praehensilis (bottom) (adopted from Sugihara et al. 2020, Fig. 2d). Sliding window analysis was conducted with 100-kb window and 20-kb step. Chromosome 11 of D. rotundata containing the sex-determining locus shows the shortest distance to that of D. abyssinica and the largest distance to that of D. praehensilis

D. alata has a male heterogametic sex-determination system (XY/XX) (Cormier et al. 2019a), which is like the majority of Dioscorea species (Table 2). D. alata belongs to the section Enantiophyllum with D. rotundata, a species with a female heterogametic sex-determination system (ZZ/ZW). Interestingly, the genomic region associated with sex of D. alata was identified on linkage group 6, which also corresponded to chromosome 6 of D. rotundata (Cormier et al. 2019a). The sex-determination locus of D. rotundata was identified on chromosome 11 (Tamiru et al. 2017), suggesting the transition of sex-determination system from XY to ZW occurred in the section Enantiophyllum. A similar transition of sex-determination system was reported in the genus Populus. Most species Populus have the XY sex-determination system, while P. alba has the ZW sex-determination system (Müller et al. 2020).

As discussed above, the genus Dioscorea contains many dioecious species with divergent sex-determination systems. Transition of sex-determination systems are presumably involved in species divergence (Kumar et al. 2014). Further genomics research in Dioscorea species will help identify the causative genes involved in sex determination and to clarify how the sex-determination systems have diversified in the genus. Such information will also allow us to manipulate sex of plants for effective cross breeding of yam crops.

7 Conclusions and Future Perspectives

Population genomics and cytogenetics studies have revealed important domestication processes in Dioscorea species, but many questions still remain. For example, we still do not know the key traits and the genes involved in yam domestications, although some studies have identified genes showing signature of selection in D. rotundata including SWEETIE gene in our study (Akakpo et al. 2017; Scarcelli et al. 2019; Sugihara et al. 2020). D. abyssinica and D. praehensilis, the wild relatives of D. rotundata, are subjected to an ongoing practice of “ennoblement.” Additionally, it has been shown that the most cultivars introduced by “ennoblement” are indeed hybrids between the cultivated and wild yams (Scarcelli et al. 2006). These findings probably indicate that the wild species cannot directly be domesticated to become cultivars and that hybridization was necessary to generate white Guinea yam cultivars. Similar interspecific hybridization was also reported in D. alata (Chaïr et al. 2016). Consequently, analyzing hybridization is important to understand what attributes characterize D. rotundata and other cultivated yams. Probably, D. rotundata was established as a cultivar as a result of heterosis derived from the hybridization between D. abyssinica and D. praehensilis.

Understanding the genomes of crop wild relatives would facilitate efficient breeding programs. Crop wild relatives are expected to have potentially beneficial alleles that are not available in the cultivars. The farmers unconsciously introduce these beneficial alleles to the cultivars presumably by “ennoblement.” Since the genomic regions containing the beneficial alleles should be affected by selective sweeps, population genomics analyses may be able to identify these regions (Akakpo et al. 2017; Scarcelli et al. 2019; Sugihara et al. 2020). Currently, there is no evidence that these candidate selective sweeps affected any phenotypes. However future functional studies of the identified genes would reveal their impact on the change of traits in the crops.

Another standing question is how many times the domestication processes occurred in the various cultivated Dioscorea species. A recent study hypothesized multiple domestication processes of D. alata in separate regions (Sharif et al. 2020). The cultivated yam landraces from Southern Ethiopian are phylogenetically close to the cultivated gene pools of D. rotundata, but they were clearly separate from Nigerian D. rotundata (Tamiru et al. 2007). Although the model-based population genetics/genomics is needed to infer the detailed demographic history, this result may suggest independent domestication processes of D. rotundata in Ethiopia (or East Africa) and Nigeria.

In this chapter, the importance of hybridization and polyploidization for the domestication of Dioscorea species has been discussed. Some of these events appear to have played an important role in yam domestication. In recent years, our knowledge of yam domestication has dramatically improved thanks to the advances in sequencing technologies and statistical methods for population genomics analysis. These developments also allowed us to identify, among others, the transition of the sex-determination system in the section Enantiophyllum. Future studies should further unravel the complex evolutionary history of Dioscorea species including hybridization, polyploidization, and sexual/asexual propagation.