Genome analyses reveal the hybrid origin of the staple crop white Guinea yam (Dioscorea rotundata) Yu Sugiharaa, Kwabena Darkwab, Hiroki Yaegashic, Satoshi Natsumec, Motoki Shimizuc, Akira Abec, Akiko Hirabuchic, Kazue Itoc, Kaori Oikawac, Muluneh Tamiru-Olic,d, Atsushi Ohtaa, Ryo Matsumotob, Paterne Agreb, David De Koeyerb,e, Babil Pachakkilf,g, Shinsuke Yamanakaf, Satoru Muranakaf, Hiroko Takagif, Ben Whiteh, Robert Asiedub, Hideki Innani, Asrat Asfawb,1, Patrick Adebolab,1, and Ryohei Terauchia,c,1 aLaboratory of Crop Evolution, Graduate School of Agriculture, Kyoto University, Kyoto 606-8502, Japan; bInternational Institute of Tropical Agriculture, Ibadan 200001, Nigeria; cIwate Biotechnology Research Center, Kitakami, Iwate 024-0003, Japan; dDepartment of Animal, Plant, and Soil Sciences, School of Life Sciences, La Trobe University, Melbourne, VIC 3086, Australia; eFredericton Research and Development Centre, Agriculture and Agri-Food Canada, Fredericton, NB E3B 4Z7, Canada; fJapan International Research Center for Agricultural Sciences, Tsukuba 305-8686, Japan; gDepartment of International Agricultural Development, Tokyo University of Agriculture, Tokyo 183-8538, Japan; hEarlham Institute, Norwich NR4 7UZ, United Kingdom; and iLaboratory of Population Genetics and Genome Evolution, The Graduate University for Advanced Studies, Hayama 240-0193, Japan Edited by Loren H. Rieseberg, University of British Columbia, Vancouver, BC, Canada, and approved November 2, 2020 (received for review July 27, 2020) White Guinea yam (Dioscorea rotundata) is an important staple praehensilis, and 34 D. abyssinica accessions suggested that dip- tuber crop in West Africa. However, its origin remains unclear. In loid D. rotundata was domesticated from D. praehensilis (10). this study, we resequenced 336 accessions of white Guinea yam Here we addressed this hypothesis using an expanded set of and compared them with the sequences of wild Dioscorea species genomes from cultivated and wild Dioscorea species. using an improved reference genome sequence of D. rotundata.In In this study, we generated an improved version of the Guinea contrast to a previous study suggesting that D. rotundata origi- yam reference genome and used it to analyze the genomes of 336 nated from a subgroup of Dioscorea praehensilis, our results sug- accessions of D. rotundata and its wild relatives. Based on these gest a hybrid origin of white Guinea yam from crosses between analyses, we attempted to reveal the history of Guinea yam do- the wild rainforest species D. praehensilis and the savannah- mestication. Our results suggest that diploid D. rotundata was adapted species Dioscorea abyssinica. We identified a greater ge- most likely derived from homoploid hybridization between D. D. abyssinica nomic contribution from in the sex chromosome of abyssinica and D. praehensilis. By evaluating the genomic con- EVOLUTION Guinea yam and extensive introgression around the SWEETIE tributions of each parental species to D. rotundata, we revealed gene. Our findings point to a complex domestication scenario for greater representation of the D. abyssinica genome in the sex Guinea yam and highlight the importance of wild species as gene chromosome of D. rotundata and a signature of extensive in- donors for improving this crop through molecular breeding. trogression in the SWEETIE gene on chromosome 17. domestication | Guinea yam | hybrid | population genomics | wild Genetic Diversity of Guinea Yam progenitors We obtained DNA samples from 336 accessions of D. rotundata maintained at the International Institute of Tropical Agriculture ams (Dioscorea spp.) are major starchy tuber crops that are Ywidely consumed in the tropics. Ten yam species are culti- Significance vated worldwide, including Dioscorea alata in Southeast Asia, Dioscorea trifida in South America, and Dioscorea rotundata in Guinea yam is an important staple tuber crop in West Africa, West and Central Africa (1). D. rotundata, also known as white where it contributes to the sustenance and sociocultural lives Guinea yam, is the most important species in West and Central of millions of people. Understanding the genetic diversity of Africa, an area accounting for 92.5% of global yam production in Guinea yam and its relationships with wild relatives is impor- 2018 (http://www.fao.org/statistics). Beyond its nutritional and tant for improving this important crop using genomic infor- food value, Guinea yam is also important for the culture of West mation. A recent genomics study proposed that Guinea yam African people (2). originated from a wild relative, the rainforest species Dio- Despite the considerable importance of Guinea yam, its origin scorea praehensilis. Our results based on sequencing of 336 has been elusive. There are two types of Guinea yam: white Guinea yam accessions do not support this notion; rather, our Guinea yam (D. rotundata) and yellow Guinea yam (Dioscorea results indicate a hybrid origin of Dioscorea rotundata from cayenensis). D. cayenensis is thought to be a triploid species of crosses between the savannah species Dioscorea abyssinica hybrid origin, with D. rotundata as the maternal parent and and D. praehensilis. Dioscorea burkilliana as the paternal parent (3, 4). In turn, the triploid D. rotundata is thought to be a hybrid between D. rotun- Author contributions: M.T.-O., R.M., D.D.K., H.T., R.A., A. Asfaw, P. Adebola, and R.T. data and Dioscorea togoensis (4). However, the origin of diploid D. designed research; Y.S., K.D., H.Y., S.N., M.S., A. Abe, A.H., K.I., K.O., M.T.-O., R.M., rotundata, which accounts for the majority of Guinea yam pro- P. Agre, D.D.K., B.P., S.Y., S.M., H.T., B.W., R.A., H.I., A. Asfaw, P. Adebola, and R.T. duction (4), has been ambiguous. Two wild species are candidate performed research; Y.S., H.Y., S.N., M.S., A. Abe, A.H., K.I., K.O., A.O., R.M., P. Agre, D.D.K., B.P., S.Y., B.W., and H.I. contributed new reagents/analytic tools; Y.S., H.Y., S.N., progenitors of diploid D. rotundata: the savannah-adapted wild M.S., A. Abe, A.O., R.M., B.P., H.T., B.W., R.A., H.I., A. Asfaw, P. Adebola, and R.T. analyzed species Dioscorea abyssinica and the rainforest-adapted wild spe- data; and Y.S., M.T.-O., R.A., H.I., A. Asfaw, P. Adebola, and R.T. wrote the paper. cies Dioscorea praehensilis (3–10). The geographical distributions The authors declare no competing interest. of D. abyssinica and D. praehensilis overlap slightly (SI Appendix, This article is a PNAS Direct Submission. Fig. S1). Based on morphological evaluation, Coursey proposed This open access article is distributed under Creative Commons Attribution License 4.0 that D. rotundata might be a hybrid between the two species (8). (CC BY). However, other reports have indicated that the origin of Guinea 1To whom correspondence may be addressed. Email: [email protected], P.Adebola@ yam is ambiguous due to the small number of markers (3–7), in- cgiar.org, or [email protected]. trogression (6, 7), or incomplete lineage sorting (7). This article contains supporting information online at https://www.pnas.org/lookup/suppl/ The whole-genome sequence of Guinea yam has been repor- doi:10.1073/pnas.2015830117/-/DCSupplemental. ted (11). A recent genome study involving 86 D. rotundata,47D. www.pnas.org/cgi/doi/10.1073/pnas.2015830117 PNAS Latest Articles | 1of6 Downloaded by guest on September 24, 2021 (IITA) in Nigeria, representing the genetic diversity of Guinea are genetically close to the D. rotundata accessions reported previ- yam landraces and improved lines from West Africa. We sub- ously (10) (Fig. 1C). However, the NJ tree showed that D. rotundata jected these samples to whole-genome resequencing on the is more closely related to D. abyssinica than to Western D. prae- Illumina sequencing platform. We aligned the resulting short hensilis (Fig. 1C), which is inconsistent with a previous finding (10) reads to the newly assembled reference genome (SI Appendix, that D. rotundata is most closely related to Western D. praehensilis. sections S1 and S2) and extracted single nucleotide polymorphism To elucidate the evolutionary relationships of the three wild (SNP) information for use in genetic diversity studies (SI Appen- Dioscorea species that are closely related to D. rotundata—D. dix, Table S1 and section S3). Based on admixture analysis with A abyssinica (designated as A), Western D. praehensilis (P), and the sNMF program (12), we defined five major clusters (Fig. 1 ). — When K = 2, cluster 1 was clearly separated from the other ac- Cameroonian D. praehensilis (C) we performed diffusion ap- ∂ ∂ cessions. Principal component analysis (PCA) also separated proximations for demographic inference ( a i) analysis (15), which cluster 1 from the rest of the clusters (Fig. 1B). Accessions in allows for estimation of demographic parameters based on an un- cluster 1 had significantly higher heterozygosity and ∼10-fold more folded site frequency spectrum. First, we tested three phylogenetic unique alleles than those in the four remaining clusters (SI Ap- models—{{A, P}, C}, {{P, C}, A}, and {{C, A}, P}—using 17,532 pendix, Figs. S2 and S3 and Table S2). Because flow cytometry SNPs that were polarized using D. alata as an outgroup without analysis confirmed that all 10 accessions analyzed in cluster 1 were considering migration among the species. Of the three models, {{A, triploids (Dataset S1), we hypothesized that cluster 1 represents P}, C} had the highest likelihood (SI Appendix,TableS4). triploid D. rotundata, a hybrid of D. rotundata and D. togoensis (4). This result is not consistent with the previous finding that {{P, After removing the cluster 1 accessions, the nucleotide diversity of − C}, A} had the highest likelihood (10), as determined using a D. rotundata was estimated as 14.83 × 10 4 (SI Appendix,Table different method with fastsimcoal2 software (16). To exactly re- S3), which is ∼1.5-fold larger than that reported previously (10), presumably because we used a larger number of samples with peat the previous analysis, we tested these three models with diverse genetic backgrounds in our study.