|GREENBERG, ANTHONY - Bayesic Research|
|MCNALLY, KENNETH - International Rice Research Institute|
|JUNG, JANELLE - Cornell University|
|KIM, HYUN-JUNG - Cornell University|
|NAREDO, ELIZABETH - International Rice Research Institute|
|BANATICLA-HILARIO, CELESTE - International Rice Research Institute|
|MCCOUCH, SUSAN - Cornell University|
Submitted to: Crop Science Society of America
Publication Type: Abstract Only
Publication Acceptance Date: 11/12/2021
Publication Date: 11/12/2021
Citation: Greenberg, A.J., Edwards, J., McNally, K.L., Jung, J., Kim, H., Naredo, E.B., Banaticla-Hilario, C.N., Eizenga, G.C., McCouch, S.R. 2021. Modeling phenotypic groups reflects genetic relationships between cultivars and their wild relatives in rice. Abstract. ASA,CSSA,SSSA International Annual Meeting. Salt Lake City, Utah.
Technical Abstract: Wild relatives of cultivated crops provide a rich gene pool that can be mined for beneficial traits. Collections of wild accessions, while often sparsely genotyped, carry passport data including heritable phenotypes. To make use of such data, we implemented a Bayesian Gaussian mixture model to infer accession groups from data on multiple traits. To test the usefulness of this approach, we applied it to accessions of the Oryza rufipogon species complex (ORSC; the wild relative of cultivated rice) that had both phenotypic and genotypic data available. We found that phenotypic groups largely reflect genotypic population structure and life habit but provide additional information that allows us to identify the effects of gene flow from cultivated O. sativa and among ORSC populations. We identify a subset of traits that effectively predict genotypic populations. An R package, MuGaMix, is available to perform these types of analyses.