Escape to Ferality: The Endoferal Origin of Weedy Rice from Crop Rice through De-Domestication

Abstract

Domestication is the hallmark of evolution and civilization and harnesses biodiversity through selection for specific traits. In regions where domesticated lines are grown near wild relatives, congeneric sources of aggressive weedy genotypes cause major economic losses. Thus, the origins of weedy genotypes where no congeneric species occur raise questions regarding management effectiveness and evolutionary mechanisms responsible for weedy population success. Since eradication in the 1970s, California growers avoided weedy rice through continuous flood culture and zero-tolerance guidelines, preventing the import, presence, and movement of weedy seeds. In 2003, after decades of no reported presence in California, a weedy rice population was confirmed in dry-seeded fields. Our objectives were to identify the origins and establishment of this population and pinpoint possible phenotypes involved. We show that California weedy rice is derived from a different genetic source among a broad range of AA genome Oryzas and is most recently diverged from O. sativa temperate japonica cultivated in California. In contrast, other weedy rice ecotypes in North America (Southern US) originate from weedy genotypes from China near wild Oryza, and are derived through existing crop-wild relative crosses. Analyses of morphological data show that California weedy rice subgroups have phenotypes like medium-grain or gourmet cultivars, but have colored pericarp, seed shattering, and awns like wild relatives, suggesting that reversion to non-domestic or wild-like traits can occur following domestication, despite apparent fixation of domestication alleles. Additionally, these results indicate that preventive methods focused on incoming weed sources through contamination may miss burgeoning weedy genotypes that rapidly adapt, establish, and proliferate. Investigating the common and unique evolutionary mechanisms underlying global weed origins and subsequent interactions with crop relatives sheds light on how weeds evolve and addresses broader questions regarding the stability of selection during domestication and crop improvement.

Fig 1. Grains of the California weedy rice ecotype and the southern weedy rice ecotypes.

California weedy rice (top panel) with representative strawhull (bottom left) and blackhull (bottom right) weedy rice ecotypes. California weedy rice has straw to golden colored hulls and long straw colored awns. Strawhull weedy rice has short to no awns and straw colored hulls. Blackhull weedy rice has long, dark awns with a dark hull color. All three ecotypes have the characteristic red colored pericarp (wherein weedy ‘red’ rice derives its common name).

Fig 2. Population structure as inferred from the best fit number of groupings in InStruct.

Each color indicates genetic membership of the specified number of groups (K). The best fit model had K = 9, shown on the right. K = 8 and K = 10 are included for comparison. K = 4 was the lowest K value maintaining California weedy rice as a separate group. Ln likelihood scores for each K value are shown in parentheses, and likelihood scores for K = 1 through K = 22 are in Table E in S1 File.

Fig 3. Posterior probability densities for time since divergence from California weedy rice estimated by the isolation-with-migration model.

Time axis is on a log scale to include all comparisons (among California US cultivars, southern US strawhull (SH) and blackhull awned (BHA) weedy rice, and O. rufipogon from China and India).

Fig 4. Relative divergence times in generations (G) (per Table 5).

Ancestral gene flow/migration (NeM) from each likely source into California weedy rice are in parentheses. Labels are as follows: Oryza sativa ancestral global pool, including Chinese O. rufipogon, aromatic, barthii, meridionalis (NG1), O. sativa global sources including Indian O. rufipogon, nivara, sativa aus, other indica-like Oryza (NG2), temperate japonica (TJ), tropical japonica (TRJ), brownhull (BrH), blackhull (BHA), or strawhull awnless (SHA) Arkansas weedy rice (AWR), California temperate japonica (CTJ), SHA (strawhull awned) California weedy rice (CWR), and mixed genotypes (e.g. red pericarp African O. glaberrima) (MX).

Fig 5. California weedy and cultivated rice phenotypic trait decomposition.

5A. PCA of California weedy and crop (medium-grain and specialty) rice. Highly correlated traits and traits with no variation excluded. 5B. PCA biplot of the refined PCA analysis of California weedy, medium-grain crop, and specialty crop rice traits which are key players in separating the three clusters. 5C. Eigenvector values for a refined California weedy and crop rice dataset of 15 traits with highly correlated measurements and those with no variation excluded (from 5B). 5D. List of traits corresponding to the numeric labels (1–35) of PCA vectors in 5A-5C. The length and direction of each vector indicate the strength and type (positive or negative) of the correlation between specific traits and the PCs.