Genomic patterns of nucleotide diversity in divergent populations of U.S. weedy rice

Abstract

Background: Weedy rice (red rice), a conspecific weed of cultivated rice (Oryza sativa L.), is a significant problem throughout the world and an emerging threat in regions where it was previously absent. Despite belonging to the same species complex as domesticated rice and its wild relatives, the evolutionary origins of weedy rice remain unclear. We use genome-wide patterns of single nucleotide polymorphism (SNP) variation in a broad geographic sample of weedy, domesticated, and wild Oryza samples to infer the origin and demographic processes influencing U.S. weedy rice evolution.

Results: We find greater population structure than has been previously reported for U.S. weedy rice, and that the multiple, genetically divergent populations have separate origins. The two main U.S. weedy rice populations share genetic backgrounds with cultivated O. sativa varietal groups not grown commercially in the U.S., suggesting weed origins from domesticated ancestors. Hybridization between weedy groups and between weedy rice and local crops has also led to the evolution of distinct U.S. weedy rice populations. Demographic simulations indicate differences among the main weedy groups in the impact of bottlenecks on their establishment in the U.S., and in the timing of divergence from their cultivated relatives.

Conclusions: Unlike prior research, we did not find unambiguous evidence for U.S. weedy rice originating via hybridization between cultivated and wild Oryza species. Our results demonstrate the potential for weedy life-histories to evolve directly from within domesticated lineages. The diverse origins of U.S. weedy rice populations demonstrate the multiplicity of evolutionary forces that can influence the emergence of weeds from a single species complex.

Figure 1

Hull morphologies of U.S. weedy rice. A) straw hull no awns; this morphology is typical of the SH population (see text for explanation), B) black hull with awns; this morphology is typical of the BHA1 and BHA2 populations (see text for explanation), C) straw hull with awns; this morphology is also found in BHA1 and BHA2 populations, D) brown or striped hull with awns: this morphology is typical of the BRH population (see text for explanation)

Figure 2

Demographic model for ABC analysis. This model assumes that the initial weedy rice founder (effective population size η_p), split from an Oryza group (effective population size η_c) in Asia at time τ_s generations in the past. At time τ_f, lineages from the founder population, η_p, were introduced to the U.S. and experienced a bottleneck of size η_b before recovering to current population size η_r, instantaneously at time τ_g.

Figure 3

Population structure assessed by InStruct. Each individual accession is represented by a colored bar, partitioned to reflect an individual's relative proportion of genetic membership in a given cluster K. A) Results of analysis for U.S. weedy rice accessions (n = 58), showing progression of clustering as K increased to optimal K = 6. The order of accessions is the same as in Additional file 1 and in all plots. B) Population structure results for the entire Oryza panel (n = 209), showing clustering at optimal K = 9. Inset plot shows clustering of U.S. weedy rice. Samples are in the same order as Additional file 2, and O. rufipogon samples are sorted by country of origin. Weedy rice accession names color coded by cytotype: DDD light green, NDD orange, NND blue, DDN red, NNN dark green, Black - no data

Figure 4

Silent site θ_w within Oryza populations. Boxplot of θ_w calculated per kb, where boxes show interquartile ranges and median. The whiskers extend to 1.5 times the interquartile range and outliers are shown as open circles.

Figure 5

Results of the ABC analysis. Priors are plotted as a dashed black line A) time of population expansion: τ_{g SH} in red and τ_{g BHA1} in green, B) time of founding in U.S. τ_{f SH} in red and τ_{f BHA1} in green, C) η_SH/η_indica in red, η_BHA1/η_aus in green. Contour plots of the approximate likelihood for current weedy population size as a function of bottleneck intensity, which is the percent decline in population size during the bottleneck, D) SH, and E) BHA1.