Genome-wide patterns of nucleotide polymorphism in domesticated rice

Abstract

Domesticated Asian rice (Oryza sativa) is one of the oldest domesticated crop species in the world, having fed more people than any other plant in human history. We report the patterns of DNA sequence variation in rice and its wild ancestor, O. rufipogon, across 111 randomly chosen gene fragments, and use these to infer the evolutionary dynamics that led to the origins of rice. There is a genome-wide excess of high-frequency derived single nucleotide polymorphisms (SNPs) in O. sativa varieties, a pattern that has not been reported for other crop species. We developed several alternative models to explain contemporary patterns of polymorphisms in rice, including a (i) selectively neutral population bottleneck model, (ii) bottleneck plus migration model, (iii) multiple selective sweeps model, and (iv) bottleneck plus selective sweeps model. We find that a simple bottleneck model, which has been the dominant demographic model for domesticated species, cannot explain the derived nucleotide polymorphism site frequency spectrum in rice. Instead, a bottleneck model that incorporates selective sweeps, or a more complex demographic model that includes subdivision and gene flow, are more plausible explanations for patterns of variation in domesticated rice varieties. If selective sweeps are indeed the explanation for the observed nucleotide data of domesticated rice, it suggests that strong selection can leave its imprint on genome-wide polymorphism patterns, contrary to expectations that selection results only in a local signature of variation.

Figure 1. Estimated Population Structure for 97 Accessions of O. sativa and O. rufipogon from 111 STS Loci

Vertical bars along the horizontal axis represent each Oryza accession; for all accessions, the proportion of ancestry under K = 7 clusters that can be attributed to each cluster is given by the length of each colored segment in a bar.

Figure 2. The Observed Marginal Derived Site-Frequency Spectra of Noncoding and Synonymous SNPs for Two Population Pairs: indica and O. rufipogon and tropical japonica and O. rufipogon

To accommodate SNPs with missing data, all spectra are plotted as the expected site frequency spectrum in a subsample of the data of size n = 16.

Figure 3. Contours of Composite Profile Log-Likelihood Surface under the Bottleneck and Migration (i.e., “Complex Demography”) Model for Three Key Demographic Parameters

Parameters include bottleneck severity, migration rate among demes (4Nm), and τ₁ (time back until start of domestication scaled in units of 2N_rufi). The maximum composite-likelihood estimate of the parameters is denoted by a red filled circle.

Figure 4. Observed and Expected Derived Site-Frequency Spectra under Various Models

The observed derived site-frequency spectrum for (A) indica and (B) tropical japonica, along with the expected site-frequency spectrum under the simple bottleneck, bottleneck plus migration demography, and bottleneck plus sweeps models. (C) Observed site-frequency spectrum for O. rufipogon and expected frequencies using a standard neutral model and a bottleneck plus migration model.

Figure 5. Composite Likelihood Surfaces in indica and tropical japonica under Models Incorporating Selection

A density plot of the marginal composite log-likelihood surface of the parameters α and κ, with the bottleneck severity υ fixed to its estimate, under the bottleneck plus sweeps model for (A) indica and (B) tropical japonica. The composite log-likelihood surface of the parameters α and κ under the pure selection model for (C) indica and (D) tropical japonica. The composite log-likelihood is represented as a deviation from the maximum log-likelihood, with lighter values representing higher composite likelihoods. Numbers above (A) and (B) indicate the total number of sweeps in the rice genome corresponding to each value of κ, and numbers to the right of (B) and (D) represent the selection coefficient, s, corresponding to each value of α, substituting an effective recombination rate of r = 10⁻¹² and ln(2N) = 10 into the expression: α ≈ rs⁻¹ ln(2N), then solving for s.