Selective and comparative genome architecture of Asian cultivated rice (Oryza sativa L.) attributed to domestication and modern breeding

Abstract

Introduction: Rice, Oryza sativa L. (Os), is one of the oldest domesticated cereals that has also gone through extensive improvement in modern breeding.

Objectives: How rice was domesticated and impacted by modern breeding.

Methods: We performed comprehensive analyses of genomic sequences of 504 accessions of Os and 456 accessions of O. rufipogon/O. nivara (Or).

Results: The natural selection on Or before domestication and the natural and artificial selection during domestication together shaped the well-differentiated genomes of two subspecies, geng(j) (japonica) and xian(i) (indica), while breeding has made apparent genomic imprints between landrace and modern varieties of each subspecies, and also between primary modern and advanced modern varieties of xian(i). Selection during domestication and breeding left genome-wide selective signals covering ∼ 22.8 % and ∼ 8.6 % of the Os genome, significantly reduced within-population genomic diversity by ∼ 22 % in xian(i) and ∼ 53 % in geng(j) plus more pronounced subspecific differentiation. Only ∼ 10 % reduction in the total genomic diversity was observed between the Os and Or populations, indicating domestication did not suffer severe genetic bottleneck.

Conclusion: Our results revealed clear differentiation of the Or accessions into three large populations, two of which correspond to the well-differentiated Os subspecies, geng(j) and xian(i). Improved productivity and common changes in the same suit of adaptive traits in xian(i) and geng(j) during domestication and breeding resulted apparently from compensatory and convergent selections for different genes/alleles acting in the common KEGG terms and/or same gene families, and thus maintaining or even increasing the within population diversity and subspecific differentiation of Os, while more genes/alleles of novel function were selected during domestication than modern breeding. Our results supported the multiple independent domestication of Os in Asia and suggest the more efficient utilization of the rich diversity within Os by exploiting inter-subspecific and among population diversity in future rice improvement.

Keywords: Domestication; Modern breeding; Novel functional variation selection; Oryza sativa; Standing functional variation selection.

Graphical abstract

Fig. 1

Phylogeny, differentiation and diversity of/among O. sativa (Os) and O. rufipogon (Or) populations. (a) The phylogenetic tree of 960 rice accessions, including 504O. sativa accessions and 456O. rufipogon/O. nivara accessions, in which the colored lines indicate different types of Os accessions with red = geng(j), green = xian(i), blue = Aus, cyan = Aro, light green = Or-XL, dark red = Or-GL, pink = Or-Int, and grey = admixtures; Or-I, Or-II and Or-III in the panel representing the three major Or populations reported by Huang et al. ; (b) the geographic distribution of the three Or populations; (c) the genetic diversity estimates, θπ, of the two Os subspecific landraces (X-LAN and G-LAN) and three Or populations (Or-XL, Or-GL and Or-Int) inferred by the phylogenetic tree; (d) a 3-dimensional presentation of the differentiation among different Or and Os populations based on the principal component analysis of SNP variation, in which the ×, y and z coordinates were indicated by PC1, PC2 and PC3, the large points represent centers of the Or and Os populations; the distance between points along line Or-GL:Or-XL (i.e. the line through points Or-GL and Or-XL, similar for others) accounts for the natural selection consequences (NSC) of two corresponding or projected populations; the dot line P1:P2 is the common vertical line between line Or-GL:Or-XL and line G-LAN:X-LAN; P3, P4, P7, P8 and P9 are the projection points of populations G-LAN, X-LAN, G-MV, X-PMV and X-AMV to the plane consisting of lines Or-GL:Or-XL and P1:P2, thus the distance between each population and its projection point accounts for the subspecies-specific artificial selection consequence (ssASC); P5, P6, P10, P11 and P12 are the projection points of P3, P4, P7, P8 and P9 to line Or-GL:Or-XL, thus the distance between each point and its projection point accounts for the subspecies-common artificial selection consequence (scASC); (e) the genomic proportions of population-specific genomic blocks of various Os populations in different Or populations from the same geographic regions of 2°×2° of latitude by longitude, where the small pies are distributions for population-specific genomic blocks. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2

Pan-genome analysis between Oryza sativa (Os) and O. rufipogon (Or). (a) The proportions of Os genes detectable in different Or populations; (b) the distribution of genes of two subspecies, geng(j) (G) and xian(i) (X), in Or-GL and Or-XL, where a, c or d in the G:X column represent absence, core or distributed genes in the corresponding subspecies, y or n in the Or(GL:XL) column represent presence or absence in the corresponding Or populations; (c)-(f) the bi-clustering analysis for genes both core in two subspecies (24,180), those core in G and distributed in X (5,114), those distributed in G and core in X (2,770), and those both distributed in two subspecies (7,998) according to their presence (red) or absence (blue) in different Or populations. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

The population structure and important agronomic phenotypes of O. sativa (Os). (a) The phylogenetic tree of 504 Os accessions, in which the marks are the same as Fig. 1a, and the colored lines represent different populations, which were inferred by population structure analysis of FRAPPE with k = 10 and labeled by symbols around the tree; (b) the step-by-step differentiation of different Os populations with increased k values; (c) and (d) the mean values and variation of plant height and grain weight per plant of different populations of xian(i) and geng(j) landraces and modern varieties, in which ^** represents significant differences in mean trait values between the modern varieties and landraces.

Fig. 4

Genome-wide selected blocks related to domestication and breeding, and cloned genes related to yield and domestication. The colored blocks in each layer from inner to outer represent respectively the domestication selected blocks detected from Oryza sativa (Os) relative to Oryza rufipogon (Or) (D1), Os landraces (LAN) relative to Or (D2), geng(j) LAN relative to geng-like Or (D3), xian(i) LAN relative to xian-like Or (D4), and the breeding selected blocks detected from geng(j) modern varieties (MV) relative to geng(j) LAN (B-G1), temperate geng(j) MV relative to temperate geng(j) LAN (B-G2), tropical geng(j) MV relative to tropical geng(j) LAN (B-G3), xian(i) primary modern varieties (PMV) relative to xian(i) LAN (B-X1), xian(i) advanced modern varieties (AMV) relative to xian(i) LAN (B-X2) and xian(i) AMV relative to PMV (B-X3); the outmost symbols are the cloned genes related to yield and domestication in the selected blocks.

Fig. 5

The scenario of agronomy-important cloned selected genes (SGs). (a) The proportion of all SGs in annotated genes and the proportion of agronomy-important cloned SGs relative to all cloned genes among different domestication or breeding events; (b) the portion of SGs subjecting to coding region selection relative to all SGs; (c) the portion of SGs subjecting to novel function selection (NFS) relative to SGs subjecting to coding region selection; (d) the regulation network of growth duration genes subjected to domestication and breeding, here, LD: long day condition, SD: short day condition, +: interaction, +p: phosphorylation, red frame: selected during geng(j) domestication, green solid or dot frame: selected during xian(i) domestication or breeding, orange filled: NFS, yellow filled: standing function selection (SFS), blank: allele-frequency-dependent selection, waved line: promoter; and (e) the cumulative percent of haplotypes with significant difference in growth duration, here, labels in the bracket indicate the types of SGs, such as B-G-Tr being the SGs during event of breeding for tropical geng(j) varieties. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6

Enrichment of pathways and gene families in different selected blocks. (a) The proportion of common terms with common genes or just common terms of pathway within two subspecies, xian(i) and geng(j) or two events (domestication and breeding), in which “To the minimum” or “To the maximum” means the proportion to the minimum or maximum number of enriched terms in the two compared types of selected blocks; and (b) the proportion of common gene families with common member genes or just common gene families within two subspecies xian(i) and geng(j) or two events (domestication and breeding).