Italian weedy rice-A case of de-domestication?

Abstract

Weedy rice is a representative of the extensive group of feral weeds that derive from crops, but has returned to the lifestyle of a wild species. These weeds develop either from a hybridization of crops with wild relatives (exoferality), or by mutation of crops to weedy forms (endoferality). Due to the close relation of weed and crop, the methods for weed-targeted containment are limited to date. A deeper understanding of the development of such weeds might help to design more efficient and sustainable approaches for weed management. Weedy rice poses a serious threat to rice yields worldwide. It is widely accepted that weedy rice has originated independently in different regions all over the world. However, details of its evolution have remained elusive. In the current study, we investigated the history of weedy rice in northern Italy, the most important rice-growing area in Europe. Our approach was to analyze genes related to weedy traits (SD1, sh4, Rc) in weedy rice accessions compared to cultivars, and to integrate these results with phenotypic and physiological data, as well as historical information about rice farming in Italy. We arrive at a working model for the timeline of evolution of weedy rice in Italy indicating that both exoferality and endoferality acted as forces driving the development of the diverse weedy rice populations found in the region today. Models of weed evolution can help to predict the direction which weed development might take and to develop new, sustainable methods to control feral weeds.

Keywords: domestication; endoferality; exoferality; single nucleotide polymorphisms; weedy rice (Oryza sativa cf. spontanea).

FIGURE 1

Structural and functional features of the locus SEMIDWARF 1 (SD1) for Italian and Indian accessions of rice. (a) Genomic structure of the SD1 locus in O. rufipogon (strain W1843); O. sativa ssp. japonica cv. “Nipponbare” (reference genome for japonica); O. sativa ssp. indica cv. “R498” (reference genome for indica), landraces from India, and Italian cultivated landraces and weedy rice with different alleles found in haplotypes H₁–H₅ across Italian cultivated landraces and weedy accessions. H₁′ is a variant of H₁ bearing a rufipogon signature in exon 3. Numbers give positions in basepairs from the start codon. Signatures specific for O. rufipogon (Oruf₁‐Oruf₄) are indicated by red arrows, the red triangle Oruf₄ indicates the position of a characteristic 11‐bp deletion downstream of the stop codon. In the japonica allele, positions of the mutations in the sativa SD1 alleles in Jikkoku (JK), Calrose 76 (CLR76), and Reimei, as well as the deletion in Dee‐Geo‐Woo‐Gen (DGWG, orange box) are indicated (Sasaki et al., 2002). (b) Haplotype map showing the single nucleotide polymorphisms (SNPs) and the resulting amino‐acid residues (aa) in the Italian haplotypes H₁‐H₅, compared to O. rufipogon (ruf), O. sativa ssp. japonica (jap), and O. sativa ssp. indica (ind). Red rectangles highlight differences between the three alleles, note that ind is equal to jap in exon 1, but equal to ruf in exon 3, the black rectangle highlights a SNP present in all Italian haplotypes (both cultivated or weedy). Colored arrows refer to the most parsimonious model (shown in c) explaining these haplotypes. (c). The respective amino‐acid substitutions and the position of the respective residue is indicated. (d) Culm length of cultivated (green bar) and weedy (pink bar) accessions collected in Italy. The difference is significant at p = .009. Data represent mean values and standard errors from at least 100 individuals

FIGURE 2

Structural and functional features of the locus SHA1/SH4 for Italian accessions of rice in comparison to the wild species O. rufipogon. (a) Haplotype map showing the single nucleotide polymorphisms (SNPs) and the resulting amino‐acid residues (aa) in the Italian haplotype H₁, compared to O. rufipogon (ruf) and O. sativa ssp. japonica (sat). The sat haplotype was seen in all 14 tested cultivated accessions, but only in four of the 24 tested weedy accessions. The haplotype H₁ was found only in the weedy, but in none of the tested cultivated accessions. (b) Classification of the abscission zone into three types depending on the depth of the cavity and the appearance of the surface. In the rough type, the caryopses remain on the ear, while in the soft type, they are readily scattered. (c) Frequency distribution of the abscission‐zone types in cultivated (green) versus weedy (red) accessions from Italy. The histogram is based on a sampled of 18 weedy and 15 cultivated accessions; the classification of each accession is based on microscopic analysis of five individual caryopses per accession

FIGURE 3

Structural and functional features of the locus Rc for Italian accessions of rice in comparison to the wild species O. rufipogon. (a) Haplotype map showing the 14‐bp deletion in exon 7 for O. sativa ssp. japonica (sat) as compared to the allele from O. rufipogon (ruf). The sat allele was exclusively seen in the cultivated accessions. In the weedy accessions, two haplotypes were found. Haplotype H₁ was identical to the rufipogon allele and was found in 25% of the tested weedy accessions, but in none of the tested cultivated accessions. Haplotype H₂ showed the 14‐bp deletion characteristic of the sat allele, but in addition carried a 1‐bp deletion 46 bp upstream. Both deletions combined will restore the reading frame downstream of the 14‐bp deletion and thus likely deliver a largely functional product. This allele was seen in 69.6% of the tested weedy accessions, but in none of the cultivated Italian accessions. (b) Quantification of proanthocyanidin content in the two weedy haplotypes compared to the Italian O. sativa ssp. japonica cultivar “Arborio” (sat). Haplotype 2 shows a slight decrease, which is, however, not significant. Values represent mean and standard error for 1 g of seed material per accession. Haplotype 1 was represented by 16, haplotype 2 by 4 accessions

FIGURE 4

Incidence of weedy alleles in domestication genes of cultivated and weedy accessions of Italian rice. (a) The (more active) rufipogon allele of sd1 is exclusively found in the cultivated accession, while all weedy accessions harbor the japonica allele or derivatives of it. (b) All tested cultigens harbor the nonfunctional allele sh4 from O. sativa japonica, the majority of weedy accessions the allele H1. (c) All tested cultigens harbor the japonica allele of Rc linked with the lack of pericarp coloration, while all weedy accessions either harbor the original (functional) rufipogon allele or the endoferal allele H2, where reading frame is restored by a second‐site deletion. (d) Coupling of weedy alleles for Rc and sh4 in cultivated and weedy accessions of Italian rice The frequency of the japonica and the feral (weedy) allele H₁ (see Figure 2) in dependence on the three alleles identified for the Rc locus. The japonica type Rc locus is tightly associated with a japonica type sh4 allele, while the endoferal Rc allele (H₂) is tightly associated with a feral sh4 allele. For a rufipogon Rc allele, both japonica and feral alleles of sh4 can occur

FIGURE 5

Timeline model for the evolution of weedy rice in Northern Italy. The data generated by this study suggest that contaminated seed stocks imported from Asia introduced rufipogon alleles for domestication traits (most prominently, for pericarp pigmentation). This step represents an exoferality event. With the establishment of rice breeding programs in Italy around 1,800 until today, several de‐domestication events of cultigens produced novel weedy traits by endoferality. These spread to a certain extent, partially in the heterozygous state (e.g., the sd1 locus). With the switch from transplanting to direct sowing in the 1960s, the selective pressure for domestication alleles was drastically reduced leading to an increase in infestation rates