Cultivated and weedy rice interactions and the domestication process



Amy Lawton-Rauh, Fax: 864 656 6879; E-mail:


Examining the targets of selection in crop species and their wild and weedy relatives sheds light on the evolutionary processes underlying differentiation of cultivars from progenitor lineages. On one hand, human-mediated directional selection in crops favours traits associated with the streamlining of controllable and predictable monoculture practices alongside selection for desired trait values. On the other hand, natural selection in wild and especially weedy relatives presumably favours trait values that increase the probability of escaping eradication. Gene flow between crops and wild species may also counter human-mediated selection, promoting the evolution and persistence of weedy forms. In this issue, two studies from a group of collaborators examine diversity and divergence patterns of genes underlying two traits associated with red rice (Oryza sp.), the conspecific relative of cultivated rice (Oryza sativa) that is a non-native weed (see Fig. 1). In the first study by Gross et al. (2010), genetic variation in the major gene underlying the hallmark red pigmentation characterizing most weedy rice (Rc) is found to have a pattern consistent with non-reversion from U.S. cultivated rice (i.e. the cultivar did not ‘go feral’). This suggests that U.S. weedy rice is not an escaped lineage derived from U.S. cultivated rice populations; weedy rice likely differentiated prior to the selective sweep occurred in this gene within cultivated rice populations. Using the major seed shattering locus sh4 gene and the neighbouring genomic region, Thurber et al. (2010) track the molecular evolutionary history of the high shattering phenotype, a trait contributing dramatically to the success of crop selection in cultivated rice as well as the persistence and expansion of weedy red rice. In this study, the shared fixation of a sh4 mutation in both cultivated rice and weedy rice indicates that weedy rice arose subsequent to the strong selective sweep leading to significant reduction in seed shattering in cultivated rice.

Figure 1.

 A weedy, brown hulled red rice individual with long awns surrounded by a field of cultivated rice (photo by A. Lawton-Rauh).

The opposing selection between favouring traits that improve cultivation and commodity value versus traits increasing survival from eradication practices in weedy conspecific species started long ago. Despite the obvious importance in the context of yield and agroecosystems management, the genetic mechanisms underlying the continuation of this war remain poorly understood. One major reason for this is that populations of wild and weedy close relatives can be independently derived from multiple, unknown source populations. Given the global nature of crop trade, seed sources, and large-scale management practices we implement to ensure high yield and predictable productivity, the potential source populations for conspecifics essentially spans rice agroecosystems around the globe. The two studies by Gross et al. (2010) and Thurber et al. (2010) address the potential origin of weedy red rice in the non-native U.S. range of this taxon and potential interactions between cultivated rice and its wild and weedy close relatives. These studies examine critically two loci that have a major impact on two critical traits associated with important differences that improve trait values in the rice crop when selected in one direction (e.g. favouring predictable, consistent seed shattering) and increase the chance of escape from removal in the weedy species in the other direction (e.g. favouring high variance and rapid shattering to escape from management practices). To examine the origins of and interactions among cultivated rice and conspecific wild and weedy relatives, these two studies sequentially address the single, independent introduction of versus continued derived source of U.S. weedy rice populations and then examine whether a key trait involved in weediness, seed shattering, evolved independently in U.S. weedy rice. Together, these results present a fascinating story regarding how traits characterizing this conspecific weed have been shaped by genetic interactions with globally diverse cultivated rice lineages and mediated by selection at genes playing pivotal roles in traits that are essential for survival from eradication practices.

Gross et al. (2010) addresses whether U.S. weedy rice is derived from either undomesticated Oryza populations related to aus and/or indica lineages that also gave rise to Asian rice domesticates or from already existing wild and/or weedy rice populations initially descended from Asian domesticates that later spread to North America. To test these alternative hypotheses, this study takes advantage of a sharply contrasting phenotype distinguishing cultivated rice from weedy red rice: pericarp pigment (the outer covering of the seed). In cultivated rice, it is desirable to have uniform size and colour of grains. Although rice with purple or red grains is cultivated in other world regions, practically all U.S. cultivated rice has purely white pericarps whereas U.S. weedy rice has red pericarp. Indeed, the zero tolerance policy for red rice contamination in the U.S. and in other countries requires cultivated rice varieties to harbor no alleles leading to pigmented pericarps. The authors further take advantage of strong selection against alleles resulting in pigmented pericarps in cultivars by examining DNA sequence variation in the 6.4 kb Rc locus plus 1.6 kb upstream and 0.8 kb downstream, respectively, of the start and stop codons of this proanthocyanidin-encoding gene. Additionally, this study also sequenced ∼500 bp fragments at points spanning 2.1 Mb upstream and 2.0 Mb downstream from the Rc gene. The size of the sequenced region is within the <1 Mb size range that a previous report showed as the range of the selective sweep in this gene (Sweeney et al. 2007). Most modern domesticated rice varieties lack pericarp proanthocyanidin pigments. Thus molecular evolutionary analyses comparing the loss-of-function alleles in cultivated rice can be used to ascertain the genetic distance and patterns of relatedness among weedy red rice conspecifics and to test whether particular U.S. weedy red rice populations originate directly from Asian cultivated lineages, U.S. cultivated lineages, or from other Asian weedy/wild populations. The authors of this study sequenced this broad region from a large sample of 156 weedy, domesticated, and wild Oryza sp. Phylogenetic analyses of these sequences clearly indicate cultivated rice varieties with white pericarps have two allele categories (rc and Rc-s) that are uniquely derived rather than from a shared common ancestral allele and that U.S. weedy red rice alleles do not have the patterns expected if reversion from crop alleles occurred. Thus, taken together, their results do not support a cultivars-gone-feral model for the origin of weedy red rice in the U.S. As the authors point out, it is still unclear if another source may be more likely: the highly diverse, widely-distributed outcrossing species Oryza rufipogon. It will be exciting to see what role, if any, that this species could have played (possibly continues to play?) in the origin of weedy and conspecific lineages in the U.S. and around the world. The anticipated expanded collection of O. rufipogon by this research group will shed light on this potential alternative explanation.

In a complementary study by the same group, Thurber et al. (2010) examine allelic diversity and divergence in the most significant gene controlling seed shattering, sh4. Seed shattering is of intense focus because low shattering leads to predictable seed release during harvest in cultivated rice and high shattering is an adaptive mechanism resulting in the persistence and expansion of weedy rice. Highly shattering genotypes release seeds with minimal contact, dispersing seeds immediately into the soil or equipment that further spreads the highly shattering genotypes across a field, onto roads, and into other fields. Without highly shattering genotypes, it is likely that weedy red rice would not be quite so widespread and less seed could return to the soil seed bank, which makes it possible to remove from cultivated rice fields. To trace the origin of highly shattering genotypes in U.S. weedy red rice, this study assessed the patterns of polymorphisms in the sh4 gene in a sample of 58 U.S. weedy rice accessions representing the rice belt distribution plus 87 diverse Oryza sp. representing potential sources of weedy red rice alleles in the accessions sampled. Sequence diversity and genealogical tracing of the sh4 gene sequence plus SNP haplotypes representing a larger, 6.2 Mb region spanning upstream and downstream from the sh4 gene indicates that the non-shattering sh4 mutation is fixed in cultivated rice and U.S. weedy rice populations sampled. Examination of the entire region revealed no haplotype sharing between the most likely U.S. cultivar Oryza sativa tropical japonica from which weedy rice was originally suspected to be derived and weedy rice, consistent with previous studies indicating that U.S. weedy rice is generally most closely related to the indica subgroup (Londo & Schaal 2007). Interestingly, these results are also consistent with observations of RiceTec hybrid rice, which is derived from an indica parent and has a high grain shattering phenotype. The uniqueness of haplotypes in weedy rice suggests that U.S. weedy rice in this sample did not capture sh4 alleles via gene exchange with U.S. cultivars grown in the region. A shared cultivar allele in combination with a highly shattering phenotype revealed in their analyses also suggests that U.S. weedy rice populations maintain an independently derived genotype or genetic mechanism that increases seed shattering. Furthermore, linkage disequilibrium mapping of the gene region using Extended Haplotype Homozygosity analysis (Sabeti et al. 2002) showed a breakdown in homozygosity in O. rufipogon (the wild ancestor of cultivated rice, Khush 1997). This pattern sharply contrasted the extensive homozygosity uncovered in cultivated rice groups. This difference in homozygosity within the region containing the sh4 gene suggests a contrasting history of genetic variation in the wild relative (O. rufipogon) and cultivated O. sativa, possibly indicating differential selection for alternative phenotypes for seed shattering since these lineages diverged as well as a potential allele fixation in cultivated rice. To connect patterns of genetic variation and divergence with the phenotype strongly shaped by the sh4 gene, this study also examined breaking tensile strength (BTS) for the release of a seed from a pedicel at the abscission layer as an estimate of potential shattering level for 90 inbred Oryza accessions. They found that cultivated Asian rice varieties have a broader range of shattering than expected, in contrast to the expected pattern of low shattering variation characterizing a fixation of alleles decreasing the potential for yield loss due to early seed release. In other cultivated groups, this expected decreased variation and lowered shattering was supported. Interestingly, O. rufipogon and O. nivara had a BTS patterns similar to that of almost all other wild Asian rice, which was consistent with high shattering and the wild origins of cultivated rice.

The genetic relatedness between cultivated species and conspecific wild species suggests a potential for highly adaptive lineages to emerge as weedy species. The strong selection practices directed towards maintaining monocultures alongside desirable agronomic and commodity trait values in cultivated species can lead to indirect selection for phenotypes that increase the persistence and spread of conspecifics. Conspecifics are similar enough to the cultivated crop to thrive in the agroecosystem conditions optimized for intensive cultivation yet either derive new or capitalize on standing phenotypic variation such as variable levels of seed dormancy and longevity that effectively keeps them feral enough to escape efforts of containment or elimination. A further potential outcome is the introgression of alleles originally maintained in cultivars for the purpose of eradicating weedy conspecifics. An example in cultivated rice and weedy red rice can be seen in F1 hybrids between herbicide-resistant cultivated rice and weedy red rice shown in Fig. 2. The two studies published in this issue shed some light on the patterns underlying a specific example in cultivated rice by examining two defined genes clearly involved in two major phenotypes characterizing marked differences between cultivated rice versus wild and weedy conspecific populations. These studies therefore shed light on both the origin of rice and how the domestication process is a continuous rather than fixed process. Furthermore, this pattern of interaction at different phases during the domestication of rice and how it continues to impact the population dynamics between cultivated rice and weedy rice suggests that we have much to learn regarding the targets of selection during domestication and the opposing evolutionary processes shaping conspecific weedy species in agroecosystem. Future work elaborating on the genetic basis for the persistence of ‘feral’ populations alongside fields of specialized monocultures will undoubtedly shed significant light onto these processes which are not only relevant to agricultural genetics, but also contribute to our fundamental understanding of speciation dynamics.

Figure 2.

 Spontaneous cross between weedy red rice and herbicide-resistant Clearfield rice (photo by N. R. Burgos).