Genomic revolution of US weedy rice in response to 21st century agricultural technologies

Understanding the genomic basis of adaptation is crucial in evolutionary biology. Crop domestication serves as a model for studying adaptive responses, and recent research on evolving crop species' genomes provides valuable insights. The evolution of agricultural weeds, particularly those closely related to crops, offers additional understanding due to the potential for genetic exchange with crop cultivars. Weedy rice (*Oryza* spp.), a de-domesticated form of cultivated rice (*O. sativa*), is a well-studied example. It's highly adapted to rice fields, where it competes aggressively with cultivated rice, causing significant yield losses and economic damage, estimated at over $45 million annually in the US and hundreds of millions worldwide. Control is challenging due to its phenotypic and genetic similarity to cultivated rice. Historically, two distinct morphotypes predominated in the southern US: strawhull (SH) and blackhull-awned (BHA). Outcrossing rates between these and US cultivars (tropical japonica) were historically low. In 2002, the introduction of non-transgenic herbicide-resistant (HR) rice cultivars (Clearfield™ rice) resistant to imidazolinone herbicides (due to ALS gene mutations) and hybrid rice technology (increasing crop volunteer rates) created a new selective environment. This spurred crop-weed hybridization, increasing the potential for herbicide resistance genes to transfer to weedy rice. As early as 2004, HR weedy rice was reported. By 2010, a significant portion of weedy rice carried HR alleles. This study investigated how the genomic composition of southern US weedy rice changed following these technological introductions, aiming to understand the genomic differences between contemporary and historical weeds, the direction of genome-wide selection post-hybridization, and the patterns of selection driving weed or crop-specific alleles to high frequency.

Previous research has extensively characterized weedy rice, focusing on its weed-adaptive features like seed dispersal and dormancy. Studies have explored its origin and population genetics in various regions. Londo and Schaal (2007) investigated the origins and population genetics of weedy red rice in the USA. Li et al. (2017) identified signatures of adaptation in the weedy rice genome. Other studies have examined the impact of weedy rice on rice yield and quality, highlighting the significant economic losses caused by this weed. The introduction of herbicide-resistant and hybrid rice cultivars has been studied in terms of their impact on weed evolution, noting the increased potential for gene flow between crop and weed. Previous work has shown the emergence of herbicide-resistant weedy rice populations following the introduction of HR cultivars, documenting the rapid spread of resistance alleles. However, this study's comprehensive genomic approach offers a more detailed picture of the evolutionary dynamics of weedy rice in the face of 21st-century agricultural technologies.

Seeds from 48 maternal weedy rice samples across five Arkansas rice fields were collected in 2018. Whole-genome sequences (>40x average coverage) were generated using leaf tissue from seedlings grown from one seed per plant. These sequences were analyzed alongside 98 previously published weedy, cultivated, and wild rice samples, resulting in a dataset of 146 samples and ~19.34 million SNPs. Wild rice was excluded from further analysis. Principal component analysis (PCA) and ADMIXTURE analysis assessed the genetic composition of contemporary weeds relative to historical strains. Loter software calculated local ancestry throughout the genome to detect bias toward crop or weed ancestry post-hybridization. F_ST was calculated between contemporary weeds and their inferred ancestors using a sliding window analysis to identify regions of selective introgression. A median-joining network tree of ALS haplotypes was constructed from manually phased consensus sequences. Herbicide resistance was phenotyped using imazethapyr application.

PCA and ADMIXTURE analysis revealed that most contemporary US weedy rice accessions (except four) are hybrid derivatives of crop-weed hybridization events. Most of these are descended from BHA × cultivated rice hybrids, with fewer derived from SH × cultivated rice. A subset of BHA-like weeds formed a genetically homogeneous group ('beta'), possibly representing an early post-hybridization population. Contemporary weeds collectively exhibited high heterozygosity, consistent with recent hybrid ancestry. However, local ancestry analysis showed a genome-wide bias toward the historical weed genome (74.1% and 69.2% for BHA-like and SH-like groups, respectively). F_ST analysis revealed strong selection for crop alleles near the ALS gene (conferring herbicide resistance), particularly in BHA-like weeds. In contrast, a weed-like *Rc* region (associated with seed dormancy) was found primarily in BHA-like weeds. ALS haplotype network analysis showed that most contemporary weeds carried the S653N resistance mutation from modern HR cultivars. However, two accessions showed the older G654E mutation, possibly representing very early hybridization events. Interestingly, two samples showed S653N resistance without evidence of hybrid origin, suggesting convergent evolution. Herbicide resistance phenotyping confirmed that most contemporary weeds showed high levels of resistance to imazethapyr.

The predominance of crop-weed hybrids among contemporary weeds is surprising, as low hybrid fitness was expected. One possibility is that while F1 hybrids have low fitness, subsequent generations, after selfing, may exhibit higher fitness as alleles segregate into favorable combinations. The consistent genome-wide bias toward weedy ancestry, irrespective of initial hybrid composition, suggests a selective maintenance of weedy genome components or a selective purge of crop alleles. The high frequency of herbicide resistance highlights the effectiveness of gene flow from HR rice into weedy rice populations and the rapid adaptation of weeds to this selection pressure. This underscores the dangers of relying on single control methods for agricultural pests.

This study reveals a genomic revolution in US weedy rice, driven by the introduction of HR and hybrid rice technologies. Most contemporary weeds are crop-weed hybrids exhibiting a genome-wide shift back towards their weedy ancestor. The widespread herbicide resistance highlights the need for integrated pest management strategies. Future research should focus on the mechanisms driving the selective maintenance of weedy alleles, the potential for further evolution of herbicide resistance, and the development of more effective weed control strategies that consider the genomic basis of weed adaptation.

The study focuses on a specific geographic region and time point. The use of greenhouse-grown plants for DNA extraction might not perfectly reflect field genotypes, especially for segregating variants. The analysis relies on short-read whole-genome sequencing, potentially missing some genomic variations. Finally, the inferences made regarding the number of generations since hybridization are based on certain assumptions about mating systems post-hybridization.