11 Scopus citations


Root system architecture (RSA) is a critical aspect of plant growth and competitive ability. Here we used two independently evolved strains of weedy rice, a de-domesticated form of rice, to study the evolution of weed-associated RSA traits and the extent to which they evolve through shared or different genetic mechanisms. We characterised 98 two-dimensional and three-dimensional RSA traits in 671 plants representing parents and descendants of two recombinant inbred line populations derived from two weed × crop crosses. A random forest machine learning model was used to assess the degree to which root traits can predict genotype and the most diagnostic traits for doing so. We used quantitative trait locus (QTL) mapping to compare genetic architecture between the weed strains. The two weeds were distinguishable from the crop in similar and predictable ways, suggesting independent evolution of a ‘weedy’ RSA phenotype. Notably, comparative QTL mapping revealed little evidence for shared underlying genetic mechanisms. Our findings suggest that despite the double bottlenecks of domestication and de-domestication, weedy rice nonetheless shows genetic flexibility in the repeated evolution of weedy RSA traits. Whereas the root growth of cultivated rice may facilitate interactions among neighbouring plants, the weedy rice phenotype may minimise below-ground contact as a competitive strategy.

Original languageEnglish
Pages (from-to)1031-1042
Number of pages12
JournalNew Phytologist
Issue number2
StatePublished - Jul 2019


  • comparative QTL mapping
  • convergent evolution
  • GiaRoots
  • parallel evolution
  • random forest
  • rice (Oryza sativa)
  • root system architecture (RSA)
  • weedy rice


Dive into the research topics of 'Convergent evolution of root system architecture in two independently evolved lineages of weedy rice'. Together they form a unique fingerprint.

Cite this