Seed dormancy shapes gene drive dynamics in plants

nature plants Article https://doi.org/10.1038/s41477-026-02256-1 Seed dormancy shapes gene drive dynamics in plants Received: 24 April 2025 Accepted: 24 February 2026 Published online: xx xx xxxx Check for updates Isabel K. Kim 1,5, Leqi Tian 1,2,5, Ryan Chaffee 3, Benjamin C. Haller Jackson Champer 4,6, Philipp W. Messer 1,6 & Jaehee Kim 1,6 , 1 Gene drives offer revolutionary potential for the management of problematic plant populations, such as invasive weeds and herbicide-resistant species, by rapidly spreading desired genetic alterations. Two recent studies have provided experimental demonstrations of engineered CRISPR gene drive systems in plants (CAIN and ClvR). However, the successful application of such systems in the field will critically depend on an accurate understanding of plant-specific life-history traits, especially seed dormancy, a ubiquitous yet frequently overlooked eco-evolutionary force. In this study, we develop a comprehensive modelling framework for gene drives in plant populations that incorporates a persistent soil seed bank. We show how the presence of a seed bank can substantially slow gene drive spread but also reduce the genetic load required to achieve population elimination. Furthermore, we show that seed banks substantially increase the required introduction frequency of threshold-dependent gene drives, which could prevent establishment in some cases, yet also provide an intrinsic biosafety mechanism for confining a highly efficient drive to a target population. Our study highlights the need to incorporate seed-bank dynamics into gene drive strategies to ensure realistic predictions and successful field applications. Weeds are among the major biological threats to global agriculture, causing annual yield losses in major crops that result in substantial economic damage and threaten food security1–5. Globally, weeds are responsible for around 10% of crop losses1, and in the USA alone, they account for more than US$26 billion a year in control expenses and lost crop yield6. The intensive and widespread use of chemical herbicides has historically provided the primary means of weed control; however, the rapid evolution of herbicide resistance poses an escalating challenge7–11. There is thus an urgent need to develop novel, effective and evolutionarily robust weed management tools. One promising avenue for next-generation weed management is the use of gene drives—selfish genetic elements that can quickly propagate through populations, even if they confer a fitness cost to individual organisms12–14. Gene drives could be used for either population suppression or population modification of problematic weeds15–20. A suppression drive, for example, could spread deleterious traits (such as sterility or non-viability) to eradicate an invasive weed population, whereas a modification drive could reverse herbicide resistance by propagating susceptibility alleles. This raises the possibility that engineered gene drives could one day provide lasting, inexpensive solutions for controlling weeds that are otherwise difficult or costly to manage. Recently, two toxin–antidote gene drive constructs were successfully developed and experimentally validated in the model plant Arabidopsis thaliana21–23: CRISPR-Assisted Inheritance using NPG1 (CAIN24; 1 Department of Computational Biology, Cornell University, Ithaca, NY, USA. 2Center for Genetic Epidemiology, Department of Population and Public Health Sciences, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA. 3Department of Molecular Biology and Genetics, Cornell University, Ithaca, NY, USA. 4Center for Bioinformatics, School of Life Sciences, Center for Life Sciences, Peking University, Beijing, China. 5These authors contributed equally: Isabel K. Kim, Leqi Tian. 6These authors jointly supervised this work: Jackson Champer, Philipp W. Messer, Jaehee Kim. e-mail: Nature Plants Article https://doi.org/10.1038/s41477-026-02256-1 a CAIN Germline Cas9 activity b ClvR Germline Cas9 activity Undisrupted target Disrupted target Wild-type Drive Meiosis Non-viable pollen Viable pollen Meiosis Viable ovule Non-viable pollen Viable pollen Non-viable ovule Viable ovule Fig. 1 | Gene drive systems. a, CAIN drive24. In the germline of a drive carrier, the drive cleaves each wild-type target allele at a rate equal to the sex-specific germline cleavage rate. The target gene is essential only for pollen germination; thus, ovules of any genotype are viable. Pollen inheriting the drive construct are always viable, since the drive contains a cleavage-resistant recoded functional target allele. A pollen grain with a wild-type allele and a disrupted target gene (leftmost gamete, bottom row) is non-viable with a probability equal to the penetrance rate of the target gene (estimated at 96%). For suppression, we define the ‘CAIN male suppression’ drive as the CAIN drive inserted into an essential haplosufficient male fertility gene, such that males with two drive alleles are sterile. b, ClvR drive25. ClvR functions similarly to CAIN but targets a gene that is essential for viability in both pollen and ovules. However, ovules inheriting disrupted alleles and no drive allele from drive-carrying females (leftmost gamete, bottom row) remain viable with a probability equal to a maternal carryover rate (estimated at 20.7%); otherwise, they are non-viable. For suppression, we define two drives. ‘ClvR male suppression’ targets an essential haplosufficient male fertility gene, such that homozygous males are sterile, and ‘ClvR female suppression’ targets an essential haplosufficient female fertility gene, such that homozygous females are sterile. Fig. 1a) and Cleave-and-Rescue gamete killer (ClvR25; Fig. 1b). Unlike traditional gene drives that depend on homology-directed repair (an inefficient mechanism in plants due to their strong tendency towards end-joining DNA repair pathways26), toxin–antidote gene drives avoid the need for homology-directed repair altogether. Instead, these drives spread by linking a toxin (Cas9 and guide RNAs (gRNAs) targeting an essential gene) to an antidote (a tightly linked, cleavage-resistant copy of the target gene), resulting in the elimination of genotypes carrying disrupted alleles but lacking the drive27. In experimental crosses, CAIN targeted NPG1, an essential gene required for pollen germination, achieving inheritance rates up to 97% through male gametes24. ClvR targeted YKT61, a gene required for viability in both pollen and ovules, achieving near-complete inheritance through males and substantial, though lower, inheritance rates through females25. Preliminary modelling in both studies indicated that such drives could spread rapidly and reliably eliminate populations under panmictic, outcrossing conditions24,25. However, these initial plant gene drive studies left out a key aspect of plant life history: seed dormancy. Dormancy allows seeds to rema (...truncated)