Revisiting the origins of glyphosate-resistant giant ragweed (Ambrosia trifida L.) in Canada

Abstract Glyphosate-resistant giant ragweed (Ambrosia trifida L.) was first identified in Canada in 2008. Although early studies attributed resistance in this species solely to non-target site mechanisms, the presence of a proline (P) to serine (S) mutation at position 106 of EPSPS2 in common and giant ragweed has recently been reported. The objective of this research was (i) to determine whether a P106S mutation is present in historical samples of giant ragweed seed collected from the site of the first report of glyphosate resistance, and (ii) to determine the frequency and distribution of P106S in resistant and susceptible biotypes collected as part of historical surveys throughout southwestern Ontario.


Introduction
Giant ragweed (Ambrosia trifida L.) is an economically important weed of row crop production systems in the mid-western USA and southern Ontario (Page and Nurse 2015;Regnier et al. 2016).In the mid-2000s, several US states reported the evolution of resistance to glyphosate in giant ragweed (Heap 2023).Shortly thereafter, in 2008, glyphosate-resistant (GR) giant ragweed was also reported in southern Ontario (Sikkema et al. 2009).Following these reports, numerous research studies were undertaken to identify alternate control options for GR giant ragweed in corn and soybean, as well as to explore the mechanism(s) conferring glyphosate resistance in this species.Researchers identified two characteristic phenotypes associated with GR giant ragweed: (i) a rapid response (RR) phenotype where the mature leaf tissue desiccates and abscises shortly after glyphosate application with new growth eventually resuming from apical and axillary meristems, and (ii) a non-rapid response (NRR) phenotype where, for the first 1-2 weeks after application, growth slows and meristematic tissues become chlorotic followed by a resumption of growth approximately 4 weeks after application (Moretti et al. 2018;Van Horn et al. 2018).These authors characterized and studied biotypes from across the mid-western USA and southern Ontario representing both RR and NRR phenotypes, uniformly reporting a lack of mutations in 5enolpyruvylshikimate-3-phosphate synthase (EPSPS), the target enzyme for glyphosate (N-(phosphonomethyl) glycine).
In members of the Asteraceae, multiple EPSPS genes are often found in the genome.In Canada fleabane (Conyza canadensis L.), for example, two full-length and one truncated EP-SPS genes can be found (Heck et al. 2003;Laforest et al. 2020), with mutations in EPSPS2 alone conferring resistance to glyphosate (Page et al. 2018;Beres et al. 2020).The recent production of chromosome-scale genomes of giant and common ragweed (Ambrosia artemisiifolia L.) identified two EPSPS genes in both species and, importantly, reported the presence of a proline (P) to serine (S) mutation at position 106 (P106S) in one of the two (Laforest et al. 2023).This EPSPS shared homology with EPSPS2 in C. canadensis and was numbered as such.When Laforest et al. (2023) aligned the primers utilised in previous studies of GR giant ragweed to the newly produced genome, it became clear that the primers selectively amplified the EPSPS gene that did not contain the P106S target-site mutation (Suppl.Fig. S1).
The observation of a P106S mutation in EPSPS2 raises the possibility that the initial reports of GR giant ragweed may have overlooked target-site mutation as a contributing mechanism in one or both of the identified resistance phenotypes.
In the current study, we revisit and genotype giant ragweed biotypes collected at or near the point of origin for the initial report of glyphosate resistance in Canada.The objectives of this research were (i) to determine whether a P106S mutation is present in historical samples of giant ragweed collected around the time of the first report of glyphosate resis-

Materials and methods
Giant ragweed involucral achenes (hereafter referred to as seeds) were collected throughout southwestern Ontario during the summers of 2009-2012 as a part of an ongoing survey following the initial report of glyphosate resistance in this species (Vink et al. 2012;Follings et al. 2013).Seedlings of these biotypes were propagated at the University of Guelph--Ridgetown and were characterized for their response to glyphosate.This phenotypic characterization followed the methods described by Van Horn et al. (2018).In brief, biotypes were classified as S, RR, or NRR based on their phenotypic response to glyphosate (0.84-1.8 kg ae ha −1 ).Following this initial characterization, seeds of these biotypes were transferred to the Harrow Research and Development Centre (RDC) and stored at 4 • C and 40% relative humidity until use.The seed collection at Harrow RDC also contained a giant ragweed biotype collected onsite in 1988 and a biotype collected by M. Cowbrough from a riparian area in Kitchener-Waterloo, Ontario, was subsequently added in 2016 (Fig. 1).This later biotype is of note because it was collected from the same riparian area as biotype 25 of Van Horn et al. (2018).
In the autumn of 2022, seeds of the above described biotypes were cold stratified following the methods of Page and Nurse (2017), and it was determined that all except the 2016 riparian biotype were non-germinable.Given this, the embryos of 20 seeds per biotype were excised in preparation for DNA extraction.Excised embryos were ground by pestle in lysis buffer and DNA was extracted using a commercially available kit following the manufacturer's protocol (Macherey-Nagel NucleoSpin Plant II kits, Macherey-Nagel Inc., Bethlehem, PA, USA).The genomic sequence of giant ragweed (Laforest et al. 2023) was used to design EPSPS2 specific primers.A 543 bp fragment encompassing the majority of exon 2 was amplified for each of the 20 embryos using the following primers: forward (5 -GGAGCTTTGAGAGCCCTA-3 ) and reverse (5 -CAAGTTGTTTAAGACCAGTGACTAA-3 ).Primers were synthesized by Integrated DNA Technologies (Coralville, IA, USA).Eluted DNA was amplified by polymerase chain reaction (PCR) with the following reaction conditions: an initial denaturation at 95 • C for 1 min, 35 cycles of 95 • C for 15 s, 56 • C for 15 s for annealing, 72 • C for 30 s, followed by a final extension at 72 • C for 7 min.PCR products were visualized on a 2.5% agarose gel containing 5% nucleic acid staining solution (RedSafe, Froggabio Vaughan, ON, Canada).Following PCR, the samples were cleaned for sequencing using GenepHlow Gel/PCR kit (Geneaid Biotech, New Taipei City, Taiwan) according to the provided protocol.Sanger sequencing of the PCR products was carried out by the London Regional Genomics Centre (Robarts Research Institute, London, ON, Canada) using the same primers from the PCR amplification.Alignment to the A. trifida EPSPS2 reference sequence (GenBank accession number: OR133708) was performed using the Sequencher software (Gene Codes, Ann Arbor, MI, USA) and analyzed for the presence of the P106S target-site mutation.

Results and discussion
Fifteen historic and current giant ragweed biotypes were genotyped for the presence of a P106S mutation in EP-SPS2.This target-site mutation was observed in five resistant biotypes in either a homozygous or heterozygous state (Fig. 2; Table S1).Of these, the allele frequency of P106S was 90% and 100% in two RR biotypes (i.e., biotypes 6 and 7), while a third RR biotype had a single heterozygous individual (i.e., biotype 8).There were also two NRR biotypes with mixtures of homozygous and heterozygous P106S individuals (i.e., biotypes 3 and 4).While these biotypes were collected at locations that roughly correspond to those described by Vink et al. (2012) in their 2008, 2009, and 2010 surveys, it is unclear whether they represent temporal collections of the same biotypes.These five biotypes possessing heterozygous or homozygous P106S individuals were spatially distributed throughout southwestern Ontario, with an average distance of approximately 50 km amongst them.
The phenotypic characterization of the historic biotypes (i.e., RR, NRR, and S) was carried out at the time of their collection and as the seeds were no longer viable, we were unable to confirm their classification.The sole exception to this, biotype 15, was collected from a riparian area in 2016 and was confirmed as susceptible to glyphosate at a dose of 0.9 kg a.e.ha −1 (E.Page, unpublished data).Based on these characterizations, it is clear that not all biotypes initially classified as GR possessed a P106S mutation.Furthermore, there was no clear association of P106S with either the NRR or RR phenotype, suggesting that other non-target site mechanisms of glyphosate resistance were also present in these biotypes.This observation supports and builds on the re-sults of Van Horn et al. (2018) and Moretti et al.(2018), who reported that non-target site resistance mechanisms were present in the initial reports of glyphosate-resistant giant ragweed.These authors concluded that non-target site resistance was present, only because they failed to identify a target-site mutation in EPSPS 1. Their published PCR primers did not amplify EPSPS2 in A. trifida.
The relative contribution of target site and non-target site mechanisms to the observed levels of resistance in glyphosate-resistant biotypes has been a source of interest and debate in the weed science community (Sammons and Gaines 2014) and no direct conclusion can be drawn from the results of the current study.Both target and non-target site mechanisms were present in biotypes collected following the initial report of glyphosate resistance in giant ragweed and many questions concerning the evolution of glyphosate resistance in this species remain to be answered.For instance, it would be of interest to know which resistance mechanism (target or non-target site) evolved initially.Do individuals genotyped as homozygous or heterozygous for P106S also possess non-target site mechanism of resistance and if so, do the resistance mechanisms act additively or synergistically?What is clear from the data is that the P106S target-site mutation was not observed in any of the S biotypes examined.It was not observed biotype 15, which was collected from the same riparian area as biotype 25 in Van Horn et al.( 2018) (M.Cowbrough, personal communication), nor was it observed in biotype 1, whose collection in 1988 predates not only the first report of glyphosate resistance in this species, but also the introduction of GR crops in North America (Green and Siehl 2021).Taken together, these results strongly suggest that P106S is associated with glyphosate resistance in giant ragweed.
The mechanism(s) conferring glyphosate resistance in giant ragweed have remained elusive since the first reports of resistance in the early 2000s.While initial studies ascribed resistance solely to non-target site mechanisms, such as reduced translocation and rapid cell death (Lespérance 2015; Moretti et al. 2018), subsequent research by Laforest et al. (2023) reported the presence of a P106S mutation in EPSPS2 of giant ragweed while producing and annotating the genomes of giant and common ragweed (Ambrosia artemisiifolia).The current research builds on these studies by clearly demonstrating that both target and non-target site resistance mechanisms were present in biotypes collected from southwestern Ontario shortly after the initial report of resistance.Furthermore, it is clear that, while P106S is not always present in resistant biotypes, it was not observed in any of the susceptible biotypes examined.Thus, we hypothesize that P106S acts additively or synergistically with other non-target site mechanisms to confer glyphosate resistance in both the NRR and RR phenotypes associated with glyphosate resistance in giant ragweed.

Fig. 1 .
Fig. 1.Riparian zone bordering the Grand River in the municipality of Kitchener-Waterloo, Ontario.Collection site for giant ragweed biotype #15.

Fig. 2 .
Fig. 2. Frequency and distribution of P106S target-site mutation in resistant (RR = rapid response, NRR = non-rapid response) and susceptible (S) giant ragweed biotypes collected across southwestern Ontario.Produced in 2024 and using data sources from the Ministries of Agriculture Food and Rural Affairs; Municipal Affairs and Housing; Agriculture and Agri-Food Canada, Esri.Projection: Lambert, Conformal Conic, NAD 1983.