A holistic approach in herbicide resistance research and management: from resistance detection to sustainable weed control

Weeds in agroecosystems face frequent, intense selection from herbicides, facilitating rapid evolution of resistance. In Amaranthus palmeri, large seed production fosters standing genetic variation, enabling multiple mechanisms (target-site mutations, gene amplification/overexpression, and non-target-site resistance) to evolve. Selection regime (dose, timing, herbicide diversity) influences which mechanisms are favored. Predicting long-term sustainability is challenging due to biological diversity, complex mechanisms, and variable human practices. Population models can explore evolutionary dynamics under management programs, but are often limited by biological knowledge and spatio-temporal representation. This study integrates field detection, mechanistic assays, alternative herbicide evaluation, and modeling to deliver a holistic understanding and management strategy for glyphosate-resistant A. palmeri in Argentina.

Prior work has documented multiple glyphosate resistance mechanisms in Amaranthus spp., including EPSPS gene amplification and target-site mutations (e.g., P106 substitutions), as well as reduced absorption/translocation. U.S. A. palmeri populations commonly exhibit EPSPS amplification, whereas target-site mutations are less frequent. Modeling frameworks have been used to illustrate principles of resistance evolution and management but often lack local parameterization and realism. Reports of PPO-inhibitor resistance in Amaranthus spp. (USA) and in other species in Brazil underscore risks to alternative chemistries. The literature emphasizes the roles of selection pressure intensity (high vs. low dose), initial resistance frequencies, and management diversity in shaping resistance trajectories.

Plant material and growth conditions: A. palmeri seeds were collected from a soybean field in Villa Valeria (Cordoba, Argentina) with 5 years of continuous glyphosate use. Standard sensitive (ApS1, Azlin Seed Service, USA), a second sensitive (ApS2), and a resistant (ApR; EPSPS gene duplication/overexpression ~9-fold vs ApS1) populations from Georgia, USA were used. Plants were grown in a glasshouse (24/18 °C day/night; 65% RH). Seedlings were used for RISQ agar tests and potted for whole-plant assays. Resistance confirmation: RISQ agar plates contained 0, 25, 50, 100 µM glyphosate; 10 seedlings per plate; survivorship scored 7–12 days post-transplanting by new root/leaf growth. Whole-plant pot tests: 14 plants per treatment; sprayed at ~8 cm height with 0, 100, 200, 400, 800, 1600, 3200, 6400 g ae ha−1 glyphosate; survival assessed 21 days after transplanting. Mechanism studies: Biokinetics assessed uptake/translocation using [14C]-glyphosate at the 4-leaf stage (20 µg glyphosate; 5 kBq per plant). Sampling at 0, 24, 48, 96 h (n=4 per time); treated leaf washed with cellulose acetate/acetone; radioactivity via liquid scintillation counting; plant parts (treated leaf, meristem, rest) oxidized and counted; uptake expressed as % applied; translocation as % absorbed. EPSPS gene duplication/expression: Sixteen untreated VV plants analyzed by qPCR; ApS2 as sensitive control; DNA and RNA extracted (KingFisher/Wizard Magnetic 96 for DNA; RNeasy for RNA; DNase treatment; cDNA by High-Capacity kit). Reference genes: ALS and CPS. Primers/probes provided for ALS, CPS, and EPSPS. Reactions in QuantStudio 7 Flex; 2 technical replicates; randomized plate layout. Copy number and expression quantified relative to reference genes. Target-site sequencing: Eighteen VV plants sequenced around EPSPS codons 102 and 106. PCR product (195 bp) amplified with specified primers; PCR cycling (95 °C 5 min; 30 cycles of 95 °C 30 s, 60 °C 30 s, 72 °C 1 min; final 72 °C 10 min); direct Sanger sequencing; alignment in DNASTAR Lasergene. Statistics: For qPCR DNA (copy number) and cDNA (expression) separately, ANOVA fitted (population effect) and t-tests for pairwise comparisons (p<0.05). Biokinetics analyzed by factorial ANOVA with factors population, time, and their interaction; where interactions existed, comparisons by time; otherwise across times. Population model: Individual-based model (NetLogo 6.0) following Liu et al. (2017), updated for A. palmeri VV population. Annual life cycle simulated over 20 years; herbicide effects depend on emergence overlap and efficacy. Parameter updates included emergence curve, seed production per plant, and annual mortality functions (as provided), and sigma of log-normal distribution for glyphosate quantitative resistance = 0.4656. Soybean planted Nov 10 (71 DASS); pre-plant burndown assumed to control pre-sowing emergence; PRE applied at 0 DAP. Scenarios: S0 glyphosate-only twice POST (20 and 40 DAP). S1–S6 focused on PPO-based programs with specified rates and residual durations (see below). PPO resistance modeled as a target-site mutation with dominance 0.75; homozygous resistant survival 100%, heterozygous 75% when exposed. Fomesafen included residual activity affecting later cohorts; lactofen had no residual. S-metolachlor and metribuzin modeled as residual-only. POST timings in S5 and S6 were 39 and 32 DAP, respectively. Initial PPO resistance proportion set to 10^-2 for some scenarios; populations assumed susceptible to S-metolachlor and metribuzin. Herbicide scenario parameters (from Table 2): S0 glyphosate POST 800 g ae ha−1 (PP efficacy 98.9%; PS 60%; SS 32.2). S1 glyphosate POST 1550 g ae ha−1 (PP 99.9%; PS 87.5%; SS 68.9). S2 fomesafen POST 319.5 g ai ha−1; residual 3.5/5.5 weeks; POST efficacy 96%. S3/S5/S6 lactofen POST 120 g ai ha−1; efficacy 99%. S-metolachlor POST 1295 g ai ha−1 (residual 3/4 weeks). S4/S5 fomesafen POST 285 g ai ha−1 (residual 3/5 weeks; 95% POST efficacy) and PRE 342 g ai ha−1 (residual 3.5/5.5 weeks). S4/S5 S-metolachlor PRE 1554 g ai ha−1 (residual 3.5/4.5 weeks). S6 S-metolachlor PRE 1572.5 g ai ha−1 (3.5/4.5 weeks) plus metribuzin PRE 372.5 g ai ha−1 (2.5/3.5 weeks).

Resistance confirmation: The standard sensitive A. palmeri (ApS1) was fully controlled at 25 µM glyphosate in RISQ and at 200 g ae ha−1 in pot tests (discriminating rates). The standard resistant (ApR) and Villa Valeria (VV) field samples survived these rates, confirming glyphosate resistance. Mechanism of glyphosate resistance: Biokinetics showed no significant difference in glyphosate uptake between VV and ApS1 (p=0.2965). Translocation impairment was not observed; movement to the meristem was significantly higher in VV than ApS1 (p=0.0141). Total glyphosate recoveries exceeded 72% at 96 h, indicating low metabolic degradation. qPCR indicated modest increases in EPSPS relative copy number and expression in VV vs ApS2: DNA EPSPS/CPS ApS2=0.98, VV=1.38 (ratio 1.41; p<0.0001); DNA EPSPS/ALS 1.05 vs 1.43 (ratio 1.36; p<0.0001); RNA EPSPS/CPS 1.24 vs 2.31 (ratio 1.86; p=0.0019); RNA EPSPS/ALS 0.95 vs 1.28 (ratio 1.34; p=0.0054). Sanger sequencing detected the EPSPS P106S target-site mutation with genotype frequencies: PP (wild-type) 6%, PS 22%, SS 72%. Conclusion: VV resistance was endowed by multiple mechanisms, predominantly P106S target-site mutation with minor EPSPS duplication/overexpression. Modeling glyphosate resistance evolution: Using dose-response parameterization from a nearby population (VM1), under glyphosate-only applied twice POST, weed density exceeded the control threshold (1 plant m−2) in an average of 9.7 years. Target-site mutation-based resistance exceeded 20% in an average of 7.4 years, while quantitative resistance remained <0.04% across all 100 replicates. Solo herbicide use tended to select primarily one mechanism, with the selected mechanism dependent on initial frequencies; TS mutation and quantitative resistance were negatively correlated in selection outcomes. Alternative herbicides: PPO inhibitors (lactofen, fomesafen) and residual herbicides (S-metolachlor, metribuzin) provided 100% control of the glyphosate-resistant VV population at field rates. Sustainability of PPO-based programs: Outcomes depended on herbicide exposure (emergence timing vs application) and PPO resistance evolution. A high-rate glyphosate+fomesafen mixture (longer residual) provided exposure but, with initial PPO resistance 10^-2, control failed within 6 years (S1). Lactofen twice POST (no residual) achieved exposure but, as a single MOA, led to control failure within 3 years in partially PPO-resistant populations (S2). Programs using S-metolachlor+fomesafen once (S3, S4) did not cover all cohorts; PRE applications (S4) performed slightly better than POST-only mixtures (S3) due to dual residuals. Applying the mixture both PRE and POST improved durability (S5). Adding chemical diversity by replacing PRE fomesafen with metribuzin (S6) further reduced PPO resistance evolution from early onset compared to S5, maintaining better long-term control. Even in effective programs (S5, S6), modeled PPO resistance frequency could build in the seedbank without immediate density signals, implying risk if programs weaken or rates are reduced.

This holistic study identified that glyphosate resistance in the Argentine VV A. palmeri population is primarily driven by the EPSPS P106S target-site mutation, contrasting with many U.S. populations where EPSPS gene amplification predominates. The selection history in Argentina (extensive solo glyphosate use) likely favored target-site mutation, with only minor quantitative resistance contribution, consistent with model results showing solo use tends to select a single mechanism based on initial frequencies. The work underscores that resistance mechanisms and their evolutionary dynamics are context- and location-specific; parameterization cannot always be transferred across regions or even within species. The success of PPO-based alternatives in bioassays demonstrates viable control options, but modeling indicates their sustainability hinges on adequate exposure through residual activity and chemical diversity, especially in the presence of emerging PPO resistance. Interdisciplinary integration—rapid detection, molecular assays, and predictive modeling—helps capture subtle biological differences that scale to significant population-level consequences over generations and informs locally tailored management strategies.

By integrating rapid diagnostics, mechanistic assays, alternative chemistry testing, and population modeling, this study provided a complete pipeline from detection to sustainable management for glyphosate-resistant A. palmeri in Argentina. The VV population’s resistance is mainly due to the EPSPS P106S mutation with minor EPSPS duplication/overexpression. PPO inhibitors and residual herbicides effectively controlled resistant plants, and modeling highlighted that chemical diversity and residual activity are key to prolonging PPO herbicide utility. The approach advocates localized, interdisciplinary programs to manage evolving weeds. Future work should expand the framework to incorporate environmental, economic, communication, and digital tools, and validate model predictions across diverse fields and seasons to refine recommendations.

- The modeling relied on parameterization of glyphosate dose–responses from a nearby population (VM1) due to similarity in mechanisms, which may introduce uncertainty when extrapolating to VV and other fields. - Assumptions in the model include PPO resistance as a single target-site mutation with dominance 0.75, initial PPO resistance frequency (e.g., 10^-2), and full susceptibility to S-metolachlor and metribuzin; deviations in real fields could alter outcomes. - Spatial heterogeneity and broader spatio-temporal variation were not explicitly represented, limiting field-specific predictions. - The relatively small numbers of plants in some assays (e.g., 16–18 individuals for molecular analyses; two technical qPCR replicates) may limit precision of frequency and expression estimates. - Generalizing recommendations is challenging due to complex agroecosystems and diverse human practices; the study emphasizes localized application rather than universal prescriptions.