Introduction
The management of Palmer amaranth (Amaranthus palmeri S. Watson) has become an obstacle for crop production in many countries, including the United States (Chahal et al. Reference Chahal, Aulakh, Jugulam, Jhala, Price, Kelton and Sarunaite2015; Gazziero et al. Reference Gazziero, Silva, Silveira, Duke and Cerdeira2023; Matzrafi et al. Reference Matzrafi, Scarabel, Milani, Iamonico, Torra, Recasens, Montull, Llenes, Gazoulis, Tataridas, Rubin, Pardo, Cirujeda, Marí and Mennan2025). The presence of A. palmeri was reported to impact row crop growth and reduce yield up to 91% due to its aggressive competitive nature, depending on the density and time of emergence (Klingaman and Oliver Reference Klingaman and Oliver1994; Massinga et al. Reference Massinga, Currie and Horak2001; Morgan et al. Reference Morgan, Baumann and Chandler2001). Overall, chemical control with herbicides is the most common method to manage weeds in crop fields in the United States (Zimdahl and Basinger Reference Zimdahl and Basinger2024). However, A. palmeri has evolved resistance to herbicides from nine site-of-action groups, and single accessions carrying resistance to herbicides from six and seven sites of action have been documented (Carvalho-Moore et al. Reference Carvalho-Moore, Norsworthy, Souza, Barber, Piveta, Meiners and Porri2025b; Heap Reference Heap2025; Shyam et al. Reference Shyam, Borgato, Peterson, Dille and Jugulam2021). Although diversified herbicide programs overlapping preemergence and postemergence herbicides with varied sites of action are recommended to manage herbicide-resistant accessions (Norsworthy et al. Reference Norsworthy, Ward, Shaw, Llewellyn, Nichols, Webster, Bradley, Frisvold, Powles, Burgos, Witt and Barrett2012), satisfactory control with herbicides is compromised in areas infested with A. palmeri accessions harboring multiple resistance.
Glufosinate (Herbicide Resistance Action Committee [HRAC]/Weed Science Society of America [WSSA] Group 10) is one of the effective postemergence herbicide options available to control herbicide-resistant A. palmeri (Cahoon et al. Reference Cahoon, York, Jordan, Everman, Seagroves, Culpepper and Eure2015; Singh et al. Reference Singh, Tyre, Perez-Jones, Krebel, Willis, Herrmann, Klingaman, Head and Aradhya2023). Susceptible biotypes are controlled by glufosinate through the inhibition of the isoforms of the enzyme glutamine synthetase (cytosolic: GS1; chloroplastic: GS2) and the production of high levels of reactive oxygen species. These events are followed by rapid cell death (Bayer et al. Reference Bayer, Gugel, Hägele, Hagenmaier, Jessipow, König and Zähner1972; Takano et al. Reference Takano, Beffa, Preston, Westra and Dayan2019, Reference Takano, Beffa, Preston, Westra and Dayan2020). The continuous use of glufosinate has led to the evolution of resistance in a few monocotyledon species, such as goosegrass [Eleusine indica (L.) Gaertn.] and Italian ryegrass (Lolium perenne L. ssp. multiflorum (Lam.) Husnot) (Heap Reference Heap2025). In dicots, the first case of glufosinate resistance was reported in A. palmeri accessions from Arkansas (Priess et al. Reference Priess, Norsworthy, Godara, Mauromoustakos, Butts, Roberts and Barber2022), with additional glufosinate-resistant accessions later encountered in Missouri and North Carolina (Jones et al. Reference Jones, Dunne, Cahoon, Jennings, Leon and Everman2024; Noguera et al. Reference Noguera, Porri, Werle, Heiser, Brändle, Lerchl, Murphy, Betz, Gatzmann, Penkert, Tuerk, Meyer and Roma-Burgos2022).
Dispersal of weeds and potential introduction of herbicide resistance in areas with susceptible plants can occur through the aid of wind, water, humans, animals, or machinery within and between fields via pollen, seeds, or vegetative structures (Bagavathiannan et al. Reference Bagavathiannan, Norsworthy, Scott and Barber2013; Jasieniuk et al. Reference Jasieniuk, Brûlé-Babel and Morrison1996; Zimdahl and Basinger Reference Zimdahl and Basinger2024). In dioecious species like A. palmeri, the reproductive organs are separated into distinct male and female individuals and have an obligatory outcrossing behavior, which aids the movement of herbicide-resistance genes and other adaptative traits (Borgato et al. Reference Borgato, Ohadi, Brunharo, Patterson and Matzrafi2025; Sauer Reference Sauer1957; Sosnoskie et al. Reference Sosnoskie, Webster, Kichler, MacRae, Grey and Culpepper2012; Stark et al. Reference Stark, Burton, Jordan, Richardson, Spears, Hoyle and Wang-Li2012; Thomson and Brunet Reference Thomson and Brunet1990). Previous research has shown the potential of A. palmeri pollen to move into adjacent fields (Sosnoskie et al. Reference Sosnoskie, Webster, Kichler, MacRae, Grey and Culpepper2012; Stark et al. Reference Stark, Burton, Jordan, Richardson, Spears, Hoyle and Wang-Li2012). Moreover, pollen from glyphosate-resistant (HRAC/WSSA Group 9) plants migrated up to 300 m into susceptible individuals, resulting in 20% of the progeny carrying resistance (Sosnoskie et al. Reference Sosnoskie, Webster, Kichler, MacRae, Grey and Culpepper2012).
Besides pollen-mediated gene flow, herbicide resistance can also move spatially through seeds. There is evidence that glyphosate-resistant A. palmeri was introduced in the Pacific Northwest states, such as Idaho and Oregon, through birdfeed, livestock feed, manure, farm equipment, or roadsides contaminated by vehicles transiting between states (Adjesiwor et al. Reference Adjesiwor, Liu, Felix and Alder2024). Herbicide-resistant A. palmeri was recently identified in northeastern states like New York and Connecticut (Aulakh et al. Reference Aulakh, Chahal, Kumar, Price and Guillard2021, Reference Aulakh, Kumar, Brunharo, Veron and Price2024; Butler-Jones et al. Reference Butler-Jones, Maloney, McClements, Kramer, Morran, Gaines, Besançon and Sosnoskie2024). Furthermore, importing contaminated machinery or grain has been suggested as the entry route for A. palmeri accessions already resistant to glyphosate or acetolactate synthase–inhibiting herbicides (HRAC/WSSA Group 2) in South America and European countries (Gazziero et al. Reference Gazziero, Silva, Silveira, Duke and Cerdeira2023; Manicardi et al. Reference Manicardi, Scarabel, Llenes, Montull, Osuna, Torra Farré and Milani2023, Reference Manicardi, Milani, Scarabel, Mora, Recasens, Llenes, Montull and Torra2025; Matzrafi et al. Reference Matzrafi, Scarabel, Milani, Iamonico, Torra, Recasens, Montull, Llenes, Gazoulis, Tataridas, Rubin, Pardo, Cirujeda, Marí and Mennan2025). Although only a limited number of seeds entered through the aforementioned routes, each emerged A. palmeri female plant has the potential to produce thousands of seeds (Borgato et al. Reference Borgato, Ohadi, Brunharo, Patterson and Matzrafi2025; Keeley et al. Reference Keeley, Carter and Thullen1987; Webster and Grey Reference Webster and Grey2015). Previous work conducted with glyphosate-resistant A. palmeri showed that 20,000 seeds in a square meter, which is less than what a single female plant can produce, led to field areas up to 0.77 ha completely infested with resistance in less than 2 yr (Norsworthy et al. Reference Norsworthy, Griffith, Griffin, Bagavathiannan and Gbur2014). Regardless of the entry route or the spreading mechanism, A. palmeri is highly mobile and adaptable to different environments.
Herbicide-resistant A. palmeri biotypes can quickly infest previously “resistance-free” fields. Even though glufosinate resistance in A. palmeri has been reported in isolated areas, great concern exists regarding the spread of the resistant biotype. The accession MSR2 was collected in 2020 from a cotton (Gossypium hirsutum L.) field in Mississippi County, AR, USA, and it was identified as highly resistant to glufosinate (24-fold) when compared with susceptible standards (Priess et al. Reference Priess, Norsworthy, Godara, Mauromoustakos, Butts, Roberts and Barber2022). Additionally, the amplification and overexpression of GS2 was identified as a resistance mechanism for the MSR2 accession (Carvalho-Moore et al. Reference Carvalho-Moore, Norsworthy, González-Torralva, Hwang, Patel, Barber, Butts and McElroy2022). In the following years, upon visiting fields adjacent to where MSR2 was collected in 2020, A. palmeri plants escaping herbicide-centric control programs were frequently observed. Although the amount of glufosinate sprayed in Mississippi County is not available, it is believed that a significant portion of corn (Zea mays L.), cotton, and soybean [Glycine max (L.) Merr.] fields in this area receive an in-crop application of this herbicide based on conversations with farmers and county extension specialists.
The initial hypothesis is that accessions collected closer to the collection site of MSR2 will likely carry similar glufosinate-resistance levels and mechanisms. Therefore, the objective of this study was to quantify the distribution of glufosinate resistance among A. palmeri accessions collected in 2023 within a 15-km radius surrounding the collection site for MSR2 in 2020, focusing on the distance and direction patterns. Additionally, the cytosolic (GS1) and chloroplastic (GS2) glutamine synthetase copy number were quantified in selected glufosinate survivors to identify any similarity in the resistance mechanism.
Materials and Methods
Collection of Amaranthus palmeri Accessions
The accessions assessed in this study were collected in 2023 from a 15-km radius around the MSR2 accession, which was identified in 2020 in Mississippi County, AR, USA (35.826167°N, 90.240389°W). It is important to acknowledge that another glufosinate-resistant A. palmeri accession (MSR1) was located 5.5 km east (35.832444°N, 90.179805°W) of the MSR2 collection. Following the methodology proposed by Burgos et al. (Reference Burgos, Tranel, Streibig, Davis, Shaner, Norsworthy and Ritz2013) for sampling size for an obligate outcrossing species, a minimum of 5 and up to 10 A. palmeri female inflorescences were collected from a total of 66 six sampling points (Figure 1). The inflorescences from each sampling site were pooled together, forming an accession. Collection sites were randomly selected to better represent the geographic area around MSR2 and were located at least 300 m apart. Even though the priority was to collect seeds from A. palmeri plants left uncontrolled inside crop fields, this scenario was not feasible in all locations. Therefore, inflorescences were collected from A. palmeri plants located in 26 fields, 24 field edges (margin area of a field commonly used for boom calibration), and 16 field ditches (channels located adjacent to agricultural fields for drainage purposes).
Figure 1. Map depicting the location where MSR2 (highly glufosinate-resistant accession) was collected and samples were collected. The square symbolizes where Mississippi County is located in Arkansas, the green star in the middle of the circle shows the location of MSR2 (35.826167°N, 90.240389°W), and the blue triangle shows the location of MSR1 (35.832444°N, 90.179805°W).
Glufosinate Screening
The inflorescences collected were threshed and stored in a cold room at 4 C. The 66 accessions were planted in individual trays filled with potting mix (Sun Gro® Horticulture, Agawam, MA, USA). The seedlings were grown in greenhouses located at the Milo J. Shult Agricultural Research and Extension Center in Fayetteville, AR, with 25 ± 5 C and 16-h photoperiod. At the cotyledon stage, A. palmeri seedlings were transplanted into 50-cell trays (Greenhouse Megastore, Danville, IL, USA) filled with potting mix with a depth of 5.9 cm and 110-cm3 volume in each cell. Germination rates varied among accessions, which impacted the number of plants sprayed in each experimental run. A minimum of 70 and a maximum of 225 plants were screened per accession, divided into at least two experimental runs. A total of 7,922 A. palmeri plants were screened across the 66 accessions. A glufosinate-susceptible standard was also included to ensure efficacy. Plants were sprayed at the 5- to 7-leaf stage with glufosinate (Liberty®, BASF Ag Products, Research Triangle Park, NC, USA) at 590 g ai ha−1. Treatments were applied using a two-nozzle spray chamber adjusted to deliver 187 L ha−1 at 1.6 km h−1 using 1100067 nozzles (TeeJet® Technologies, Glendale Heights, IL, USA).
The number of A. palmeri plants alive was counted before spraying and 21 d after glufosinate treatment (DAT) to calculate mortality. Mortality was calculated using Equation 1:
$$\begin{align}\scriptstyle{{\rm{Mortality}}\left( {\rm{\% }} \right) = \left[ {{{{\rm{No}}.{\rm{\;of\;plants\;alive\;prior\;to\;treatment}} - {\rm{no}}.{\rm{\;of\;plants\;alive\;at\;}}21{\rm{\;DAT}}}}\over{{{\rm{No}}.{\rm{\;of\;plants\;alive\;prior\;to\;treatment}}}}} \right] \times 100}\end{align}$$
Wind speed (m s−1) and direction (blowing from) data from the MSR2 collection site were obtained from the National Aeronautics and Space Administration (NASA) Prediction of Worldwide Energy Resources (POWER) project v. 2.4.9 (NASA 2025). Following the pattern of A. palmeri pollen dispersion in Arkansas, data were used only from July, August, September, and October (Figure 2). Wind rose plots were constructed for each month of interest using the information of the years 2019 (a year before the report of putative resistance), 2020 (year of collection of MSR2), 2021, 2022, and 2023 (year of collection), using WR View Plot freeware v. 8.0.2 (Lakes Software, Waterloo, ONT, Canada).
Figure 2. Wind rose plot showing the distribution and frequency (%) of monthly (July, August, September, and October) average weed speed (m s−1) and direction (blowing from) from 2019 to 2023 at the MSR2 location. The data used to produce plots were obtained from the National Aeronautics and Space Administration (NASA) Prediction of Worldwide Energy Resources (POWER) project v. 2.4.9 (NASA 2025).
Gene Copy Number Assay
Following the glufosinate screening, leaf samples were collected from survivors of accessions with mortality rates less than 96% (n = 46 accessions). Approximately 100 mg of leaf tissue was collected from a total of 251 survivors with a minimum of three biological replicates per selected accession, and genomic DNA was extracted using a modified cetyltrimethylammonium bromide (CTAB) protocol (Doyle and Doyle Reference Doyle and Doyle1987). Additionally, DNA from two well-characterized susceptible standards collected in South Carolina in 1986 (S1) and in Arkansas in 2001 (S2) were included for comparison. For each susceptible standard, a total of five biological replicates were used.
To quantify the GS1 (GS1.1 and GS1.2) and GS2 (GS2.1 and GS2.2) copy number variation among survivors, a nanowell-based digital PCR (dPCR) was conducted. The dPCR reaction volume (12 µl) consisted of 1.48 µl (6 ƞg) of DNA, 0.48 µl (0.2 µM) of specific primers (Table 1), 0.25 µl (0.2 µM) of probe (biomers.net GmbH, Ulm, Germany), 3 µl of QIAcuity Probe PCR Kit mix (Qiagen, Hilden, Germany), and 3.92 µl PCR-Grade H2O for dPCR. The assay was performed in a dPCR thermal cycler (QIAcuity One, 5plex Device, Qiagen, Hilden, Germany) in a 96-well nanoplate with 8,500 nanowells for each sample (QIAcuity Nanoplate 8.5k 96-well) under the following conditions: 2 min at 95 C and 55 cycles of 15 s of denaturation at 95 C; 40 s of annealing, elongation, and detection at 60 C. Partitions were imaged with the following conditions: FAM and HEX, 500-ms exposure time, gain set to 6; ROX 400-ms exposure time, gain set to 6. Qiagen’s QIAcuity Software Suite (v. 2.1.8) was used to determine sample thresholds using positive, negative, and no-template control wells, as well as the copy number variation.
Table 1. Digital and quantitative PCR primer information.
a Abbreviations: GS1.1 and GS1.2, cytosolic glutamine synthetase isoforms; GS2.1 and GS2.2, chloroplastic glutamine synthetase isoforms; qPCR, real time quantitative PCR; dPCR, digital PCR.
To determine the amplification of the GS2.1 and GS2.2 isoforms, TaqMan™ technology was used. A multiplex approach for the target and reference (Actin) genes (Table 1) was used in these assays. The real-time quantitative PCR (qPCR) was performed in a final volume of 25 µl with 6.25 µl of DNA, 1 µl (0.2 µM) of specific primers (Table 1), 0.25 µl (0.2 µM) of probe (biomers.net GmbH, Ulm, Germany), 12.5 µl of SensiFAST Real-Time PCR Kit (Meridian Bioscience, Luckenwalde, Germany), and 2.5 µl PCR-Grade H2O. Three technical replicates were used for each sample. The assay was performed in a qPCR thermal cycler (CFX96 Touch Real-Time PCR Detection System, Bio-Rad Laboratories GmbH, Germany) under the following conditions: 5 min at 95 C and 35 cycles of 10 s of denaturation at 95 C; followed by 30 s at 60 C for annealing, elongation, and detection. The evaluation, according to the
$\;{2^{ - \Delta \Delta {C_T}}}$
method, was carried out with the software Bio-Rad CFX Maestro 2.2 v. 5.2.008.0222.
Data Analysis
Data visualization and analysis were conducted using R 4.4.1 (R Core Team 2024), JMP® Pro 18.0.2 (SAS Institute, Cary, NC, USA), and SigmaPlot 15.0 (Systat Software, San Jose, CA, USA). The relationships between the response variables, mortality and GS2 (GS2.1 and GS2.2) copy number, and the covariates, direction, and distance in relation to the location of MSR2 collection were investigated using generalized linear models (GLMs). The GLMs were fit using the glmmTMB (Brooks et al. Reference Brooks, Kristensen, Benthem, Magnusson, Berg, Nielsen, Skaug, Mächler and Bolker2017) library in the R environment. The mortality and copy number were modeled using a beta and a lognormal distribution, respectively (Gbur et al. Reference Gbur, Stroup, McCarter, Durham, Young, Christman, West and Kramer2012). Therefore, the link functions used in the modeling process were a logit function for the mortality response and a log function for the GS2 copy number. The fitting process included building models in increasing order of complexity, going from including only an intercept to having direction, distance, and the interaction term between the two. Model residuals were visually inspected for patterns and spatial clustering. The generalized models can be described as follows:
$$\matrix{ {g\left( {{Y_i}} \right) = {{\rm{\Phi }}_i}} \cr {{{\rm{\Phi }}_i} = {{\rm X}_i}{\rm \beta} + { \in _i}} \cr {\in \,\sim N\left( {0,{\rm{\Sigma }}} \right)} \cr }$$
where the response variable Y i is mortality or gene copy number, g(Y i) is the link function, Φi models the linear combination of the covariate-specific parameters in the parameter vector β (i.e., direction and distance), and X i is the observed data in the design matrix. The errors (ϵi) are assumed to be normally distributed and spatially related by the variance–covariance matrix Σ. The inclusion of covariates in the model was carried out using a model comparison approach between models containing all combinations of the covariates. The most parsimonious models were selected based on the lowest Akaike information criterion (AIC). When the selected model included the effect of covariate, multiple-comparison tests were conducted to compare the expected means for different cardinal directions using Bonferroni’s adjustment for error rate control.
Results and Discussion
Glufosinate mortality varied among the A. palmeri accessions, ranging from 100% to 34% (Figure 3). Out of 66 accessions, high efficacy (mortality ≥ 99%) was observed for 20 accessions, and glufosinate resistance is unlikely to be present in these fields. According to Frans et al. (Reference Frans, Talbert, Marx, Crowley and Camper1986), a satisfactory response is observed when a herbicide provides ≥80% control of the weed species studied. Using this scale, glufosinate obtained satisfactory control of most of the A. palmeri accessions (51 out of 66 samples) collected around the MSR2 collection site. However, it is important to note that the growing environment (temperature, humidity, and light availability) and spraying conditions for accessions in the greenhouse were optimal, which may not always be representative of field conditions. Glufosinate may underperform in some of these areas. Concerningly, there were 15 accessions having a mortality of <80%.
Figure 3. Mortality distribution from Amaranthus palmeri accessions (n = 66 accessions) collected around MSR2 (highly glufosinate-resistant accession). The box plot was generated using the mortality data collected from 66 accessions, with the center line representing the median, box limits representing the upper and lower quartiles, whiskers representing the 1.5× interquartile, and points representing the outliers.
Different from the initial hypothesis, the distance from MSR2, at least out to 15 km, did not influence the mortality response (Table 2). For mortality, the most parsimonious model (Model 2) detected the covariate direction as a positive response predictor for glufosinate mortality (Table 2; Figure 4). Accessions collected north (315° to 45°) relative to MSR2 tended to have lower average mortality (83%), which was statistically different from accessions collected east of it (95%). Recent multistate screenings showed that glufosinate obtained satisfactory control of several accessions of A. palmeri or its relative, waterhemp [Amaranthus tuberculatus (Moq.) Sauer] (Adjesiwor et al. Reference Adjesiwor, Liu, Felix and Alder2024; Mahoney et al. Reference Mahoney, Jordan, Roma-Burgos, Jennings, Leon, Vann, Everman and Cahoon2020; D Singh et al. Reference Singh, Tyre, Perez-Jones, Krebel, Willis, Herrmann, Klingaman, Head and Aradhya2023; N Singh et al. Reference Singh, Peters, Miller, Naeve and Sarangi2024; Williams et al. Reference Williams, Marshall, Cutulle, Plumblee and Bridges2024). However, the result of this study shows that putative glufosinate-resistant A. palmeri is likely distributed into a larger number of locations across the investigated range than previously determined, highlighting the need for region-specific resistance management efforts. Like MSR2, the accession MSR1 was also detected in 2020. As previously mentioned, the MSR1 collection site was located only 5.5 km east of the MSR2 location. The presence of two A. palmeri accessions harboring glufosinate resistance in proximity hints to the possibility of additional fields infested with resistant biotypes as early as 2020, albeit unreported.
Table 2. Models generated for mortality.a
a Abbreviations: AIC, Akaike information criterion; BIC, Bayesian information criterion. Yes or No indicates whether a parameter was or was not included in the model fit, respectively.
Figure 4. Glufosinate mortality (%) of Amaranthus palmeri accessions (n = 66 accessions) collected around MSR2 (highly glufosinate-resistant accession). Bars with the same lowercase letter are not statistically different according to multiple comparisons tests (α = 0.05) using Bonferroni’s adjustment for error rate control. Abbreviations: E, east (46° to 135° from MSR2); N, north (316° to 45° from MSR2); S, south (136° to 225°); W, west (226° to 315°).
Besides glufosinate mortality, assays were conducted to estimate the copy number of glutamine synthetase isoforms in survivors (n = 251 survivors) of selected accessions (n = 46 accessions). No variation was observed in the susceptible samples for any of the genes tested (Figure 5). Within and among accessions, A. palmeri survivors showed no copy number variation for GS1.1 and GS1.2, while the GS2.1 and GS2.2 copy numbers varied considerably (Figure 5). Values for GS2.1 and GS2.2 ranged from 0.8 to 42 and 0.8 to 18 copies, respectively. Similar to mortality, the distance from MSR2 did not influence the copy number of either GS2 isoform (Tables 3 and 4). For GS2.1, the most parsimonious model (Model 2) detected the covariate direction as a positive response predictor (Table 3; Figure 6). Accessions collected north from MSR2 had a higher average number of GS2.1 copies, statistically different from accessions collected in any other direction relative to MSR2. For comparison, the GS2.1 averaged 2.3 copies for survivors from accessions collected north of MSR2, and it ranged from 1 to 1.2 copies for survivors in the other directions (east, south, or west). For GS2.2, the model with the best fit (Model 1) did not include distance or direction relative to the MSR2 collection site as covariates (Table 4), which indicates that there is no covariate influencing this response.
Figure 5. Glutamine synthetase copy number distribution among (A) Amaranthus palmeri survivors from accessions (n = 46 accessions) collected around MSR2 (highly glufosinate-resistant accession) following glufosinate screening; (B) nontreated plants from susceptible standards (S1 and S2). Abbreviations: GS1.1 and GS1.2, cytosolic glutamine synthetase isoforms; GS2.1 and GS2.2, chloroplastic glutamine synthetase.
Table 3. Models generated for GS2.1.a
a Abbreviations: AIC, Akaike information criterion; BIC, Bayesian information criterion. Yes or No indicates whether a parameter was or was not included in the model fit, respectively.
Table 4. Models generated for GS2.2.a
a Abbreviations: AIC, Akaike information criterion; BIC, Bayesian information criterion. Yes or No indicates whether a parameter was or was not included in the model fit, respectively.
Figure 6. Chloroplastic glutamine synthetase isoform (GS2.1) copy number of Amaranthus palmeri survivors from accessions (n = 46 accessions) collected around MSR2 (highly glufosinate-resistant accession). Means followed by the same lowercase letter are not statistically different according to multiple comparisons tests (α = 0.05) using Bonferroni’s adjustment for error rate control.
The covariate direction was a positive predictor for both mortality and GS2.1 copy number, with survivors collected north of MSR2 exhibiting the lowest accession average mortality (83%; Figure 4) and highest GS2.1 copy number (2.3 copies; Figure 6). Conversely, Sosnoskie et al. (Reference Sosnoskie, Webster, Kichler, MacRae, Grey and Culpepper2012) observed that direction did not affect resistance transfer via pollen from resistant to susceptible A. palmeri plants. In the same study, distance significantly impacted the resistance spreading, with a higher percentage of resistant individuals being at closer distances (up to 5 m). A different study evaluating pollen dispersal in cotton fields observed that 82% of pollen captured was within 2 m of the A. palmeri source with no correlation to direction (Stark et al. Reference Stark, Burton, Jordan, Richardson, Spears, Hoyle and Wang-Li2012). Interestingly, when examining the 5-yr average predominant wind direction for months when A. palmeri pollen dispersal will likely occur in Arkansas (July, August, September, and October), wind blowing toward the north of the MSR2 collection site was observed only near the end of the crop season in October (Figure 2). Therefore, wind patterns in this region may not be involved in the movement of resistance, assuming that MSR2 was the origin of resistance. The absence of a distance-dependency effect and strong evidence that direction influences the glufosinate mortality response in this study suggest that glufosinate resistance may be spreading through routes beyond localized pollen or seed dispersal.
Gene flow in plants can occur via pollen or seed dispersal (Ennos Reference Ennos1994), and herbicide resistance migration can be transmitted over long distances in A. palmeri. In a study evaluating the pollen-mediated movement of glyphosate resistance between susceptible and resistant A. palmeri biotypes, moderate outcrossing (20%) occurred at 300 m (Sosnoskie et al. Reference Sosnoskie, Webster, Kichler, MacRae, Grey and Culpepper2012). Similarly, the pollen of A. tuberculatus remained viable to at least 800 m up to 120 h after dispersal (Liu et al. Reference Liu, Davis and Tranel2012). Besides the mobility via pollen, A. palmeri seeds are small and easily transported as well. For instance, the entry of herbicide-resistant A. palmeri to different countries has been linked to imported grain or machinery contaminated with seeds (Gazziero et al. Reference Gazziero, Silva, Silveira, Duke and Cerdeira2023; Manicardi et al. Reference Manicardi, Scarabel, Llenes, Montull, Osuna, Torra Farré and Milani2023). The spread of contaminated residues in production areas is also a possibility. In Arkansas, viable A. palmeri seeds were found in composted cotton gin trash, which is usually spread onto fields during fallow months (Norsworthy et al. Reference Norsworthy, Smith, Steckel and Koger2009). Seeds of Amaranthus species, including A. palmeri, were also present in the surface water of irrigation canals and were recovered from the digestive tracts of migratory birds (Farmer et al. Reference Farmer, Webb, Pierce and Bradley2017; Kelley and Bruns Reference Kelley and Bruns1975; Wilson Reference Wilson1980). Although the wind has less impact on the dispersal of A. palmeri seeds compared with pollen, the introduction of this species in previously non-infested areas in Texas was connected to a hurricane in 1980 (Menges Reference Menges1987).
An extrachromosomal circular DNA structure co-amplifying both GS2.1 and GS2.2 isoforms has been characterized and validated in MSR2 plants (Carvalho-Moore et al. Reference Carvalho-Moore, Borgato, Cutti, Porri, Meiners, Lerchl, Norsworthy and Patterson2025a). Despite being collected near the site for MSR2, only four accessions showed amplification of both isoforms among the survivors evaluated (data not shown). This result hints that the existence of additional arrangements might be driving the amplification of the GS2 gene. In fact, different amplification patterns were identified in the aforementioned MSR1 glufosinate-resistant accession and in an accession from Missouri, where only the GS2.1 isoform showed gene amplification (Carvalho-Moore et al. Reference Carvalho-Moore, Borgato, Cutti, Porri, Meiners, Lerchl, Norsworthy and Patterson2025a; Noguera et al. Reference Noguera, Porri, Werle, Heiser, Brändle, Lerchl, Murphy, Betz, Gatzmann, Penkert, Tuerk, Meyer and Roma-Burgos2022). Additionally, survivors from 27 accessions (out of 46 selected accessions), with mortality ranging from 57% to 96%, did not show amplification of any of the GS2 isoforms (data not shown). This result suggests that an additional resistance mechanism, other than GS2 amplification, might be involved.
It is important to re-emphasize that the MSR2 accession was collected in 2020, whereas the accessions analyzed in this study were collected in 2023. As a result, there is a gap in knowledge regarding the management practices used in this region from 2020 to 2023, which may have impacted the selection and spread of resistant mechanisms and resistant individuals. Moreover, volunteer A. palmeri plants were observed on roadsides throughout the collection region (PC-M, personal observations). Previous studies have shown that A. palmeri accessions collected from roadsides, field edges, or ditches harbored herbicide resistance, reflective of chemical failures that often occurred in adjacent fields, and acted as carriers of resistance (Bagavathiannan and Norsworthy Reference Bagavathiannan and Norsworthy2016; Vieira et al. Reference Vieira, Samuelson, Alves, Gaines, Werle and Kruger2018). Moreover, the recurring exposure to sublethal herbicide doses in field edges can increase the tolerance of problematic weeds (Tehranchian et al. Reference Tehranchian, Norsworthy, Powles, Bararpour, Bagavathiannan, Barber and Scott2017; Vila-Aiub and Ghersa Reference Vila-Aiub and Ghersa2005). Zero tolerance is strongly recommended to manage resistant populations, especially species with high seed production (Keeley et al. Reference Keeley, Carter and Thullen1987; Norsworthy et al. Reference Norsworthy, Ward, Shaw, Llewellyn, Nichols, Webster, Bradley, Frisvold, Powles, Burgos, Witt and Barrett2012, Reference Norsworthy, Griffith, Griffin, Bagavathiannan and Gbur2014). Hand weeding is practiced by some growers within the radius of fields sampled here (JK Norsworthy, personal observation).
Glufosinate is a valuable postemergence herbicide. However, the findings presented here show that putative glufosinate-resistant A. palmeri populations are present in more areas than initially detected and are a threat to the stewardship of this and other technologies. Ideally, management practices to minimize farther movement of the resistant biotype need to be applied soon, because resistance management is easier with smaller or localized populations (Adjesiwor et al. Reference Adjesiwor, Liu, Felix and Alder2024; Norsworthy et al. Reference Norsworthy, Ward, Shaw, Llewellyn, Nichols, Webster, Bradley, Frisvold, Powles, Burgos, Witt and Barrett2012). In 2021, the state of Minnesota reported the complete eradication of A. palmeri infestations that were detected in 2016. Yu et al. (Reference Yu, Blair, Hardel, Chandler, Thiede, Cortilet, Gunsolus and Becker2021) reported an aggressive protocol that included intensive scouting, area burning, torching, and herbicide applications, as well as regulatory support and collaboration with agencies of interest. Although this is an ambitious and utopian approach for Arkansas due to the high presence of A. palmeri in the state, the high collaboration and communication between different entities, followed by rapid response, is valuable. Zero tolerance with the physical removal of any A. palmeri field escapes and control of plants in nonagricultural areas (roadsides and ditches) is crucial to reduce resistance spreading and perpetuation by avoiding seed deposition and pollen migration (Norsworthy et al. Reference Norsworthy, Ward, Shaw, Llewellyn, Nichols, Webster, Bradley, Frisvold, Powles, Burgos, Witt and Barrett2012; Sosnoskie et al. Reference Sosnoskie, Webster, Kichler, MacRae, Grey and Culpepper2012; Vieira et al. Reference Vieira, Samuelson, Alves, Gaines, Werle and Kruger2018; Webster and Nichols Reference Webster and Nichols2012). Crop and herbicide rotation and a foundational residual program are recommended, especially in the fields where putative glufosinate resistance was detected in A. palmeri. Future investigations should broaden the evaluated radius to assess the extent of glufosinate resistance among A. palmeri accessions throughout Arkansas. Also, it is crucial to unravel the additional mechanisms involved in the response of accessions with low mortality and no variation in gene copy number.
Acknowledgments
The authors thank the University of Arkansas Division of Agriculture for the research support. Also, we acknowledge the colleagues from the Weed Science group, especially Rodrigo Botelho, for their assistance with greenhouse experiments.
Funding statement
BASF Corporation funded this project.
Competing interests
The authors AP, IM, and JL are affiliated with BASF Corporation and BASF SE. The other authors declare no conflicts of interest.
Open access
