Field Survey

A field survey was conducted before the sugar beet harvest in SE Montana (Big Horn, Carbon, Custer, Rosebud, Stillwater, Treasure, and Yellowstone counties) and NW Wyoming in August 2021 (Big Horn, Hot Springs, Park, and Washakie counties) (Figure 1A). Before the survey, the coordinates of sugar beet fields were acquired from local agronomists and loaded into a handheld GPS device (Garmin GPSmap® 76C, Olathe, KS, USA). The northernmost sugar beet field (46.299637°N, 107.226228°W) was visited first, and weed populations were collected from adjacent sugar beet fields every 8 km. If no sugar beet field was found at the 8-km mark, weed populations were collected from the nearest sugar beet field. The agronomists also indicated fields with suspected herbicide resistance; therefore, weed populations were collected from these fields, even if they did not fall within the survey design. This systematic collection approach enabled us to sample across the survey area, capturing a representative distribution of weed populations without excessive sampling. Based on the infestation level, the occurrence of weed populations was visually classified into four different categories within each field: (A) weed-free (i.e., no visible weeds; Figure 2A), (B) isolated (i.e., singular plant in 5-m diameter; Figure 2B), (C) clustered (i.e., multiple plants in 5-m diameter; Figure 2C), and (D) widespread (i.e., plants were spread across the field; Figure 2D).

Figure 1. Infestation pattern of the weed species during an August 2021 survey within the sugar beet cropping systems in southeast Montana and northwest Wyoming: (A) surveyed counties highlighted in orange; (B) number of weed species present (species richness); infestation level and spatial pattern of (C) Bassia scoparia; (D) Amaranthus retroflexus; and (E) Chenopodium album.

Figure 2. Categories of sugar beet fields based on weed infestation visually identified during a survey conducted in southeast Montana and northwest Wyoming counties: (A) weed-free, (B) isolated, (C) clustered, and (D) widespread. Black arrows indicate isolated and clustered infestations of weeds.

Seed Collection and Storage

The seeds of three major weed species, B. scoparia, A. retroflexus, and C. album, and three minor species, wild oats (Avena fatua L.), green foxtail [Setaria viridis (L.) P. Beauv.], and barnyardgrass [Echinochloa crus-galli (L.) P. Beauv.], were collected by detaching inflorescences with a pruning shear and placing them in paper bags. Major and minor species were classified based on their abundance and impact. For each weed species, seeds were collected from at least 2 to 30 individuals within fields and along the field margins. Seeds collected from fields and margins were kept in separate bags. Seeds from each individual within a population were bulked to achieve a representative seed lot. All within-field collections were conducted leaving ≥50 m from field margins. While seeds were collected by walking in a zigzag pattern across the sugar beet fields for weed populations that showed a clustered and widespread dispersal, a targeted collection was used for isolated populations. Seed bags were kept in a hot-air oven at 35 C for a week to prevent mold development due to excessive moisture. Mature seeds were threshed, cleaned, and stored in paper bags at 15 to 17 C in dark conditions for ∼1 to 2 yr at the greenhouse facility of Southern Agricultural Research Center (SARC), Huntley, MT, until use.

Glyphosate, Glufosinate, and Dicamba Screening

To answer our first question, all herbicide screenings were conducted in a greenhouse at SARC, Huntley, MT, from July 2022 to August 2023. Seeds of B. scoparia, A. retroflexus, C. album, A. fatua, S. viridis, and E. crus-galli populations were individually sown on the surface of thermoformed trays (53 by 28 cm, 1020 Heavy duty, CN-FLHD-X1, Greenhouse Megastore, Danville, IL, USA) filled with potting mix (Pro-Mix BX^™, Premier Tech Home and Garden, Quakertown, Pennsylvania, USA). Seedlings of individual populations, upon emergence, were transplanted into plastic trays (53 by 28 cm) containing 32 subunits, with one seedling per subunit and one species per tray. In addition, seeds of B. scoparia, A. retroflexus, and C. album susceptible to glyphosate, dicamba, and glufosinate were sown and transplanted following the method described earlier. The seedlings were irrigated daily and fertilized weekly (Miracle-Gro® water-soluble fertilizer [24N-8P-16K], Scotts Miracle-Gro Products, Marysville, OH, USA). Throughout the study, the greenhouse was maintained at 25/20 ± 2 C day/night and a 16/8-h photoperiod. These conditions closely mimic the environmental characteristics of SE Montana and NW Wyoming during late May and early June, which is the usual time frame for the first herbicide application.

Individual trays were treated with either glyphosate (Roundup PowerMax®, Bayer CropScience, St Louis, MO, USA), glufosinate (Liberty® 280 SL, BASF, Research Triangle Park, NC, USA), or dicamba (Clash®, Nufarm American, Alsip, IL, USA) to assess herbicide resistance. Herbicides were uniformly applied to 5- to 10-cm-tall seedlings using a stationary spray cabinet (Research Track Sprayer, De Vries Manufacturing, Hollandale, MN, USA) calibrated to deliver 187 L ha⁻¹ through flat-fan nozzles (TeeJet® XR 8002VS, Spraying System, Wheaton, IL, USA). Given the documented existence of resistance (Hall et al. Reference Hall, Beckie, Low, Shirriff, Blackshaw, Kimmel and Neeser2014; Kumar et al. Reference Kumar, Jha and Reichard2014), glyphosate was applied at 1.3 kg ae ha^{− 1} for B. scoparia, whereas a 0.8 kg ae ha^{− 1} rate was used for A. retroflexus and C. album. Glufosinate at 0.6 kg ai ha^{− 1} and dicamba at 0.6 kg ae ha^{− 1} were applied to all dicot weed species. Monocot weed species were excluded from dicamba screening. Glyphosate applications were made at 0.9, 0.5, and 0.7 kg ae ha^{− 1} for A. fatua, S. viridis, and E. crus-galli, respectively. Glufosinate was applied at 0.6 kg ai ha^{− 1} for all monocot species. Ammonium sulfate (20 g L^{− 1}) was added as an adjuvant to glyphosate and glufosinate, and a non-ionic surfactant ( Induce®, Helena Agri-Enterprises, Collierville, TN, USA) (0.25 % v/v) + ammonium sulfate (20 g L^{− 1}) was added to dicamba. Plants were not watered for 12 h following herbicide applications to meet the required rainfast period.

Given the humidity-dependent nature of glufosinate (Coetzer et al. Reference Coetzer, Al-Khatib and Loughin2001), greenhouse relative humidity was elevated to 60% to 80% using an automated misting system (Homenote, Guangzhou, China) before spraying. Plants treated with glufosinate were maintained under elevated humidity conditions throughout the entire experimental period. Specifically, following a 12-h rainfast period, the greenhouse misting system was set to activate at 10-min intervals to maintain >60% humidity throughout the study.

The number of surviving plants (i.e., with green tissue) per population was recorded and converted into percentages to represent survival rates at 28 d after glyphosate and glufosinate applications and 35 d after dicamba application. The populations were subsequently classified into three different categories based on their survival percentages: (1) susceptible (0%), (2) developing resistance (1% to 20%), and (3) resistant (>20%) (scale modified from Owen et al. [Reference Owen, Walsh, Llewellyn and Powles2007]). All herbicide screenings were repeated three times. The screening protocol was collaboratively designed by our team following Simões Araujo et al. (Reference Simões Araujo AL Westra, Shergill and Gaines2024).

Seedling Emergence Dynamics

To address our second and third questions, a greenhouse experiment was conducted at SARC from March to May 2023 in a randomized complete block design with four replications, each consisting of 13, 11, and 4 field and margin populations of B. scoparia, A. retroflexus, and C. album, respectively. The populations were selected based on the presence of field and margin pairs across the surveyed area. The greenhouse was maintained at 22/17 ± 2 C day/night temperatures and a 16/8-h photoperiod. These conditions closely mimic the environmental characteristics of SE Montana and NW Wyoming during late May and early June, which is the usual time frame for the first herbicide application.

Twenty-five seeds of each population were manually counted and stored in paper envelopes. Thermoformed trays (53 by 28 cm, 1020 Heavy duty, CN-FLHD-X1, Greenhouse Megastore) with eight subunits were filled with the same potting mix used in the herbicide screening. Each subunit represented one replication, and each tray contained eight different populations. Twenty-five seeds per population were uniformly placed on the surface of a subunit filled with the potting mix and slightly incorporated. Seeds were not buried into the potting mix due to the surface-emerging nature of all tested species.

Seedling emergence counts were recorded daily for 40 d, and emerged seedlings were removed using forceps. Seedlings were classified as “emerged” when their plumules were visible. Daily emergence counts were aggregated to derive cumulative emergence counts, from which the cumulative emergence percentage (CE%) was calculated (Equation 1).

([1]) $${\rm{CE\% }} = \left( {{{\mathop \sum \nolimits_{i = 1}^t {E_i}}}\over{N}} \right) \times 100$$

where CE% is the cumulative emergence percentage, $\mathop \sum \nolimits_{i=1}^t {E_i}$ is the sum of the emerged seedling(s) from day 1 to day t, and N represents the total number of seeds per experimental unit (i.e., 25) at the start of the experiment. The second run was repeated 20 d after the completion of the first run. Each run was concluded upon the emergence of all seedlings (i.e., 100% emergence).

Statistical Analysis

The data were analyzed in R Studio (v. 4.2.1; R Development Core Team) (RStudio Team, 2020). For herbicide screening, survival (%) data were plotted against the populations of three major weed species. A three-parameter log-logistic regression (Equation 2) was fit to model the cumulative emergence using the drm function from the drc package (Knezevic et al. Reference Knezevic, Streibig and Ritz2007). No violations, including monotonicity and homoscedasticity, were observed in diagnostic plots; therefore, nontransformed data were used for analysis. Data were pooled across two runs, as no significant difference was observed between them. Species differed in their emergence response; therefore, the model was computed separately for each species to capture the unique characteristics of their emergence patterns. In these models, the response, predictor, and categorical variables were CE%, days after sowing (DAS), and population, respectively.

([2]) $${Y} ={ {d}\over{{1 + exp\left[ {b\left( {{\rm{log}}x - log{{\rm{E}}_{50}}} \right)} \right]}}}$$

where Y, x, b, and E₅₀ represent CE%, DAS, the slope of the regression curve (rate of emergence), and x value (DAS) required to reach 50% emergence, respectively. A smaller b value indicates a slow emergence rate. The upper limit d was fixed at 100, because the emergence of all populations was 100%. Additionally, the DAS required for 10% and 90% emergence (E₁₀ and E₉₀, respectively) was quantified with a 95% confidence interval using the ED function of the drc package. The Akaike information criterion (AIC) was used for model selection. A lack-of-fit test (P > 0.05) was performed using the modelFit function to determine whether the selected model adequately described the emergence response. The parameters estimated by the three-parameter log-logistic model were compared based on an approximate t-test using the CompParm and EDcomp functions.

To assess our second question, the relationship between phylogenetic distance and emergence patterns of A. retroflexus, B. scoparia, and C. album, a phylogenetic tree was generated using the V.PhyloMaker2, a package that uses three databases (i.e., TPL, LCVP, and WP, The Plant List, Leipzig Catalogue of Vascular Plants, and World Plants) to create large phylogenies for vascular plants (Jin and Qian Reference Jin and Qian2022). In conjunction, the cophenetic.phylo function in the ape package was used to compute pair-wise interspecific phylogenetic distance (hereafter referred to as InterspecificPD) based on the sum of branch lengths of the shortest path. E₁₀, E₅₀, and E₉₀ data of the three tested weed species were subjected to a one-way ANOVA using the aov function. Mean emergence parameters were separated across the three weed species based on Fisher’s Protected LSD test (α = 0.05). Results of one-way ANOVA and phylogenetic analysis were plotted adjacent to each other using the phytools package.

Weed Species Richness and Infestation Pattern

A total of 72 sugar beet fields were visited (Figure 1B), of which 25% were weed-free, 31% were infested by all three major weed species (B. scoparia, A. retroflexus, and C. album), while 25% and 7% had infestations of two and one of three major species, respectively (Figure 1B). Among the surveyed fields, 60%, 56%, and 38% were infested by A. retroflexus, B. scoparia, and C. album, respectively. In four fields (3%), the three other weed species, S. viridis, E. crus-galli, or A. fatua, were observed (Figure 1B).

For B. scoparia, 57.5% of the field populations were isolated, 17.5% were clustered, and 25% were widespread (Figure 1C). Amaranthus retroflexus field populations were present in isolated (37%) and clustered (30%) conditions, while 33% of populations were present in widespread conditions (Figure 1D). For C. album, 63% of the field populations occurred in isolation, 26% in clusters, and 11% in widespread conditions (Figure 1E). These categories provide useful insights into the spatial distribution of weed species, which is crucial for designing effective management strategies. The observed differences in infestation levels between species hold broader ecological and management implications, discussed in the “Management Implications of Triple-stacked Sugar Beet Systems” section. The presence of these weed species before sugar beet harvesting could be attributed to two different evolutionary adaptations: herbicide resistance and/or a temporal shift in emergence dynamics (Essman et al. Reference Essman, Loux, Lindsey, Ackley and Regnier2021), as well as application conditions such as drought stress and/or herbicide application error.

Glyphosate, Glufosinate, and Dicamba Resistance

Bassia scoparia

Out of the 16 populations found in field and margin pairs, 25% showed two-way resistance (glyphosate + dicamba), and 50% of pairs had one-way resistance to glyphosate (Figure 3). Among 24 populations found only in fields, 46% and 42% of them displayed two-way resistance (glyphosate + dicamba) and one-way resistance to glyphosate, respectively (Figure 3). Two-way resistance (glyphosate + dicamba) and one-way glyphosate resistance were observed in 17% and 33% of populations found only in margins, respectively.

Figure 3. Percentage survival of Bassia scoparia populations from (A) fields and (B) margins collected during August 2021 when sprayed with dicamba (0.6 kg ae ha^{− 1}), glufosinate (0.6 kg ae ha^{− 1}), and glyphosate (1.3 kg ae ha^{− 1}). The numerical values preceded by the letter K represent the surveyed field number. The horizontal truncated black line differentiates populations based on developing resistance (1–20%) and resistance (20%). Isolated, cluster, and widespread indicate the infestation level during the August 2021 survey, while “S” represents the susceptible populations from North Dakota used as a reference line.

Population K13 exhibited >20% survival after being sprayed with glyphosate and glufosinate (Figure 3A). To our knowledge, there is no documented evidence of glufosinate-resistant B. scoparia. Therefore, pending inheritance studies, dose-response assays, and molecular validation, this could be the first glufosinate-resistant B. scoparia case. Additionally, field populations K11, K44, K17, K20, and K23 demonstrated a low level of survival (3% to 9%) to glufosinate, suggesting that a few individuals in these populations may have resistance to glufosinate (Figure 3A), with further investigation of heritability and dose-response required.

The prevalence of glyphosate and dicamba resistance in margin populations, where herbicides are not usually applied, suggests pollen- and seed-mediated gene flow (Beckie et al. Reference Beckie, Blackshaw, Hall and Johnson2016; Jhala et al. Reference Jhala, Norsworthy, Ganie, Sosnoskie, Beckie, Mallory-Smith, Liu, Wei, Wang and Stoltenberg2021; Sarangi et al. Reference Sarangi, Tyre, Patterson, Gaines, Irmak, Knezevic, Lindquist and Jhala2017; Stallings et al. Reference Stallings, Thill, Mallory-Smith and Shafii1995). In B. scoparia, pollen-mediated gene flow can occur up to 100 m due to its highly cross-pollinated nature and protogynous flowers (Beckie et al. Reference Beckie, Blackshaw, Hall and Johnson2016). Also, seed-mediated gene flow can reach up to 1,000 m in B. scoparia, attributed to its tumbling characteristic (Beckie et al. Reference Beckie, Blackshaw, Hall and Johnson2016). In addition, glyphosate-resistant populations K02 and K55 and two-way resistant (glyphosate + dicamba) population K30 were found only in the margins (Figure 3B). Due to the existence of pollen- and seed-mediated gene flow of B. scoparia, herbicide-resistant traits present in margin plants could infest sugar beet fields over time. These results highlight the importance of managing B. scoparia in field margins to minimize the spread of resistant biotypes.

In our study, while glyphosate resistance was observed in 93% of B. scoparia populations, 40% of populations exhibited resistance to dicamba (Figure 3A). These data indicate a rapid evolutionary ability, as glyphosate and dicamba resistance was reported in ∼30% and ∼8% of B. scoparia populations, respectively, in a survey conducted across Colorado from 2012 to 2014 (Westra et al. Reference Westra, Nissen, Getts, Westra and Gaines2019). To date, dicamba resistance has been reported in the United States and Canada in six different dicot weed species, including B. scoparia (Heap Reference Heap2025). Although synthetic auxins such as dicamba have been utilized for >70 yr, resistance to this herbicide class is not as widely documented as other modes of action (e.g., acetolactate synthase, photosystem II, and 5-enolpyruvylshikimate-3-phosphate synthase inhibitors). Relative to glyphosate, the limited distribution of dicamba resistance could be attributed to the associated fitness penalty. For example, in Montana, reduced germination and 39% lower fecundity were observed in the dicamba-resistant B. scoparia compared with the dicamba-susceptible population, suggesting a potential fitness penalty, which may have restricted the spread of dicamba resistance in SE Montana and NW Wyoming (Kumar and Jha Reference Kumar and Jha2016). However, B. scoparia does not necessarily express such fitness penalties in field situations (Menalled and Smith Reference Menalled and Smith2007), suggesting that other demographic mechanisms could also be conditioning the dynamics of dicamba resistance in the region.

Amaranthus retroflexus

Despite the occurrence of A. retroflexus, no populations were identified as resistant to glyphosate, glufosinate, and dicamba (Figure 4A). However, 27% of the 44 populations showed developing resistance (1% to 20% survival) to glyphosate (Figure 4A). Out of the total populations, 7% showed one-way (dicamba) and two-way (glyphosate + dicamba) developing resistance (Figure 4A). We did not detect herbicide resistance in margin populations (n = 16) (data not shown).

Figure 4. Percentage survival of (A) Amaranthus retroflexus and (B) Chenopodium album populations collected from sugar beet fields during the August 2021 survey when sprayed with dicamba (0.6 kg ae ha^{− 1}), glufosinate (0.6 kg ae ha^{− 1}), and glyphosate (0.8 kg ae ha^{− 1}). The numerical values preceded by the letters P or L represent the surveyed field number. The horizontal truncated black line differentiates populations based on developing resistance (1–20%) and resistance (20%). Isolated, cluster, and widespread indicate the infestation level during the August 2021 survey, while “S” represents the susceptible populations from Montana used as a reference line.

Seedling Emergence Dynamics

The CE% response of B. scoparia, A. retroflexus, and C. album was adequately described by the three-parameter log-logistic model (P > 0.05) (Table 1). No consistent emergence response was observed among populations collected from fields and margins for all tested species. All B. scoparia and C. album populations reached 10% emergence (E₁₀) within 4 to 6 d (Table 1). In contrast, A. retroflexus populations required 4 to 22 d to reach 10% emergence (Table 1). Similarly, E₅₀ values for B. scoparia and C. album populations ranged from 2 to 11 d, while A. retroflexus populations exhibited a wider range, requiring 8 to 32 d (Table 1). Except for K23-field, K32-field, and K47-margin B. scoparia populations, all populations reached E₉₀ within 20 d. Similarly, C. album populations required <20 d to achieve E₉₀. In contrast, all A. retroflexus populations, except P56-field, took >25 d to reach E₉₀. These results suggest that a large proportion of A. retroflexus seedlings tend to emerge later compared with the other two weed species. The delayed and/or extended emergence response of A. retroflexus populations was also reflected by the smaller b values ranging from −2.1 to −17.1. Similar emergence patterns were also documented in giant ragweed (Ambrosia trifida L.) populations growing under the selective pressure of repeated cultivation (Davis et al. Reference Davis, Clay, Cardina, Dille, Forcella, Lindquist and Sprague2013; Hovick et al. Reference Hovick, McArdle, Harrison and Regnier2018; Schutte et al. Reference Schutte, Regnier, Harrison, Schmoll, Spokas and Forcella2008).

Table 1. Regression parameters estimated by a three-parameter log-logistic model { $Y = d/1 + {\rm{exp}}\left[ {b\left( {{\rm{log}}x - {\rm{log}}{{\rm{E}}_{50}}} \right)} \right]$ } for cumulative emergence percentage of Bassia scoparia, Amaranthus retroflexus, and Chenopodium album populations collected in southeast Montana and northwest Wyoming, USA.

^a Letters represent the common names of the weed species: K, kochia, Bassia scoparia; P, redroot pigweed, Amaranthus retroflexus; and L, common lambsquarters, Chenopodium album. The numerical values preceded by letters represent the surveyed field number; for instance, K14 was collected from field 14 and the margins of field 14.

^b b, the slope of the curve indicating the emergence rate of populations over time; E₁₀, E₅₀, and E₉₀ represent days required to reach 10%, 50%, and 90% emergence, respectively. Values in parentheses are standard errors of the mean.

Based on the phylogenetic tree (Figure 5), the InterspecificPD values between B. scoparia and C. album and B. scoparia/C. album and A. retroflexus are 80.5 and 87.3, respectively. These data suggest that B. scoparia and C. album are more closely related compared with A. retroflexus. In response to our second research question and in agreement with our second hypothesis, we detected a higher resemblance of emergence patterns between B. scoparia and C. album compared with A. retroflexus (Figure 5; Table 1), showing evidence of phylogenetic niche conservatism (i.e., a positive relationship between phylogenetic distance and functional dissimilarity). Despite the positive relationship between the phylogenetic distance and dissimilarity in emergence patterns across three tested species, the phylogenetic niche conservatism hypothesis is not ubiquitous in trait and lineage and may not be universally justified (Crisp and Cook Reference Crisp and Cook2012; E-Vojtkó et al. Reference E-Vojtkó, de Bello, Lososová and Götzenberger2023). Therefore, a comprehensive analysis exploring the relationship between various weed emergence (dis)similarities and phylogenetic relatedness would provide in-depth insights into weed management, plant systematics, and ecology.

Figure 5. Phylogenetic relatedness and (dis)similarity of emergence pattern across Bassia scoparia, Chenopodium album, and Amaranthus retroflexus. Different colored boxes represent E₁₀, E₅₀, and E₉₀, which are days required to reach 10%, 50%, and 90% emergence, respectively, estimated by a three-parameter log-logistic model at 22/17 ± 2 C day/night temperatures and a 16/8-h photoperiod. Similar letters denoting nonsignificant differences based on Fisher’s Protected LSD test (α = 0.05) are for E₁₀, E₅₀, and E₉₀ across species.

Unlike B. scoparia and C. album, herbicide resistance was not as prevalent in A. retroflexus (Figure 4A). However, there was a late-season occurrence of 43 populations, potentially reflecting the delayed and extended emergence of the A. retroflexus populations. Additionally, the temperature increases in Montana and Wyoming in late June and early July favor the rapid emergence of A. retroflexus, given its ability to germinate at a higher rate at 35/30 C (Guo and Al-Khatib Reference Guo and Al-Khatib2003). These findings address our third research question and support the third hypothesis, suggesting that a dual evolutionary adaptation—herbicide resistance and delayed/extended emergence—can enable weed species to escape early-season weed management in sugar beet cropping systems.

Most of the B. scoparia populations reached E₉₀ within 20 d, except K23-field, K32-field, and K47-margin, which required 31, 24, and 29 d, respectively (Table 1). However, they reached E₁₀ within 4 d, indicating an extended emergence period. The smaller b values associated with the K23-field (−3.9), K32-field (−3.6), and K47-margin (−3.4) further support the extended emergence of these populations. Given the prevalent herbicide resistance observed (Figure 3) and the extended emergence of these B. scoparia populations, our results support the possibility of dual evolutionary adaptation. In C. album, no population exhibited a temporal shift in emergence (Table 1). Nevertheless, 78% of C. album populations showed reduced susceptibility to glyphosate (Figure 4B), explaining the late-season detection of 27 populations.

The delayed and extended emergence of A. retroflexus and the prolonged emergence period of some B. scoparia populations observed in this study could be a form of phenological adaptation (Délye et al. Reference Délye, Menchari, Michel, Cadet and Le Corre2013; Kumar et al. Reference Kumar, Jha, Dille and Stahlman2018; Mortimer Reference Mortimer1997; Recasens et al. Reference Recasens, Calvet, Cirujeda and Conesa2005) that allows these weed species to evade the burndown herbicide applications, preplanting tillage operations, and in-crop herbicide applications. Therefore, attention should also be given to late-season weed management strategies to manage and prevent seedbank replenishment from the late-emerging cohorts (Kumar and Jha Reference Kumar and Jha2015; Wilson and Sbatella Reference Wilson and Sbatella2011). Considering the short stature of sugar beet, weed electrocution could be a potential nonchemical method for late-emerging and herbicide-resistant weeds (Schreier et al. Reference Schreier, Bish and Bradley2022; Slaven et al. Reference Slaven, Koch and Borger2023). For example, in sugar beet, Peters et al. (Reference Peters, Borgen and Lystad2020) reported 80% and 76% control of escaped waterhemp [Amaranthus tuberculatus (Moq.) Sauer] and B. scoparia using weed electrocution, respectively.

From an eco-evolutionary perspective (Neve et al. Reference Neve, Vila-Aiub and Roux2009), the extended emergence patterns of some B. scoparia populations could lead to sympatric speciation over time, a process by which a new species forms when barriers to gene flow, such as reproductive isolation, develop between groups of populations. Later-emerged B. scoparia seedlings may occupy niches at different times, indicating temporal niche divergence (Vaissi and Rezaei Reference Vaissi and Rezaei2022), which could lead to reproductive isolation (Rundle and Nosil Reference Rundle and Nosil2005). From a weed management perspective, the extended emergence pattern could potentially be exploited by adjusting the crop planting dates to promote size-asymmetric competition, that is, the unequal distribution of resources among consumer species (Weiner Reference Weiner1990) that usually occurs due to initial size advantage and overtopping (Freckleton and Watkinson Reference Freckleton and Watkinson2001). Early planting may enhance crop competitiveness, as crops gain an initial size advantage and significantly suppress the growth of later-emerged weed seedlings (Beckie et al. Reference Beckie, Blackshaw, Leeson, Stahlman, Gaines and Johnson2018).

Management Implications of Triple-stacked Sugar Beet Systems

Sustainable management of herbicide resistance relies on minimizing herbicide usage by integrating other tactics (Hawes et al., Reference Hawes, Haughton, Osborne, Roy, Clark, Perry, Rothery, Bohan, Brooks and Champion2003). In this context, the prevention of weed seedbank replenishment is a fundamental approach for effective herbicide-resistant weed management (Schwartz-Lazaro and Copes Reference Schwartz-Lazaro and Copes2019). During the survey, a large fraction of herbicide-resistant B. scoparia (57.5%) and C. album (63%) populations were present in isolation (Figure 1C and 1E). Given the greater fecundity of B. scoparia and C. album, these isolated populations could be precursors to pervasive infestation if not managed at an early stage. It would be of utmost importance for sugar beet growers to manually uproot these populations or conduct precision applications of alternative herbicides to prevent herbicide-resistant seedbank replenishment.

The clusters of B. scoparia (17.5%) and C. album (26%) (Figure 1C and 1E) could be problematic hotspots due to the accumulation of herbicide-resistant biotypes, suggesting a localized gene flow. Late-season herbicide-based rescue treatments, mechanical removal, and electric weeding could be effective management strategies to prevent the spread of resistance alleles to adjacent fields. Within the sugar beet fields, several B. scoparia populations, namely K07, K17, K23 K47, K54, K59, K68, and K69, existed in clustered and widespread conditions. Interestingly, these populations demonstrated developing resistance to dicamba with <20% survival, suggesting that only a few plants within these populations can produce dicamba-resistant progenies. Fields with widespread herbicide-resistant B. scoparia (25%) and C. album (11%) populations may exceed economic thresholds sooner compared with those with clustered and isolated populations. These fields need immediate interventions and an integrated weed management approach (e.g., crop rotations, multiple tillage operations, layered herbicide applications, and late-season rescue treatments) to ensure long-term and profitable sugar beet productions. Even though herbicide resistance is not as prevalent in A. retroflexus as in B. scoparia and C. album, fields with widespread infestation of A. retroflexus require season-long management to secure production goals due to this weed’s delayed and extended emergence.

The introduction of the glyphosate-resistant sugar beet cultivar provided short-term efficient weed control. Consequently, there has been a shift in philosophy from managing weeds based on an economic threshold (Wilkerson et al. Reference Wilkerson, Wiles and Bennett2002) to zero-tolerance for weed seedbank replenishment (Brunharo et al. Reference Brunharo, Gast, Kumar, Mallory-Smith, Tidemann and Beckie2022). Almost 100% sugar beet production in SE Montana and NW Wyoming is Roundup Ready®. Glyphosate is the predominant herbicide in the Roundup Ready® sugar beet. Other postemergent herbicides (phemedipham + desmedipham [discontinued in the United States], clopyralid, ethofumesate, and triflusulfuron) are rarely used (Peters et al. Reference Peters, Lystad, Khan and Boetel2024). These changes in weed management principles led to overreliance on glyphosate, which escalated the evolution of glyphosate-resistant B. scoparia populations (Kumar et al. Reference Kumar, Jha, Jugulam, Yadav and Stahlman2019). In accordance with our results, the majority of B. scoparia populations in the Great Plains region possess some level of glyphosate resistance (Godar et al. Reference Godar, Stahlman, Jugulam and Dille2015; Kumar et al. Reference Kumar, Jha, Jugulam, Yadav and Stahlman2019; Wiersma et al. Reference Wiersma, Gaines, Preston, Hamilton, Giacomini, Robin Buell, Leach and Westra2015). No effective postemergent options are available for B. scoparia resistant to glyphosate and triflusulfuron (Lawrence and Kniss Reference Lawrence and Kniss2021). Our study also highlights the widespread occurrence of dicamba resistance in 88% of populations of B. scoparia, including the resistance recorded at the developing stage. In the Northern Great Plains, sugar beet is typically cultivated in a 3-yr rotation cycle with corn. While dicamba is not a widely used herbicide in sugar beet, it has been one of the major herbicides used to manage glyphosate-resistant B. scoparia in corn for >30 yr (Wicks et al. Reference Wicks, Martin and Mahnken1993), resulting in the evolution of two-way (glyphosate + dicamba) resistance.

This study determined that a few B. scoparia populations displayed developing resistance to glufosinate (Figure 3), a herbicide not extensively used in the surveyed area. A low level of glufosinate resistance in B. scoparia populations suggests that glufosinate might still be effective against two-way (glyphosate + dicamba) resistant B. scoparia populations if applied as a standalone herbicide in triple-stacked sugar beet. However, previous studies have shown that glufosinate requires ≥60% relative humidity for optimum activity (Coetzer et al. Reference Coetzer, Al-Khatib and Loughin2001; Ramsey et al. Reference Ramsey, Stephenson and Hall2002). In the Northern Great Plains, where relative humidity often ranges from 25% to 35% during summer, suboptimal weed management following glufosinate application is a recurrent issue. Therefore, the long-term stewardship of glufosinate requires integrated weed management that encompasses, but is not limited to, the adoption of novel application technologies, crop rotations, preplanting tillage, herbicide diversification, weed electrocution, rescue herbicide treatment, and manual uprooting. In conclusion, widespread two-way (glyphosate + dicamba) herbicide resistance, developing glufosinate resistance in B. scoparia, and the humidity-dependent nature of glufosinate raise ecological, evolutionary, and management concerns about the long-term suitability of the triple-stacked sugar beet cultivar in semiarid regions of the Northern Great Plains.