Introduction
The Amaranthus genus consists of approximately 50 species that are native to the Americas (Kigel Reference Kigel2018). Among these, the two dioecious species, Palmer amaranth and waterhemp, are consistently ranked among the most problematic weeds in U.S. cropping systems, especially soybean (Van Wychen Reference Van Wychen2022). Palmer amaranth densities of 0.33 to10 plants m−1 of row at 8 to 12 wk after emergence have been reported to reduce soybean yield by 17% to 68%, respectively (Klingaman and Oliver Reference Klingaman and Oliver1994). Soybean yield losses of 43% were reported after 10 wk of waterhemp interference at densities of 89 to 362 plants m−2 (Hager et al. Reference Hager, Wax, Stoller and Bollero2002).
Herbicide resistance was first reported in Palmer amaranth in 1989 in South Carolina and in waterhemp in 1993 in Illinois (Heap Reference Heap2024). The evolution of herbicide-resistant biotypes of Palmer amaranth and waterhemp has contributed to an increase in soybean production costs and the need to make complicated weed management decisions. Currently, Palmer amaranth populations in the United States have been confirmed with resistance to herbicides from nine sites of action (SOAs). These include inhibitors of acetolactate synthase (ALS; Group 2), microtubule assembly (Group 3), photosystem II (PS II; Group 5), enolpyruvylshikimate-3-phosphate synthase (EPSPS; Group 9), protoporphyrinogen oxidase (PPO; Group 14), very-long-chain fatty acids (VLCFAs; Group 15), 4-hydroxyphenylpyruvate dioxygenase (HPPD; Group 27), glutamine synthetase (GS; Group 10), and synthetic auxin herbicides (SAHs; Group 4) (Heap Reference Heap2024). (Note: herbicide group numbers are assigned by the Herbicide Resistance Action Committee [HRAC] and the Weed Science Society of America [WSSA].) Waterhemp resistance in the United States has also been reported to herbicides from seven different SOAs including those that inhibit EPSPS, ALS, PS II, PPO, HPPD, VLCFAs, and SAHs (Heap Reference Heap2024).
Introduction of glyphosate resistant crops in the late 1990s propelled weed management efforts to focus on using postemergence herbicides in place of combinations of preemergence plus postemergence options, thus accelerating the evolution of glyphosate resistance (Givens et al. Reference Givens, Shaw, Johnson, Weller, Young, Wilson and Jordan2009; Powles Reference Powles2008). Oliveira et al. (Reference Oliveira, Feist, Eskelsen, Scott and Knezevic2017) reported benefits of using preemergence herbicides to control annual broadleaf and herbicide-resistant weeds, including Palmer amaranth and waterhemp. Preemergence herbicides provide weed control during the first 3 to 4 wk after crop planting but they also reduce the selection pressure on postemergence herbicides (Butts et al. Reference Butts, Miller, Pruitt, Vieira, Oliveira, Ramirez and Lindquist2017; Knezevic et al. Reference Knezevic, Pavlovic, Osipitan, Barnes, Beiermann, Oliveira and Jhala2019; Tursun et al. Reference Tursun, Datta, Sakinmaz, Kantarci, Knezevic and Chauhan2016). Additionally, effective preemergence herbicides were found to be an important strategy to control other herbicide-resistant weeds (Norsworthy et al. Reference Norsworthy, Ward, Shaw, Llewellyn, Nichols, Webster and Barrett2012). The current scenario of herbicide resistance by weeds has led to fewer postemergence weed control options, prompting growers to increase their use preemergence herbicides. Between 2006 and 2017, the hectarage of soybean fields treated with metribuzin, sulfentrazone, and S-metolachlor has increased by 16%, 21%, and 15%, respectively (USDA-NASS 2017).
Metribuzin is an asymmetric triazine herbicide. As a systemic herbicide metribuzin is readily absorbed by the roots and inhibits the flow of electrons through PS II, generating reactive oxygen species, which ultimately leads to plant death. Metribuzin was registered in the United States in 1973 and was soon adopted as a major preemergence herbicide for use in soybean production. However, use of metribuzin significantly dropped during the 1990s with total treated hectarage decreasing from 3.6 million ha in 1990 to 1.82 million ha in 1999 (US EPA 2003). Recent research has shown that metribuzin can provide good to excellent residual control of herbicide-resistant waterhemp and Palmer amaranth, including populations with metabolic resistance to atrazine (Meyer et al. Reference Meyer, Norsworthy, Young, Steckel, Bradley, Johnson, Loux, Davis, Kruger, Bararpour, Ikley, Spaunhorst and Butts2015; Vennapusa et al. Reference Vennapusa, Faleco, Vieira, Samuelson, Kruger, Werle and Jugulam2018; Vyn et al. Reference Vyn, Swanton, Weaver and Sikkema2007). Metribuzin is a common component of commercially available premixes for soybean because it can be naturally metabolized by soybean plants. However, when metribuzin is used in a premix combination, the dose (210 to 420 g ai ha-1) is frequently too low to achieve the needed duration of residual control of Amaranthus species. For soybean producers who adopt metribuzin as a preemergence herbicide, there is often concern about early season crop injury, which may potentially lead to yield reductions. A typical symptom of metribuzin injury includes interveinal leaf chlorosis that progresses to necrosis, which is primarily evident on unifoliate and first trifoliate leaves. Risk of soybean injury increases with high soil pH (>7) and/or soils with low organic matter (OM) (<1%) due to greater availability of the herbicide (Hartzler Reference Hartzler2017; Shaner Reference Shaner2014). Additionally, soybean injury also depends on variety and temperature, as cooler temperatures tends to reduce soybean emergence vigor and the plant’s ability to metabolize the herbicide (Hartzler Reference Hartzler2017).
Similar to the most soil-applied herbicides, the biologically effective dose of metribuzin is influenced by the interaction between the herbicide and other edaphic and environmental parameters that influence herbicide availability, retention, adsorption, and transport. Soybean tolerance to metribuzin is greatly influenced by metribuzin dose, soil OM, and amount of rainfall following application (Coble and Schrader Reference Coble and Schrader1973). Thus, we hypothesize that metribuzin rates greater than the current commercial premixes (210 to 420 g ai ha-1) can effectively control Palmer amaranth and waterhemp with no adverse effects on soybean growth and development. This study was designed to evaluate the entire use range of metribuzin across 13 treatments, with three major objectives: 1) determine the optimal metribuzin dose for residual control of herbicide-resistant Palmer amaranth and waterhemp under varied soil and environmental conditions; 2) evaluate early season growth and development of soybean following preemergence application of metribuzin at multiple doses; and 3) determine whether potential early season herbicide-induced injury could affect soybean yield.
Materials and Methods
Site Description
Field experiments were conducted during 2022 and 2023 in 15 states in the United States: Arkansas (Experiments AR'22 and AR'23); Illinois (IL'22, IL'23, SIL'22, and SIL'23); Indiana (IN'22 and IN'23); Iowa (IA'23); Kansas (KS'22 and KS'23); Kentucky (KY'22 and KY'23); Louisiana (LA'22 and LA'23); Michigan (MI'22 and MI'23); Mississippi (MS'22 and MS'23); Missouri (MO'22 and MO'23); Nebraska (NE'22 and NE'23); North Dakota (ND'22 and ND'23); Ohio (OH'22 and OH'23); Tennessee (TN'22 and TN'23); and Wisconsin (WI'22 and WI'23). Additional information on management practices for each site is summarized in Table 1. Soil characteristics and rainfall data for the duration of the experiment are summarized in Table 2. Site-years offered variability in soil texture, OM (1.5% to 6.6%), pH (5.5 to 8.0), interval between application and first precipitation (0 to 18 d after application of herbicide; DAA) and cumulative precipitation 42 DAA (46 to 343 mm). Naturally occurring infestations of herbicide-resistant Palmer amaranth or waterhemp at each site were evaluated for control (Table 1).
Table 1. Management practices for 32 site-years included in this study. a
a Abbreviations: PA, Palmer amaranth; WH, waterhemp.
b Resistance profile numbers represent herbicide Group numbers as designated by the Herbicide Resistance Action Committee and Weed Science Society of America.
Table 2. Soil characteristics and precipitation records across 27 site-years. a
a Abbreviation: DAA, days after application of herbicide.
Experimental Design and Treatments
Experiments were conducted in a randomized complete block design with three or four replications. Plot size was approximately 3 m by 9.1 m and differed slightly among site-years primarily due to row spacing and management practice unique to each site. Soybean were planted at 300,000 to 390,000 seeds ha-1 following preplant tillage to prepare a weed-free site at trial establishment. Tillage, planting, and herbicide application was completed within a 24-to 48-h window; therefore, all further assessments use herbicide application date as reference. A total of 17 single active–ingredient treatments were evaluated to determine the effective dose of metribuzin to control Palmer amaranth and waterhemp. Treatment information is summarized in Table 3, which includes 13 treatments of metribuzin ranging from 210 to 841 g ai ha-1 (subsequent treatments were in increments of 52 g ai ha-1), along with two comparison herbicides, sulfentrazone (420 g ai ha-1) and S-metolachlor (1,790 g ai ha-1), and nontreated and weed-free controls. Herbicides were applied within 1 d after planting using a CO2-pressurized backpack sprayer calibrated to deliver 140 L ha-1 at most locations, except at the Illinois and Michigan sites in both years (Supplementary Table S1). The application rate and nozzle used at each site are listed in Supplementary Table S1. Weed-free plots were maintained that way either by mechanical or chemical weed control options unique to each site to reduce the effects of weed interference on soybean growth and development (Supplementary Table S1).
Table 3. Herbicide treatments applied before soybean emergence.
a Trade names for metribuzin include Tricor 4F, Tricor 75DF, Metricor DF, Glory 4L, and Metribuzin 75WG, manufactured by UPL (Cary, NC), Syngenta Crop Protection (Greensboro, NC), Loveland Products (Loveland, CO). Sulfentrazone is sold under the trade name Spartan 4F by FMC (Philadelphia, PA). S-metolachlor is sold by Syngenta Crop Protection under the trade names Dual Magnum and Dual II Magnum.
Data Collection
Weed parameters, including weed control, weed density, and weed biomass, were recorded only for Palmer amaranth or waterhemp at respective sites and will collectively be referred to as Amaranthus weeds considering their physiological, morphological, and biological similarities. Weed control was visually assessed and recorded at 14, 28, and 42 DAA of herbicide. Each experimental unit was visually evaluated on a scale of 0% to 100% relative to nontreated plots, with 0% representing no weed control and 100% representing complete weed control. Weed emergence was recorded as the number of days after planting until the first emergence of Amaranthus weeds. Additionally, weed density and aboveground weed biomass per square meter were collected using four 0.25-m2 quadrants per plot at 28 DAA. Aboveground weed biomass was harvested separately from each plot in paper bags and dried in an oven at 65 C for 72 h. Soybean injury was visually assessed and recorded at 14, 28, and 42 DAA on a scale of 0% to 100% injury relative to nontreated plots, with 0% representing no injury and 100% representing complete mortality of soybean plants. Injury symptoms included physiological growth, emergence, interveinal chlorosis, and necrosis. The height of six randomly selected soybean plants per plot was measured to calculate the average soybean height. Additionally, soybean yield was recorded at maturity (Supplementary Table S2).
Statistical Analyses
Data from IL'22, IA'23, LA'23, MS'22, and MS'23 were excluded from further analysis due to technical errors or experimental failure, making the data insufficient for drawing any conclusions. All data were processed and analyzed using R Studio software with respective packages (R Core Team Reference Core Team2021). Soil moisture at the time of application was a categorical variable reported in seven levels (very dry, dry, fair, adequate, good, damp, and moist) and so were reduced to three levels: dry for very dry and dry site-years; fair for fair, adequate, and good site-years; and moist for damp and moist site-years (Table 2). Soybean planting date was transformed to day of the year format to maintain consistency. Weather data collected from the respective weather stations were used to calculate precipitation values. Based on research by Meyer (Reference Meyer2023), who identified 12.7 mm of precipitation within the first 2 wk after application as being critical for optimizing herbicide performance, this value was used as a threshold for further analysis. A principal component analysis (PCA) was conducted using the prcomp function to evaluate the differences across site-years (Venables and Ripley Reference Venables and Ripley2002). The following variables were included for calculating eigenvectors: soil texture (sand, silt, and clay %); soil OM; soil temperature at the time of herbicide application; soil pH; interval between application and first precipitation; amount of first precipitation; days between application and cumulative 12.7 mm precipitation; and cumulative precipitation at 42 DAA. Following PCA, a random forest approach was used to predict the importance score of covariates determining weed control using the randomForest package (Liaw and Wiener Reference Liaw and Wiener2002). Weed control was selected as the predicted variable for the random forest model and was evaluated as a linear function of all the variables used for PCA in addition to metribuzin dose, weed control days after application, seeding rate, row spacing, soil moisture at application, and planting date.
Weed control was analyzed using the generalized additive model (GAM) from the mgcv package specifying a beta regression family with a probit link to account for the bounded nature of percent control data (Wood Reference Wood2011). Weed control was evaluated as a function of metribuzin dose, soil OM, soil pH, clay content, interval between application and first precipitation, amount of first precipitation, days between application, cumulative 12.7 mm precipitation, cumulative precipitation at 42 DAA, soil temperature, and soil moisture at application. A smooth term was included for metribuzin dose to capture the potential nonlinearity in the dose-response relationship. Weed density and weed biomass data were analyzed separately using GAM as a function of metribuzin dose. Crop injury data were analyzed separately at 14, 28, and 42 DAA using a GAM with crop injury as a function of metribuzin dose. Predicted soybean response was then plotted against the metribuzin dose. The emmeans package was used to estimate marginal means for soybean height and yield data across treatments (Lenth Reference Lenth2023). The estimated means were then subjected to pairwise comparison using Tukey’s honestly significant difference (HSD) test (α = 0.05). Additionally, weed control with sulfentrazone and S-metolachlor was compared with that of 525 g ai ha-1 of metribuzin by subjecting the estimated means to pairwise comparisons using Tukey’s HSD (α = 0.05). Crop injury and all weed parameters except weed control were evaluated separately for sulfentrazone and S-metolachlor treatments similar to soybean height and yield as outlined previously.
Results and Discussion
Principal Component and Random Forest Analyses
A PCA biplot was developed to summarize the geographical differences in terms of soil and precipitation parameters. Dimensions 1 and 2 combined explain 48.4% variability among site-years. In the PCA biplot the IL'23 and MI'23 site-years were distantly separated from other sites along the principal components, primarily driven by interval between application and first precipitation and days between application and cumulative 12.7 mm precipitation (Figure 1). The IL'23 experiment received its first precipitation 18 DAA and required 23 d to accumulate 12.7 mm of precipitation, whereas the MI'23 experimental plots received their first precipitation 19 DAA and required 32 d to accumulate 12.7 mm of precipitation (Table 2). Data from IL'23 and MI'23 were analyzed separately from other site-years due to the delayed timing of precipitation following herbicide application. Precipitation within the first 2 wk of herbicide application is an important factor for herbicide incorporation and subsequent weed control (Landau et al. Reference Landau, Hager, Tranel, Davis, Martin and Williams2021; Meyer Reference Meyer2023). Because rain was substantially delayed at these two sites, they were categorized as representing a delayed precipitation condition. In contrast, the data from all site-years, except IL'23 and MI'23, were clustered together and labeled as having optimum precipitation conditions (Figure 1). This classification was based solely on the time required to accumulate 12.7 mm of precipitation, rather than total rainfall. The optimum precipitation condition subset was used for all further analyses. In contrast, the delayed precipitation condition subset was analyzed only for weed control due to its limited variability and small data set size.
Figure 1. Principal component biplot for soil texture (sand, silt and clay), soil organic matter (OM), soil temperature at herbicide application, soil moisture, soil pH, precipitation interval after herbicide application (interval of first precipitation), amount of first precipitation, interval for cumulative 12.7 mm precipitation, and cumulative precipitation 42 d after application for all site-years. Site-years are represented by state location followed by experimental year: Arkansas (AR'22, AR'23), Illinois (IL'23, SIL'22, SIL'23), Indiana (IN'22, IN'23), Kansas (KS'22, KS'23), Kentucky (KY'22, KY'23), Louisiana (LA'22), Michigan (MI'22, MI'23), Missouri (MO'22, MO'23), Nebraska (NE'22, NE'23), North Dakota (ND'22, ND'23), Ohio (OH'22, OH'23), Tennessee (TN'22, TN'23) and Wisconsin (WI'22, WI'23).
In the context of random forest analysis, a variable importance plot quantifies the contribution of each variable to the overall predictive accuracy of the model. As expected, metribuzin dose was the largest factor in predicting weed control followed by weed control DAA (Figure 2). Weed control days after application is an indication of how long into the growing season weed control is achieved (i.e., 14, 28, or 42 DAA). Cumulative precipitation and amount of first precipitation were the two biggest co-variates among the precipitation parameters (Figure 2). Soil OM, soil texture, and soil pH were the major co-variates among the soil parameters. Seeding rate and row spacing had a minimal effect on predicting weed control. Therefore, these two variables were excluded from further analysis (Figure 2). Previous research suggests that seeding rate and row spacing have inconsistent effects on weed control in soybean, although some studies reported reduced weed density and increased weed control with higher seeding rates of soybean (Arce et al. Reference Arce, Pedersen and Hartzler2009; Place et al. Reference Place, Reberg-Horton, Dunphy and Smith2009). A meta-analysis revealed that narrow rows reduced weed biomass by 71%, yet 36% of soybean trials showed no significant weed suppression (Singh et al. Reference Singh, Thapa, Singh, Mirsky, Acharya and Jhala2023).
Figure 2. Variable importance plot for covariates determining weed control based on random forest model for all site-years except Illinois 2023 and Michigan 2023. Abbreviations: DAA, days after application of herbicide; OM, organic matter.
Weed Control
Optimum Precipitation Condition. Amaranthus weed control was assessed using a GAM, and all the co-variates had a significant effect in predicting weed control except soil pH and clay content (Table 4). This is likely due to a lack of variability in soil pH and clay content among site-years. The smooth term for dose was highly significant (effective degrees of freedom [edf] = 3.76, P < 0.001), indicating a flexible, nonlinear relationship. Derivatives of the smooth function revealed that the rate of increase in control was highest at low doses and plateaus at higher doses. Weed control was directly related to metribuzin dose. Effectiveness of metribuzin to control weeds decreased as time progresses in the growing season, with the overall greatest control achieved at 14 DAA and the least control at 42 DAA (Figure 3). At 14 DAA, 473 g ai ha-1 of metribuzin provided 95% weed control with the maximum weed control of 98% achieved with 841 g ai ha-1 of metribuzin (Figure 3). At 28 DAA, weed control was <95% for all metribuzin doses. However, 90% weed control was achieved using 630 g ai ha-1 metribuzin with a marginal increase in weed control at greater doses of metribuzin at 28 DAA (Figure 3). For the lower doses of metribuzin (210 to 420 g ai ha-1) weed control decreased at 42 DAA compared with control at 14 and 28 DAA. Eighty percent and 85% weed control was achieved with 578 and 683 g ai ha-1, respectively, at 42 DAA. Weed control gradually increased with an increasing dose of metribuzin at 42 DAA with a maximum control of 89% achieved when 841 g ai ha-1 was applied (Figure 3).
Table 4. Estimated parameter values of the generalized additive model for weed control as a function of predicting variables for all site-years except Illinois 2023 and Michigan 2023. a
a Adjusted R 2 = 0.41; deviance explained = 59%.
Figure 3. Soybean injury and Amaranthus weed control across metribuzin doses at 14, 28, and 42 d after application (DAA) for all site-years except Illinois 2023 and Michigan 2023. The shaded area around the regression line indicates the 95% confidence interval. The dotted black line represents 5% crop injury levels, and the dotted purple lines represent 100%, 95%, 90%, and 85% weed control, respectively. Each dot represents an individual weed control data point across treatments.
Metribuzin (525 g ai ha-1) and sulfentrazone (420 g ai ha-1) consistently provided similar weed control at 14, 28, and 42 DAA. In contrast, S-metolachlor (1,790 g ai ha-1) resulted in significantly lower weed control at all evaluation times. At 14 DAA metribuzin and sulfentrazone provided >92% weed control compared with 88% control for S-metolachlor. Although metribuzin and sulfentrazone provided >78% weed control at 42 DAA compared with 59% control achieved with S-metolachlor (Table 5).
Table 5. Percent weed control estimates as a function of herbicide treatment for all site-years except Illinois 2023 and Michigan 2023.a,b
a Abbreviation: DAA, days after application of herbicide.
b Means within a column followed by lowercase letters represent significant differences identified by separation of means for each interval using Tukey’s honest significant difference test (α = 0.05).
c Weed control estimates as a result of metribuzin treatment were evaluated when the herbicide was applied at 525 g ai ha-1.
Delayed Precipitation Condition. Control of Amaranthus weed species was analyzed separately for IL'23 and MI'23 using a GAM. Metribuzin dose, weed control DAA, soil temperature, days between application, and cumulative 12.7 mm precipitation had a significant effect in predicting weed control. In previous research, soil temperatures were found to affect early season weed control only when rainfall was scant during the first 15 DAA (Landau et al. Reference Landau, Hager, Tranel, Davis, Martin and Williams2021). The smooth term for dose was highly significant (edf = 1.92, P < 0.001), indicating a flexible, nonlinear relationship. Weed control at 14 DAA was likely affected due to dry and fair soil moisture at the time of herbicide application followed by no precipitation up to 19 and 32 DAA. Weed control increased with an increasing dose of metribuzin. The effectiveness of metribuzin to control weeds decreased over time with overall greatest control occurring at 14 DAA with a metribuzin dose of 841 g ai ha-1. At 14 DAA, 630 g ai ha-1 of metribuzin provided 80% weed control with a maximum weed control of 91% achieved with 841 g ai ha-1 of metribuzin. While 80% weed control was achieved at 28 DAA using 578 g ai ha-1 metribuzin, this increased to 88% when 841 g ai ha-1 metribuzin was applied. Weed control at 14 and 28 DAA was comparable with applications of 525 to 841 g ai ha-1 metribuzin, while weed control decreased later into the growing season at 42 DAA. A maximum weed control of 76% was achieved at 841 g ai ha-1 42 DAA (Figure 4).
Figure 4. Amaranthus weed control across metribuzin doses at 14, 28, and 42 d after application (DAA) at the Illinois 2023 and Michigan 2023 site-years. The shaded area around the regression line indicates the 95% confidence interval. The dotted purple lines represent 100%, 90%, and 80% weed control, respectively. Each dot represents an individual weed control data point across treatments.
At the IL'23 and MI'23 site-years, metribuzin applied at 525 g ai ha-1 provided significantly greater (82%) weed control than either sulfentrazone applied at 420 g ai ha-1, which provided 30% control, or S-metolachlor at 1,790 g ai ha-1, which provided 33% when assessed at 14 DAA (Table 6). However, by 28 and 42 DAA, no significant differences were observed among treatments. Weed control at 28 DAA ranged from 79% to 84%, and decreased slightly across treatments by 42 DAA, ranging from 64% to 74% (Table 6).
Table 6. Percent weed control estimates as a function of herbicide treatment for Illinois 2023 and Michigan 2023 site-yearsa,b.
a Abbreviation: DAA, days after application of herbicide.
b Means within a column followed by lowercase letters represent significant differences identified by separation of means for each interval using Tukey’s honest significant difference test (α = 0.05).
c Weed control estimates as a result of metribuzin treatment were evaluated when the herbicide was applied at 525 g ai ha-1.
Amaranthus Weed Emergence . Weed emergence was recorded for 9 site-years (Supplementary Table S2). Weeds emerged as early as 6 DAA in some nontreated plots. An overall delayed weed emergence for the site-years in Missouri and Ohio was observed (data not shown). Sulfentrazone and S-metolachlor applications resulted in delayed weed emergence at 27 and 21 DAA, respectively (Figure 5). Metribuzin doses of 630 to 841 g ai ha-1 delayed weed emergence longer than sulfentrazone and S-metolachlor, whereas 841 g ai ha-1 metribuzin delayed emergence to an average of 33 DAA.
Figure 5. Amaranthus weed emergence from 9 site-years across metribuzin doses. The regression line represents weed emergence as a function of metribuzin dose. The shaded area around the line indicates the 95% confidence interval. The dashed lines represent mean weed emergence for sulfentrazone (S; 27 d after planting), and S-metolachlor (M; 20 d after planting). Each dot represents an individual weed emergence data collection point.
Amaranthus Weed Density and Biomass . Weed density was recorded for 20 site-years at 28 DAA (Supplementary Table S2). The lowest weed density was observed at ND'23, with an average of 9 plants m−2, and the greatest density was observed at MO'23 where approximately 1,300 plants m−2 were counted in the nontreated plot. We observed the lowest weed densities when 525 to 841 g ai ha-1 of metribuzin were applied, which was similar when sulfentrazone was applied (Figure 6). Weed density after S-metolachlor treatment was similar to that when lower doses of metribuzin (210 to 315 g ai ha-1) were applied (Figure 6).
Figure 6. Amaranthus weed density from 20 site-years across metribuzin doses 28 d after application. The regression line represents weed density as a function of metribuzin dose. The shaded area around the line indicates the 95% confidence interval. The dashed lines represent mean weed density for sulfentrazone (S; 21 plants) and S-metolachlor (M; 66 plants). Each dot represents an individual weed density data collection point.
Weed biomass was recorded for 16 site-years (Supplementary Table S2). Nontreated KS'23 and MO'23 plots accumulated the greatest weed biomass. We observed less weed biomass at 28 DAA when 525 to 841 g ai ha-1 of metribuzin was applied (Figure 7), whereas similar mean weed biomass measurements of 6 and 9.4 g per m−2 were recorded after applications of sulfentrazone and S-metolachlor, respectively, when assessed at 28 DAA.
Figure 7. Amaranthus weed biomass from 16 site-years across metribuzin doses 28 d after application. The regression line represents weed biomass as a function of metribuzin dose. The shaded area around the line indicates the 95% confidence interval. The dashed lines represent mean weed biomass for sulfentrazone (S; 6 g) and S-metolachlor (M; 9 g). Each dot represents an individual weed biomass data collection point.
In this research, effective control of Amaranthus weeds occurred when precipitation conditions were optimal, particularly when higher doses of metribuzin (578 to 841 g ai ha-1) were used. A maximum control of 98% was achieved with applications of 841 g metribuzin ha-1; when assessed at 14 DAA. The efficacy of metribuzin declined later in the growing season. At 28 DAA, 90% control was achieved with 630 g ai ha-1; of metribuzin. Meyer et al. (Reference Meyer, Norsworthy, Young, Steckel, Bradley, Johnson, Loux, Davis, Kruger, Bararpour, Ikley, Spaunhorst and Butts2016) reported 69% control of Palmer amaranth 4 wk after treatment with 420 g ai ha-1 of metribuzin, which was less than that when 1,068 g ai ha-1 of S-metolachlor was applied, which provided 89% control in a coarse-textured soil with little OM. However, both S-metolachlor and metribuzin treatments provided similar reductions in Palmer amaranth and waterhemp density (Meyer et al. Reference Meyer, Norsworthy, Young, Steckel, Bradley, Johnson, Loux, Davis, Kruger, Bararpour, Ikley, Spaunhorst and Butts2016). In our study, by 42 DAA the effectiveness of metribuzin decreased, especially at lower doses of 210 to 420 g ai ha⁻1. However, the highest dose of 841 g ai ha⁻1; still provided 89% control. In contrast, S-metolachlor exhibited a sharp decline in efficacy, dropping to 59% control at 42 DAA. Additionally, results suggest that metribuzin (525 g ai ha-1) and sulfentrazone (420 g ai ha-1) offer superior residual Amaranthus control compared to S-metolachlor (1,790 g ai ha-1). However, previous research suggests that sulfentrazone (280 g ai ha-1) and S-metolachlor (1,787 g ai ha-1) provided better residual control of Palmer amaranth than metribuzin (563 g ai ha-1) (Ribeiro et al Reference Ribeiro, Oliveira, Smith, Santos and Werle2021b). That conclusion was based on the greenhouse bioassays conducted on silty loam soils at a depth of 0 to 10 cm with OM ranging from 2.6% to 3.1% and pH of 6.5 and 7 (Ribeiro et al Reference Ribeiro, Oliveira, Smith, Santos and Werle2021b).
When precipitation was delayed, we observed that metribuzin was more effective, potentially due to its higher solubility, whereas sulfentrazone and S-metolachlor exhibit reduced early season activity until adequate the soil was adequately moist. The solubility and soil half-life of these herbicides influence their water requirements for activity and persistence in the soil. Metribuzin, with its high solubility (1,100 mg L−1 at 20 C) and moderate soil half-life (30 to 60 d) requires less initial moisture for activity and can quickly move into the weed seed zone (Shaner Reference Shaner2014). However, its mobility also increases the risk of leaching when rain is excessive. S-metolachlor, with lower solubility (488 mg L−1 at 20 C) and a shorter soil half-life (15 to 50 d), requires more consistent and timely precipitation for optimal activity and is less prone to leaching (Shaner Reference Shaner2014). Sulfentrazone, with moderate solubility (780 mg L−1 at pH 7) and a long soil half-life (121 to 302 d), is more persistent but may require higher moisture levels for activity, making its efficacy more dependent on sufficient and timely precipitation (Shaner Reference Shaner2014). Landau et al. (Reference Landau, Hager, Tranel, Davis, Martin and Williams2021) also reported that the probability of effective control of three annual weed species increased as rainfall increased to a threshold of 10 mm. Additionally, herbicide combinations achieved maximum efficacy when rainfall was low (Landau et al. Reference Landau, Hager, Tranel, Davis, Martin and Williams2021).
Furthermore, this study highlights the efficacy of metribuzin to control herbicide-resistant Amaranthus weeds. Palmer amaranth and waterhemp populations at experimental sites were resistant to multiple herbicides (mostly to Group 2 and Group 9 herbicides) as listed in Table 1. Specifically, populations of Palmer amaranth and waterhemp in plots in Illinois, Indiana, and North Dakota were resistant to postemergence-applied Group 5 herbicides (potentially atrazine), yet weeds were effectively controlled with metribuzin. This suggests that metribuzin remains an effective herbicide for managing multiple herbicide–resistant populations of Palmer amaranth and waterhemp.
Soybean Response
Crop injury increased with an increasing dose of metribuzin with predicted injury of no more than 5% even at the highest dose of metribuzin (841 g ai ha-1; Figure 3). Crop injury as high as 10% was recorded at LA'22, MI'22, OH'22, and TN'23 at 14 DAA, which recovered at 28 and 42 DAA (<5%; data not shown); whereas injury ranging up to 20% was reported at 42 DAA at the AR'22 and LA'22 plots (data not shown). There was no injury with applications of either sulfentrazone (420 g ai ha-1) or S-metolachlor (1,790 g ai ha-1; Table 7).
Table 7. Percent crop injury estimated as a function of herbicide treatment for all site-years except Illinois 2023 and Michigan 2023a,b.
a Abbreviation: DAA, days after application of herbicide.
b Means within a column followed by lowercase letters represent significant differences identified by separation of means for each interval using Tukey’s honest significant difference test (α = 0.05).
Soybean Height . Soybean height was recorded for 20 site-years at 28 DAA (Supplementary Table S2). Soybean growth varied across site-years depending on the relative weather conditions and soybean cultivar. Overall, there was no difference in soybean height among all treatments, especially comparing the nontreated and weed-free plots with those that received herbicide treatments, suggesting that herbicides had no effect on soybean height (Figure 8).
Figure 8. Soybean height from 20 site-years across herbicide treatment 28 d after application. Each dot represents an average soybean height data point. The boxes represent the 25th to 75th percentiles of interquartile ranges with the horizontal line inside each box indicating the median yield. The whiskers extend to the smallest and largest values within 1.5 times the interquartile range. Abbreviations: M, S-metolachlor; ns, no significant differences (identified by separation of means using Tukey’s honestly significant difference test; α = 0.05); NT, nontreated; S, sulfentrazone; WF, weed-free.
Soybean Yield . Soybean yield was recorded for 7 site-years at crop maturity (Supplementary Table S2). A minimum yield was recorded at KS'23, averaging 531 kg ha-1 from nontreated plots. A maximum yield was recorded from weed-free plots in NE'23, averaging 4,300 kg ha-1. Despite the absence of postemergence herbicide applications, preemergence herbicide treatments resulted in soybean yields that were similar to those observed in the weed-free control (Figure 9). This is likely due to reduced competitiveness of the weeds that survived preemergence herbicides. Previously, Adcock and Banks (Reference Adcock and Banks1991) reported reduced weed competition and lowered water use after an application of 0.4 kg ha of metribuzin. Soybean yield following metribuzin application at 315 to 841 g ai ha-1 and sulfentrazone applications were similar to yield from the weed-free plots. Lower yields were observed from plots that received S-metolachlor treatment and 210 g ai ha-1 of metribuzin (Figure 9).
Figure 9. Soybean yield from 7 site-years plotted with herbicide treatment. Lowercase letters represent significant differences identified by separation of means using Tukey’s honestly significant difference test (α = 0.05). Each dot represents an individual soybean yield data point. The boxes represent 25th to 75th percentiles of interquartile ranges with the horizontal line inside each box indicating the median yield. The whiskers extend to the smallest and largest values within 1.5 times the interquartile range. Abbreviations: M, S-metolachlor; NT, nontreated; S, sulfentrazone; WF, weed-free.
There was minimal soybean injury resulting from the application of metribuzin, even at the highest dose of 841 g ai ha-1. The variability in injury at the OH'22, TN'23, AR'22, and LA'22 plots could be attributed to low OM and heavy rainfall following herbicide application, as herbicide adsorption is known to be strongly correlated with soil OM and texture (Blumhorst et al. Reference Blumhorst, Weber and Swain1990). Additionally, extended periods of cool, wet soil conditions during crop emergence are known to reduce soybean’s ability to metabolize preemergence herbicides, potentially leading to soybean injury (Moomaw and Martin Reference Moomaw and Martin1978; Osborne et al. Reference Osborne, Shaw and Ratliff1995). Soybean plants, due to their indeterminate growth habit, are known to compensate for herbicide injury occurring during early developmental stages (Cox and Cherney Reference Cox and Cherney2011, Ribeiro et al Reference Ribeiro, Maia, Arneson, Oliveira, Read, Ané and Werle2021a, Weidenhamer et al. Reference Weidenhamer, Triplett and Sobotka1989). This might be why we observed no negative effects on soybean height and yield in our study following metribuzin application. In previous studies, metribuzin (560 g ai ha-1) and sulfentrazone (280 g ai ha-1) applied preemergence did not reduce soybean yield on loam soil texture with soil OM ranging from 1.7% to 2.2% and pH from 6.7 to 7.5 (Arsenijevic et al Reference Arsenijevic, de Avellar, Butts, Arneson and Werle2021). However, sulfentrazone resulted in a 22% reduction in green canopy at the V2 growth stage and a 10% reduction in plant stand at maturity compared with a nontreated control. Despite these effects, overall soybean yield was increased by 3% with metribuzin and sulfentrazone treatments, and this can be attributed to a higher number of seeds produced per plant. Interestingly, in our study, sulfentrazone and S-metolachlor did not cause concerning crop injury levels. This contrasts with findings by Taylor-Lovell et al. (Reference Taylor-Lovell, Wax and Nelson2001), who reported up to 61% soybean injury across 15 varieties following sulfentrazone (446 g ai ha-1) application, with greater injury observed under prolonged wet and cool conditions after planting. Previous research has reported differential tolerance of soybean varieties to preemergence herbicides (Taylor-Lovell et al. Reference Taylor-Lovell, Wax and Nelson2001); however, this was beyond the scope of our study, and necessary comparisons were not included in the experimental design.
Practical Implications
Weed control remains a top priority for soybean producers throughout the United States. The challenge is accompanied by accelerating evolution of herbicide-resistant weeds. This multistate study demonstrates that metribuzin, a long-established soil-residual herbicide, remains a viable option for residual control of Palmer amaranth and waterhemp. Results suggest that metribuzin can be safely applied at higher rates than those commonly included in commercial premixes, particularly in optimum precipitation conditions. Doses ranging from 578 to 841 g ai ha-1 can be judiciously incorporated into herbicide rotation strategies for effective control of Amaranthus weeds. Even under delayed precipitation conditions, metribuzin retained activity better than rates used for sulfentrazone (420g ai ha-1) and S-metolachlor (1,790g ai ha-1), likely due to metribuzin’s higher solubility. The study also confirms effectiveness of metribuzin on weed populations that have become resistant to multiple herbicide SOAs, including PS II inhibitors such as atrazine.
Preemergence herbicides have regained importance for building effective weed management strategies. The weed control efficacy of preemergence herbicides likely outweighs the concerns associated with early season soybean injury. Incorporating higher rates of metribuzin into preemergence herbicide program can improve early season weed control, delay the critical period of weed control, and reduce the selection pressure on postemergence herbicides.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/wet.2025.10047.
Acknowledgments
We thank all academic staff and students involved in establishing and collecting data from this project and Mr. Aiden Kerns for his assistance with statistical analyses.
Funding
This project was funded by United Soybean Board.
Competing Interests
The authors declare they have no competing interests.
Open access
