Introduction
The objective of POST herbicide application is to deliver the proper amount of spray solution to the leaf surface of targeted plant species to obtain optimum levels of control (Ennis and Williamson Reference Ennis and Williamson1963). Ideally, the amount of solution applied by the sprayer should provide uniform spray deposition across the target, maximizing the amount of herbicide available for uptake (Shaw et al. Reference Shaw, Morris, Webster and Smith2000). Proper spray droplet size could secure herbicide efficacy through optimizing the chemical deposition on the plant surface. Previous research has demonstrated that spray application is effective but in several cases inefficient (Beyer Reference Beyer1991; Knoche Reference Knoche1994). Normally, only a small fraction of the total active ingredient being applied is responsible for the desired plant response (Pimentel Reference Pimentel1995). During POST application of pesticides, most of the spray solution is either intercepted by crop canopy or deposited on the soil surface. Additionally, a small percentage may be susceptible to off-target movement (Knoche Reference Knoche1994; Matthews Reference Matthews2008). Therefore, application efficiency to reduce the amount of pesticide needed for optimal control would be economically and ecologically beneficial.
Several factors affect deposition and retention of pesticide spray droplets—all of which can influence the effectiveness of a given herbicide: (1) meteorological factors such as wind speed, air temperature and humidity, and atmospheric stability; (2) application factors such as sprayer and nozzle type, orifice size, application pressure, spray boom height, driving speed, and spray exit angle; and (3) chemical formulation (Carlsen et al. Reference Carlsen, Spliid and Svensmark2006). Previous research has demonstrated that among these factors, droplet size is critical to spray deposition and drift (Taylor et al. Reference Taylor, Womac, Miller and Taylor2004; Whisenant et al. Reference Whisenant, Bouse, Crane and Bovey1993; Yates et al. Reference Yates, Akesson and Bayer1976).
The droplet spectra of pesticides are composed of spray droplets with various sizes and are characterized by the volume median diameter (VMD) of the spray solution (Meyer et al. Reference Meyer, Norsworthy, Kruger and Barber2015). The VMD represents the median droplet size where half of the spray volume is composed of spray droplets with diameters that are smaller, and the other half with diameters larger than the median, and typically classified by their diameter in microns (μm) (Creech et al. Reference Creech, Henry, Fritz and Kruger2015; Meyer et al. Reference Meyer, Norsworthy, Kruger and Barber2015). The number of spray nozzles used in agriculture is vast. According to Bouse et al. (Reference Bouse, Kirk and Bode1990), the different nozzle designs used in agriculture produce spray droplets ranging from 10 (Extremely Fine) to 1,000 μm (Ultra Coarse). Therefore, comprehending the impact of environmental factors on different droplet sizes is essential to understanding product efficacy but also off-target movement. Previous research demonstrated that a 100-μm (Very Fine) droplet may travel 7.5 times farther off-target compared to a 500-μm (Very Coarse) droplet given a 5-km h–1 wind speed (Bode Reference Bode, McWhorter and Gebhardt1987; Creech et al. Reference Creech, Henry, Fritz and Kruger2015).
Typically, nozzles that produce smaller spray droplets increase the efficacy of nonsystemic (contact) herbicides. Atomization of spray solution into smaller droplets results in greater coverage of the target tissue, thus maximizing herbicide activity (Ennis and Williamson Reference Ennis and Williamson1963; Lake Reference Lake1977; McKinlay et al. Reference McKinlay, Brandt, Morse and Ashford1972). Rogers and Maki (Reference Rogers and Maki1986) reported that smaller spray droplets provide greater spray deposition when compared to larger droplets. Furthermore, larger spray droplets have greater kinetic energy and velocity, which decreases adhesion and increases droplet bouncing off the leaf surface and shattering (Spillman Reference Spillman1984). Research conducted by Shaw et al. (Reference Shaw, Morris, Webster and Smith2000) indicated that acifluorfen applied with 250-μm (Medium) spray droplets provided the greatest control of common cocklebur (Xanthium strumarium L.). Additionally, reduced glufosinate and paraquat control of broadleaf signalgrass [Urochloa platyphylla (Munro ex C. Wright) R.D. Webster] and common cocklebur has been observed with increased spray droplet sizes (Etheridge et al. Reference Etheridge, Hart, Hayes and Mueller2001). In contrast, Sikkema et al. (Reference Sikkema, Brown, Shropshire, Spieser and Soltani2008) and Berger et al. (Reference Berger, Dobrow, Ferrel and Webster2014) observed similar control of common lambsquarters (Chenopodium album L.) and Palmer amaranth following fomesafen and lactofen application using different spray droplet sizes. The convoluted results found in the literature regarding efficacy of contact herbicides could be attributed to specific relationships between biotic and abiotic factors such as plant species, population genetics and density, climate, and soil type. Chachalis et al. (Reference Chachalis, Reddy, Elmore and Steele2001) reported that leaf structure differences were responsible for higher droplet contact angle following acifluorfen application on ivyleaf morningglory [Ipomoea hederacea (L.) Jacq.] compared to pitted morningglory (Ipomoea lacunosa L.) and smallflower morningglory [Jacquemontia tamnifolia (L.) Griseb.].
The development of digital application technology has allowed the implementation of pulse width modulation (PWM) systems into agricultural sprayers (Bode and Bretthauer Reference Bode and Bretthauer2007). PWM sprayers increase application flexibility by delivering a specific droplet size distribution at different driving speeds while variably controlling flow (Butts et al. Reference Butts, Samples, Franca, Dodds, Reynolds, Adams, Zollinger, Howatt, Fritz, Hoffmann and Kruger2018, Reference Butts, Samples, Franca, Dodds, Reynolds, Adams, Zollinger, Howatt, Fritz, Hoffmann, Luck and Kruger2019a, Reference Butts, Butts, Luck, Fritz, Hoffmann and Kruger2019b). In PWM systems, each spray nozzle is equipped with an electronically powered solenoid that typically pulses on a frequency of 10 pulses s–1 (Giles and Comino Reference Giles and Comino1989). The relative proportion of time each valve remains open, called the duty cycle, allows for variable flow rate. In comparison to conventional spray systems, PWM sprayers provide flow rate changes without altering application pressure and nozzle type, allowing the operator to make applications at different speeds while maintaining a constant droplet size distribution and carrier volume (Anglung and Ayers Reference Anglung and Ayers2003). Previous research has shown that PWM duty cycle has minimal to no effect on spray droplet size when using non-Venturi nozzles (Butts et al. Reference Butts, Butts, Luck, Fritz, Hoffmann and Kruger2019b; Giles et al. Reference Giles, Henderson, Funk, Robert, Rust and Larson1996). In addition, research has demonstrated that when operated at or above 40% duty cycles, PWM sprayers do not affect spray pattern and coverage (Mangus et al. Reference Mangus, Sharda, Engelhardt, Flippo, Strasser, Luck and Griffin2017). Therefore, PWM sprayers are a reliable system to perform spray applications where a desired spray droplet size distribution is held constant.
The development and dissemination of Palmer amaranth populations resistant to multiple herbicide modes of action has complicated control practices in many areas of the United States. Palmer amaranth populations resistant to at least one of the following eight herbicide sites of action have been documented: 5-enolpyruvylshikimate-3-phosphate synthase (EPSP) inhibitors, acetolactate synthase (ALS) or acetohydroxyacid synthase (AHAS) inhibitors, photosystem II (PSII) inhibitors, synthetic auxins, 4-hydroxyphenylpyruvate dioxygenase (HPPD) inhibitors, microtubule inhibitors, very-long-chain fatty acid (VLCFA) inhibitors, and protoporphyrinogen oxidase (PPO) inhibitors (Heap Reference Heap2019). Biotypes resistant to PPO-inhibiting herbicides were first documented in Arkansas in 2011. Consequently, populations resistant to acifluorfen, lactofen, and fomesafen were reported in 2016 (Heap Reference Heap2019). Research has shown that Palmer amaranth populations from fields located in the northern Mississippi Delta region are likely to be infested with biotypes resistant to PPO-inhibiting herbicides (Bond et al. Reference Bond, Reynolds and Irby2016). This scenario has raised concern, as PPO-inhibiting herbicides such as acifluorfen are important POST herbicides for growers to control Palmer amaranth and other broadleaf species in soybean, peanut, and rice (Sweat et al. Reference Sweat, Horak, Peterson, Lloyd and Boyer1998). Given the rapid spread of PPO-resistant Palmer amaranth, cost-effective means of application are needed that maximize acifluorfen effectiveness for Palmer amaranth control and mitigate spray drift. Nonlinear regression methods such as generalized additive models (GAMs) are an attractive alternative to evaluate weed control response to spray droplet size. These models provide more data flexibility in suggesting the form of the function instead of imposing a specific function type on the data, an advantage that could be utilized to generate models for weed control (Rosenheim and Meisner Reference Rosenheim and Meisner2013). Therefore, the objectives of this research were to evaluate the influence of spray droplet size on acifluorfen efficacy on Palmer amaranth control and develop prediction models to optimize spray droplet size recommendations for Palmer amaranth control.
Materials and Methods
Experiment Design and Establishment
Experiments were conducted in 2016, 2017, and 2018 in a non-crop environment in Dundee, MS (2016, 2017, and 2018) on a Sharkey clay soil; in Beaver City, NE (2016 and 2017) on a Holdrege silt loam; and in Robinsonville, MS (2017 and 2018) on a Commerce silt loam. Seven site-years of research were conducted, and individual site information including GPS coordinates, elevation, Palmer amaranth population density, application date, and weather conditions at time of application are presented in Table 1. Herbicide resistance to glyphosate and ALS-inhibiting herbicides has been reported across all experimental locations. Nevertheless, no level of PPO resistance was observed prior to and throughout the years in which experiments were conducted. Acifluorfen (Ultra Blazer®, 0.24 kg ai L–1; UPL Ltd., King of Prussia, PA) at 0.42 kg ai ha–1 plus crop oil concentrate (Agri-Dex®; Helena Chemical Co., Collierville, TN) at 1% v/v were applied to 15-cm-tall Palmer amaranth plants. Treatments consisted of six targeted droplet size distributions (150, 300, 450, 600, 750, and 900 µm) determined from the VMD of the measured droplet size distribution. One nontreated control per site-year was used for treatment comparison. Plot dimensions were 4 m wide by 12 m long, and treatments were arranged in a randomized complete block design with four replications. Buffer strips measuring 1 m were used between plots to minimize the effect of any potential spray drift on surrounding plots. Treatments were applied using a tractor-mounted sprayer equipped with a Pin Point® PWM system (Capstan Ag Systems, Inc., Topeka, KS) using non-Venturi Wilger™ precision technology spray nozzles (Wilger Inc., Lexington, TN) operated at 4.8 km h–1 and spray volume of 140 L ha–1 (Figure 1).
Table 1. Location, latitude, longitude, elevation, year, Palmer amaranth density, application date, and weather conditions at the time of acifluorfen application.a
a AMAPA, Amaranthus palmeri S. Wats., Palmer amaranth.
b Palmer amaranth population density was collected 1 wk prior to herbicide application.
c Wind direction at time of application.
d Precipitation totals (30 d) starting at 15 d prior to herbicide application date.
Figure 1. Tractor-mounted sprayer equipped with Pin Point® pulse width modulation (PWM) system (A) and non-Venturi Wilger™ precision technology spray nozzles (B) in Dundee, MS.
Prior to experiment initiation, the spray droplet size spectrum of acifluorfen was characterized in a low-speed wind tunnel located at the Pesticide Application Technology Laboratory in North Platte, NE. Nozzle type, orifice size, and application pressure necessary to produce the aforementioned droplet size treatments were determined using a Sympatec HELOS-VARIO/KR laser diffraction system (Sympatec Inc., Clausthal-Zellerfeld, Germany) equipped with an R7 lens capable of detecting particle sizes ranging from 18 to 3,500 µm (Table 2). The laser diffraction instrument was placed 30 cm downwind from the nozzle and 38 cm above the bottom of the wind tunnel. The spray plume was oriented perpendicular to the instrument and traversed through the laser beam with assistance of a mechanical linear actuator. The actuator moves the nozzles at a constant speed of 0.2 m s–1 such that the entire spray plume would pass through the laser beam. During application, each nozzle–pressure combination was traversed through the laser beam three times, with each pass serving as one repetition for VMD determination. Creech et al. (Reference Creech, Henry, Fritz and Kruger2015) describe in detail the operation of the low-speed wind tunnel at the Pesticide Application Technology Laboratory. Droplet size classifications were assigned in accordance with ASABE S572.1 (ASABE 2009).
Table 2. Nozzle type, application pressure, and droplet size classification for acifluorfen spray droplet size treatments.
a Flat-fan, non-Venturi nozzles; Wilger™ precision technology spray tips (Wilger Inc., Lexington, TN).
b Target spray droplet sizes used in data analysis.
c Spray classification according to ASABE S572.1.
Data Collection
Visual evaluation of Palmer amaranth control was carried out 28 d after herbicide treatment. Palmer amaranth control was evaluated on a scale of 0 (no control) to 100% (complete death of all plants) relative to the nontreated check (Frans et al. Reference Frans, Talbert, Marx, Crowley and Camper1986). Prior to herbicide application, 10 plants per plot were tagged at the soil surface for aboveground biomass evaluation. To facilitate evaluation of Palmer amaranth control in response to droplet size, plants were allowed to grow up to 15 cm (Butts et al. Reference Butts, Samples, Franca, Dodds, Reynolds, Adams, Zollinger, Howatt, Fritz, Hoffmann and Kruger2018, Reference Butts, Samples, Franca, Dodds, Reynolds, Adams, Zollinger, Howatt, Fritz, Hoffmann, Luck and Kruger2019a, Reference Butts, Samples, Franca, Dodds, Reynolds, Adams, Zollinger, Howatt, Fritz, Hoffmann, Luck and Kruger2019c). Rulers were used to ensure that tagged plants did not exceed 15 cm in height. Plants were selected in the center of each plot, preferably in between tractor wheel tracks to avoid running over plants. Furthermore, tagged plants were used to assist with evaluation of visible Palmer amaranth control in plots where new emergence or regrowth occurred. At 28 d after herbicide treatment, tagged plants were harvested, placed in paper bags, removed from the experimental area, and dried in a forced-air dryer at 55 C for 72 h to constant mass. The dry Palmer amaranth plants were pooled into one dry biomass measurement per plot and divided by 10 for average dry-shoot biomass per plant. Dry biomass measurements were converted into percent dry biomass reduction as shown in Equation 1.
$$100 - \left( {{X \times 100}\over{Y}} \right)$$
where X is the average dry biomass of an individual experimental unit and Y is the dry biomass of the nontreated unit. To minimize any block effect, percent biomass reduction was calculated by comparing dry-weight measurements to the nontreated check within each block.
Statistical Analysis
GAM analysis was conducted in R x64 3.4.3 using the mgcv package to provide estimated optimum droplet size for Palmer amaranth control and dry biomass reduction (Crawley Reference Crawley2013). The nontreated was included in the experiment for comparison but was not included in GAM analysis for either response to allow better separation between spray droplet sizes. To meet model assumptions, Palmer amaranth control and percent dry biomass reduction were converted to a beta distribution (Price et al. Reference Price, Shafii and Seefeldt2012; Zuur and Ieno Reference Zuur and Ieno2016), where data were bound between 0 and 1. Response variables were Palmer amaranth control and percent dry biomass reduction. Response variables were subjected to one smooth variable (spray droplet size) as presented in Equation 2.
$${\rm{Response\;variable}}\ \tilde\ \ {\rm{s}}\left( {{\rm{Target\;droplet\;size}}} \right)$$
GAMs were used to predict optimal droplet size that provided maximum Palmer amaranth control and dry biomass reduction. In addition, predicted spray droplet sizes were used to calculate the range in which 90% of maximum Palmer amaranth control and dry biomass reduction are sustained.
Results and Discussion
Pooled Site-Year Analysis
GAM analysis for Palmer amaranth control and dry biomass reduction across seven site-years is presented in Figure 2. Smooth-term estimated degrees of freedom (EDF) and deviance explained values are shown in Table 3. A smooth-term EDF of 1 indicates minimum GAM model fluctuation and characterizes a linear model (Butts et al. Reference Butts, Samples, Franca, Dodds, Reynolds, Adams, Zollinger, Howatt, Fritz, Hoffmann and Kruger2018, Reference Butts, Samples, Franca, Dodds, Reynolds, Adams, Zollinger, Howatt, Fritz, Hoffmann, Luck and Kruger2019a, Reference Butts, Samples, Franca, Dodds, Reynolds, Adams, Zollinger, Howatt, Fritz, Hoffmann, Luck and Kruger2019c). Deviance explained values provide a model-fitting estimation between predicted values and actual observations, with greater percentages representing an overall better model fit.
Figure 2. Proportion of visible AMAPA (Amaranthus palmeri S. Wats., Palmer amaranth) control (dotted line) and dry biomass reduction (solid line) following acifluorfen application predicted using generalized additive model (GAM) analysis across seven site-years. The gray-shaded area represents the 95% confidence intervals.
Table 3. Generalized additive model (GAM) smooth parameters and deviance explained for Palmer amaranth control and dry biomass reduction pooled across all site-years.
a Smooth-term estimated degrees of freedom (EDF) provides an estimation of model fluctuation for a response variable. Smooth-term EDF values of 1 represent a linear model.
b Deviance explained value represents the variability of a given response variable that is due to spray droplet size.
Smooth-term EDF values of 1 were observed for both Palmer amaranth control and dry biomass reduction (Table 3). Deviance explained of 7.2% was observed for Palmer amaranth control, meaning that 7.2% of differences in Palmer amaranth control were due to spray droplet size. In terms of dry biomass reduction, deviance explained was 1%, indicating that 99% of dry biomass reduction must be explained by factors other than spray droplet size. Similar deviance explained values resulting from prediction models developed across site-years have been reported by Butts et al. (Reference Butts, Samples, Franca, Dodds, Reynolds, Adams, Zollinger, Howatt, Fritz, Hoffmann and Kruger2018, Reference Butts, Samples, Franca, Dodds, Reynolds, Adams, Zollinger, Howatt, Fritz, Hoffmann, Luck and Kruger2019a, Reference Butts, Samples, Franca, Dodds, Reynolds, Adams, Zollinger, Howatt, Fritz, Hoffmann, Luck and Kruger2019c). Weather conditions, geographic location, soil type, fertility levels, weed density, and population genetics should be investigated in future research to implement acifluorfen model assumptions for Palmer amaranth control and dry biomass reduction across multiple locations.
GAM predicted optimum Palmer amaranth control and dry biomass reduction with acifluorfen using Fine sprays (Table 4). Models suggest that maximum Palmer amaranth control and dry biomass reduction could be achieved with 150-μm (Fine) droplets. These results agree with the general hypothesis that smaller droplets provide greater coverage, thus increasing the efficacy of contact herbicides such as acifluorfen. Rogers and Maki (Reference Rogers and Maki1986) reported increased acifluorfen deposition when spray droplet size was reduced from 410 (Very Coarse) to 130 μm (Fine). Furthermore, similar research reported increased herbicide phytotoxicity with spray droplet size as small as 100 μm (Very Fine) (McKinlay et al. Reference McKinlay, Brandt, Morse and Ashford1972; Prasad et al. Reference Prasad1987). According to GAM models, spray droplet sizes ranging from 150 (Fine) to 425 μm (Very Coarse) could be used to maintain at least 90% of maximum Palmer amaranth control. Therefore, the use of 425 μm (Very Coarse) spray droplets could be implemented as part of a drift mitigation approach.
Table 4. Generalized additive model (GAM) analysis for maximum Palmer amaranth control and dry biomass reduction pooled across all site-years.
a Spray droplet sizes required to achieve maximum Palmer amaranth control and dry biomass reduction.
b Spray classification according to ASABE S572.1.
c Larger spray droplet size recommended to maintain 90% of maximum Palmer amaranth control and dry biomass reduction.
Although finer sprays provided greater dry-biomass reduction, GAM results suggest that 755-μm (Ultra Coarse) spray droplets could be used to maintain at least 90% of maximum dry biomass reduction (Table 4). Based on these results, the use of Ultra Coarse sprays would be recommended to minimize spray drift potential. Nevertheless, the small deviance explained value (1%) indicates large dry biomass variability in response to spray droplet size across all site-years. Interpretations from GAM models across all site-years suggest that dry biomass is not an accurate predictor of Palmer amaranth control. Despite the greater deviance explained that was observed in GAM models using Palmer amaranth control data, spray droplet size recommendations across a wide range of geographic areas must be made with caution. The presented model suggests that 7.2% of acifluorfen efficacy variability could be attributed to spray droplet size. Thus, model predictions were presumed to be inaccurate, emphasizing the necessity of a location-specific analysis approach. Factors that also influence herbicide efficacy, such as weather conditions, weed density, time of day, and geographic location (Kudsk Reference Kudsk, Hatcher and Froud-Williams2017) may have been more detrimental to final acifluorfen efficacy than droplet size. These factors should be considered as parameters in future research for the development of stronger models using acifluorfen across different locations.
Location-Specific Analysis
Location-specific analyses were conducted to minimize discrepancies between GAM model predictions and observed values across all site-years. GAMs for Palmer amaranth and dry biomass reduction for Dundee, MS, Beaver City, NE, and Robinsonville, MS, across years are presented in Figures 3, 4, and 5, respectively. GAM smooth-term EDF and deviance explained values for each location pooled across years are presented in Table 5.
Figure 3. Proportion of visible AMAPA (Amaranthus palmeri S. Wats., Palmer amaranth) control (solid line) and dry biomass reduction (dotted line) following acifluorfen application predicted using generalized additive model (GAM) analysis for Dundee, MS, in 2016, 2017, and 2018. The gray-shaded area represents the 95% confidence intervals.
Figure 4. Proportion of visible AMAPA (Amaranthus palmeri S. Wats., Palmer amaranth) control (solid line) and dry biomass reduction (dotted line) following acifluorfen application predicted using generalized additive model (GAM) analysis for Beaver City, NE, in 2016 and 2017. The gray-shaded area represents the 95% confidence intervals.
Figure 5. Proportion of visible AMAPA (Amaranthus palmeri S. Wats., Palmer amaranth) control (solid line) and dry biomass reduction (dotted line) following acifluorfen application predicted using generalized additive model (GAM) analysis for Robinsonville, MS, in 2017 and 2018. The gray-shaded area represents the 95% confidence intervals.
Table 5. Generalized additive model (GAM) analysis for Palmer amaranth control and dry biomass reduction for each location pooled over years.
a Smooth-term estimated degrees of freedom (EDF) provides an estimation of model fluctuation for a response variable. Smooth-term EDF values of 1 represent a linear model.
b Deviance explained value represents the variability of a given response variable due to spray droplet size.
Smooth-term EDF values of 4.9 and 3.3 indicate a nonlinear characterization of visible Palmer amaranth control GAM models for Dundee, MS, and Robinsonville, MS, respectively (Table 5; Figures 3 and 5). Location-specific analysis for all locations significantly increased deviance explained values to 49.6% (7-fold increase) in Dundee, MS, and 20% (3-fold increase) in Robinsonville, MS, which indicates better model performance in predicting Palmer amaranth control as influenced by spray droplet size. Unlike previous models, GAM analysis for Palmer amaranth control in Beaver City, NE, pooled over 2 yr had a linear pattern (smooth-term EDF = 1) and slightly increased deviance explained value (8.9%) when compared to GAM model developed across all site-years (Table 5; Figure 4). The small deviance explained increase (1.7%) observed in Beaver City, NE, may be a result of Palmer amaranth density differences observed at this location between 2016 and 2017. Although experimental areas were in near proximity, Palmer amaranth density was significantly lower in 2017 compared to 2016 (Table 1).
Deviance explained values from GAM models using dry biomass reduction data were 1.8%, 1.4%, and 9.7% for Dundee, MS, Beaver City, NE, and Robinsonville, MS, respectively (Table 5). Although GAM models developed using a location-specific approach increased accuracy of prediction models, especially in Robinsonville, MS, deviance explained values observed in GAM models using dry biomass reduction remained significantly lower compared to GAM models using Palmer amaranth control evaluations. Deviance explained differences that were observed between response variables could be attributed to the method in which visible control evaluations are made and the mode of action of acifluorfen. Following acifluorfen application, it was common to see similar plant damage and dry biomass reduction across a range of droplet sizes. However, upon closer visual inspection, regrowth was often observed in plants sprayed especially with larger droplet sizes, resulting in dry biomass reduction without accounting for actual plant mortality. This research indicates that care should be taken in future weed control modeling research to determine weed mortality as opposed to dry biomass reduction to fully evaluate acifluorfen effectiveness. Therefore, spray droplet sizes that would provide maximum control were calculated using visual control models from each location (Table 6).
Table 6. Spray droplet size prediction based on generalized additive model (GAM) analysis to reach and maintain 90% of maximum Palmer amaranth control with acifluorfen in Dundee, MS, Beaver City, NE, and Robinsonville, MS, pooled across years.
a Droplet size required to achieve maximum Palmer amaranth control.
b Spray classification according to ASABE S572.1.
c Larger spray droplet size recommended to maintain 90% of maximum Palmer amaranth control.
GAM for Dundee, MS, using Palmer amaranth control observations pooled across 3 yr suggests that maximum control could be achieved with 250-μm (Medium) spray droplets. Previous research conducted by De Cock et al. (Reference De Cock, Massinon, Salah and Lebeau2017) reported that spray droplets ranging between 200 (Fine) and 250 μm (Medium) provided moderate kinetic energy, which led to increased spray deposition. Additionally, spray deposition was increased up to 80% with 250-μm (Medium) spray droplets. Furthermore, similar research conducted by Shaw et al. (Reference Shaw, Morris, Webster and Smith2000) reported that 250-μm (Medium) droplets optimized acifluorfen efficacy for common cocklebur control. The 250-μm (Medium) droplet size probably allowed acifluorfen to be dispersed uniformly over a larger portion of the plant tissues in Dundee, MS; however, leaf coverage was not evaluated in this research. Prediction models suggested that 150-μm (Fine) and 370-μm (Coarse) droplets could be used to achieve maximum acifluorfen efficacy on Palmer amaranth in Beaver City, NE, and Robinsonville, MS, respectively (Table 6). The convoluted observations from this research could be associated with not only weather conditions and weed density at the time of application, but also morphological characteristics from each Palmer amaranth population. Differences in Palmer amaranth leaf morphology and growth traits have been previous reported by Bravo et al. (Reference Bravo, Leon, Ferrell, Mulvaney and Wood2018). Acifluorfen applied with 150-μm (Fine) droplets must have provided greater plant tissue coverage in the less dense Palmer amaranth area in Beaver City, NE, whereas the larger, heavier 370-μm (Coarse) droplets must have contributed to greater herbicide deposition in the thicker Palmer amaranth leaf canopy from Robinsonville, MS. Analysis of plant canopy and acifluorfen deposition is necessary in future research so as to give a more detailed explanation of these relationships. According to prediction models, medium sprays could be used to sustain 90% of maximum Palmer amaranth control in Dundee, MS (310 μm, Medium) and Beaver City, NE (340 μm, Medium) (Table 6). In addition, 90% of maximum Palmer amaranth control could be maintained using Ultra Coarse sprays in Robinsonville, MS (Table 6). These observations may be useful to determine the spray droplet buffer in which 90% of optimal Palmer amaranth control is sustained in each location. Considering that current spray drift mitigation efforts primarily focus on increasing spray droplet size to minimize off-target movement (Bueno et al. Reference Bueno, da Cunha and de Santana2017), the findings of this research could provide location-specific information regarding the droplet size range in which greater levels of weed control are sustained (Creech et al. Reference Creech, Moraes, Henry, Luck and Kruger2016). In addition, this research has provided proof of concept that the use of PWM sprayers, paired with appropriate nozzle–pressure combinations for acifluorfen, could be effectively integrated into precision agricultural practices to apply optimum droplet sizes in a location-specific weed management approach.
Conclusions drawn from GAM analysis identified that prediction models suggest the use of 150-μm (Fine) droplets to achieve optimal acifluorfen efficacy on Palmer amaranth across different locations in Mississippi and Nebraska. However, acifluorfen efficacy using a broad range of spray droplet sizes could only be predicted with 7.2% accuracy when analyzed using a pooled site-year approach. More precise acifluorfen applications could be achieved through precision agricultural methods by applying specific droplet sizes in a location-specific approach. GAM accuracy increased significantly using a tailored, location-specific analysis for Palmer amaranth management. According to prediction models, acifluorfen applied with 250-μm (Medium) droplets provide the greatest level of Palmer amaranth control in Dundee, MS, and 90% of maximum weed control could be maintained with 310-μm (Medium) spray droplets. Conversely, Palmer amaranth control was greater when acifluorfen was applied with 150-μm (Fine) droplets in Beaver City, NE, and 90% of maximum control was sustained with 340-μm (Medium) droplets. For Robinsonville, MS, GAM models predicted that 370-μm (Coarse) droplets should be used to achieve optimal Palmer amaranth control with acifluorfen, and 680-μm (Ultra Coarse) droplets could be used to sustain 90% of total control. These differences in optimum droplet sizes were probably due to leaf shape and architecture and weed density at each experimental location. Previous research reported different leaf shapes in indigenous Palmer amaranth populations across the southern United States (Gray et al. Reference Gray, Shaw, Bond, Stephenson and Oliver2007), and these characteristics may influence droplet deposition and retention and thereby herbicidal efficacy (Lake Reference Lake1977; Massinon et al. Reference Massinon, De Cock, Forster, Nairn, McCue, Zabkiewicz and Lebeau2017; Spillman Reference Spillman1984). However, several other factors, such as weather conditions at application time, soil type, fertility, time of day, and light incidence, may have played a significant role in final acifluorfen efficacy. Future research should identify and investigate the influence of environmental and agronomic factors on prediction modeling so as to improve herbicide application efficiency and to develop more robust Palmer amaranth control models with a wide range of spray droplet sizes.
Acknowledgments
This research was partially funded by Monsanto Company as part of the Will D. Carpenter Distinguished Field Scientist Graduate Assistantship. The authors would like to thank all the undergraduate and graduate research assistants at Mississippi State University and University of Nebraska who helped with the performance of this research. The authors would further like to thank Capstan Ag Systems, Inc., for providing assistance with the PWM system, and Wilger Inc., for supplying the nozzles used in this research. No conflicts of interest have been declared.
