2.1. Deposition
Table 1 shows the median deposition of the sprayed herbicide solution. Greater deposition was achieved with remotely piloted aircraft (RPA) (3.466 μg cm
−2) than with ground application via a backpack sprayer pressurized with CO
2 (2.242 μg cm
−2). There were no significant differences for the different nozzles used in the application. The XR 11001 yielded a median deposition of 2.919 μg cm
−2 and the AirMix 11001 yielded 2.372 μg cm
−2. Jeevan et al. [
28] observed similar behaviour. The authors explain that applications via RPA showed higher deposition due to the reduced application rate, with a higher concentration of the active ingredient in the droplets and lower runoff to the soil surface. Other authors observed that the downwash effect had a positive influence on deposition, in which downward pressure on the droplets favoured deposit and decreased drift [
29,
30]; these observations may explain what occurred in the current study.
It is important to remember some points. Low-volume applications require a more efficient deposition of droplets on the target surface [
31] due to the use of more concentrated solution and the volume distributed over the area, so that there is no loss of product, reduced effectiveness, or increased environmental risks. The deposition characteristics may be different for different RPA models. Sinha et al. [
32] found that although nozzles located under each rotor are quite common in commercial RPA, nozzles arranged in a boom yielded a lower drift potential. The authors attribute this performance to the aerodynamic differences for the nozzle arrangements in relation to the rotors of the RPA.
Control charts are tools employed to analyse whether or not a process is under statistical process control and they are commonly used to monitor undesirable changes [
33]. Despite the technological advances in application technology, the spraying process may exhibit deficiencies, so it can be beneficial to obtain application data in the area. Deposition analysis through control graphs can show how well the deposition is controlled, as shown in
Figure 1.
The graphs show that applications via RPA (
Figure 1a,b) showed greater variation in relation to the average value. Application via RPA with the XR 11001 nozzle (
Figure 1a) presented, in replicate six, a deposition of 5.128 μg cm
−2, while the overall mean value of the treatment was 2.836 μg cm
−2, constituting a variation of 2.292 μg cm
−2. When the application was performed with the AirMix 11001 nozzle (
Figure 1b), for the same sprayer, the greatest variation was 2.659 μg cm
−2; in this treatment, the mean was 4.834 μg cm
−2 and in replicate six, the deposition was 2.175 μg cm
−2.
The ground application resulted in less deposition variation in relation to the mean value; that is, the data for this application method tended to be closer to the mean value of the treatment. When the application was performed with the XR 11001 nozzle (
Figure 1c), the largest difference (1.273 μg cm
−2) occurred between the sample (replicate six, 3.876 μg cm
−2) and the mean (2.603 μg cm
−2). For application with the AirMix 11001 nozzle (
Figure 1d), replicate three presented a deposition of 2.825 μg cm
−2, and the treatment mean was 2.245 μg cm
−2, with a variation of 0.580 μg cm
−2. The variation between repetition six (2.245 μg cm
−2) and the mean (2.245 μg cm
−2) was equal to zero, with no variation between the evaluated point and the treatment mean.
Although the results showed variations in application, the process was well controlled for all treatments according to the statistics obtained. As seen in the graphs; the data are all positioned between the upper (UCL) and lower (LCL) limits, indicating a quality application from the point of view of the distribution of the spray in the area. There were no patterns of nonrandomness. A process is out of control when points in the control chart are outside the upper or lower limits. Thus, the process changes from an average with projected variation over the sampling [
34]. High variability in deposition may be responsible for excessive or insufficient product application within the applied area. Process analysis facilitates intervention and corresponding changes in process parameters for better performance [
35].
Notably, although the results reveal that the process was under control, the values obtained are not necessarily the desired values. The control graphs show that the ground application yielded less variation but also lower mean values of deposition. Jorani et al. [
36] state that the detection metrics of the X bar graph (the one used in this study) only take information about new data samples to the decision-making process and ignore other information, such as ideal deposition values.
2.2. Droplet Spectrum
Table 2 shows the average coverage (%), droplet density (droplets cm
−2), and relative amplitude (RA) obtained by the application of the herbicide spray (glyphosate). The coverage, which represents the proportion of the area occupied on the surface of the target by the droplets, was greater for the ground application, at 18.1%. Application via RPA presented 5.1% coverage. There was no significant difference in coverage with the different nozzles. The XR 11001 standard flat fan showed 10.7% coverage, and the coverage of the AirMix 11001 air induction flat fan was 12.5%. As expected, an increase in the application rate provided an increase in coverage. It should be noted that higher application rates may pose the risk of runoff from the deposition area (grass), although it offers a greater probability of target coverage [
37].
Subr, Al-Ahmadi, and Abbas [
38] reported that a standard flat spray nozzle (12003 Agroplast
®, Sawin, Poland) produced greater coverage than an air induction nozzle (8MS 03C Agroplast
®, Sawin, Poland), with 11.5% and 5.9% coverage, respectively, for application to eggplant (
Solanum melongena L.) via a pressurized backpack sprayer (284 L ha
−1). Ya, Yu, Kang, and Lee [
27] evaluated application via RPA for rice (
Oryza sativa L.) and soybean (
Glycine max (L.) Merrill) plants with air induction flat fan (AI) nozzles designed by the authors and standard flat fan spray nozzles (XR 110015VS). Regarding coverage, values of 5.1% and 3.4% for the AI nozzle and 2.1% and 1.4% for the XR 110015VS nozzle were obtained for rice and soybean, respectively. The larger droplet size produced by the AI apparently reduced the drift, increasing the number of droplets that reached the target, thus increasing the coverage. Abdelmotalib et al. [
39] reported that when the droplets generated by air induction nozzles impact the target surface, they are fragmented into smaller droplets, which reduces the drift potential and allows for better controlled application, which may improve coverage.
The density, which represents the number of droplets in an area, was higher for the method with the higher rate (100 L ha
−1), ground application via a backpack sprayer, with 239.9 droplets cm
−2; application via RPA presented a density of 55.4 droplets cm
−2. This effect has also been observed by different researchers. Chen et al. [
40] performed applications of thidiazuron and ethephon solution for cotton (
Gossypium hirsutum L.) defoliation via RPA at different application rates (15, 18, 22.5, 25.5, 29.1, and 30 L ha
−1) and observed that, as the volume of spray solution increased, the average droplet density gradually increased, with a mean of approximately 17 droplets cm
−2 at the lowest rate and 27 droplets cm
−2 at the highest rate. Önler et al. [
41] also observed that the density increased as the application rate increased. For the different nozzles used, application with the standard flat fan XR 11001 yielded the highest droplet density, with 213.9 droplets cm
−2, which can be attributed to the smaller droplet size. The flat fan nozzle with air induction yielded 81.4 droplets cm
−2. This can be explained by the findings of Mur et al. [
42] that smaller droplets are more likely to be deposited on the target than droplets of the same volume distributed in larger droplets. Considering the indications of Syngenta Crop Protection AG (Basel, Switzerland), which recommends at least 30–40 droplets cm
−2 for postemergence herbicide applications [
43] and evaluating only the density found for the different nozzles in this study, both have the potential to facilitate efficiency in weed control. Urach Ferreira [
44], evaluating applications with the herbicide pyroxasulfone, found that higher droplet densities did not result in better weed control efficacy.
Note that the values of coverage and droplet density should be viewed with caution because there are no standardized ranges of values that inform the ideal percentage of coverage nor the number of droplets cm
−2 for each active ingredient correlated with biological efficacy. There are indications [
37], as previously discussed, but efficacy studies are important to better understand the disease, pest, or weed control.
The RA indicates the homogeneity in the droplet size [
45]. The lower the RA value is, the lower the amplitude between the values of the diameters of the droplets produced during spraying and, consequently, the better the quality of the application, as the process is more easily controlled. In this study, the RA was 1.0 for the application via RPA, denoting lower dispersion in the droplet spectrum, indicating that the uniformity of the droplet size distribution was greater with this application method. The ground application presented an RA of 1.3. For the different nozzles used in the application, XR 11001 and AirMix 11001, there was no significant difference in RA. Xue et al. [
46] suggest that RA values lower than one (1) indicate good homogeneity in droplet size distribution, as was found for the application via RPA. A spray with a small RA is important to increase application efficiency, as there will be a greater number of droplets sized according to the target’s needs [
47].
Researchers [
48,
49] have shown that the herbicide glyphosate added to water spray has the ability to reduce the surface tension of fluid, and Xue et al. [
50] reported that decreasing the surface tension of the spray solution can inhibit its shape retention capacity and promote its rupture, modifying the droplet spectrum and leading to a more uniform size distribution. This may have occurred in the present study, where the application at a rate of 10 L ha
−1 yielded a higher concentration of herbicide and greater homogeneity than the spray solution of 100 L ha
−1. This higher concentration may have a greater effect on the physical characteristics of the spray solution, resulting in greater alteration in the droplet spectrum.
Table 3 shows the means of the volume median diameter (VMD) and the percentage of droplets smaller than 100 μm, where the interaction between the method and nozzle factors was significant for these variables. The AirMix 11001, which introduces air into the droplets, exhibited higher VMD values regardless of the application method, with average values of 393.4 and 436.0 μm for applications via RPA and the backpack sprayer, respectively, than the XR 11001. Furthermore, this same nozzle (AirMix 11001) had a lower VMD value for application via RPA (393.4 μm) than for ground application (436.0 μm), demonstrating that this type of nozzle may be more influenced by the different application methods.
Gibbs, Peters, and Heck [
51] observed that, compared with ground sprays, applications via RPA produced smaller droplets. This may occur due to a secondary breakdown of the droplets promoted by the air flow from the blades. Furthermore, during the droplet formation process, when air is incorporated into the droplet, the bubbles inside the droplets form weaker points that facilitate disintegration [
52,
53]. This may explain what occurred in this study when the AirMix 11001 nozzle was used.
Note that in herbicide applications, especially by aerial methods, the droplet size should be chosen very carefully. There is a compromise between the greatest effectiveness, sometimes achieved as a function of better leaf coverage using smaller droplet sizes, and minimizing the risk of herbicide exodrift, which is achieved using larger droplet sizes [
54]. However, if the droplet diameters are very large, the spray is not uniformly deposited on the target, which may favour runoff due to the action of gravity [
55]. Butts et al. [
56] showed that droplets sized 395 μm maximize weed mortality when a dicamba and glyphosate herbicide mixture is applied at 94 L ha
−1 and that the droplet size can be increased to 620 μm while maintaining 90% weed control and mitigating the potential drift risk, suggesting the use of air induction nozzles.
The drift potential, expressed as the percentage of droplet volume smaller than 100 μm, was greater for applications with the standard XR 11001 flat fan nozzle (2.4% and 4.1%) than for those with the AirMix 11001 flat fan with air induction (0.8% and 0.5%) for the different application methods. This presents reductions of 66.8% and 86.8% for aerial and terrestrial applications, respectively, with an air induction nozzle, when compared to the application with the standard flat fan nozzle. Therefore, the air induction nozzle presented fewer smaller droplets, resulting in a lower exodrift potential; thus, these droplets were less likely to deviate from the target area or evaporate into the atmosphere. The application with the XR 11001 had a greater percentage of droplets smaller than 100 μm for application via a ground sprayer (4.1%) than for application via RPA (2.4%). This may have been due to the droplets being closer to the value of the VMDs found in the application via RPA, which were above 200 μm. Despite the difference between some treatments, notably, the percentage of droplets smaller than 100 μm did not reach 5%. When the AirMix 11001 was used, there was no significant difference in the volume of droplets smaller than 100 μm for the different application methods.
Feng et al. [
57] observed that downwash may result in a more uniform and regular spray distribution. Notably, the droplets sprayed by a hydraulic nozzle, under the influence of downwards airflow, reach a certain speed that can result in the droplets reaching the water-sensitive paper, producing a trail that can increase the identified droplet size. Moreover, measuring droplet sizes less than 50 μm is not feasible for this methodology, so the identified droplet spectrum may be increased [
58]. This may also have contributed to the lower percentage volume of droplets smaller than 100 μm in the application via RPA in this study.
2.3. Effectiveness in Controlling Urochloa Decumbens
Means and medians of the percentage of effectiveness in controlling
Urochloa decumbens can be seen in
Table 4. There was no difference in the control effectiveness between the application methods, RPA and ground. At 14 days after application (DAA), the control means were 77.2% and 75.5%, respectively; at 21 DAA, the control medians were 93.0% and 95.0%, respectively; and at 28 DAA, the control medians were 90.0% and 88.0% for application via RPA and ground, respectively. This finding is interesting because the use of a reduced spray volume, without loss of control effectiveness, may result in cost reduction [
59]. Jeevan et al. [
28] evaluated the effect of different herbicide application rates for weed control in transplanted rice. For applications via RPA, rates of 15, 20, and 25 L ha
−1 were studied, and for ground application with a manual backpack sprayer, a rate of 500 L ha
−1 was used. Based on the results, the authors concluded that the application of 25 L ha
−1 via RPA provided better weed control efficiency and higher crop yield, possibly due to better deposition.
Compared with the AirMix, the XR11001 had greater weed control effects on all the days evaluated, with 78.1% weed control at 14 DAA, 95.0% weed control at 21 DAA, and 90.0% weed control at 28 DAA. The AirMix 11001 showed a weed control of 74.6% at 14 DAA, 90.0% at 21 DAA, and 85.0% at 28 DAA. Ferguson et al. [
60] reported that some herbicides are less effective when sprays have larger droplets because these droplets may not be deposited on the target plant, especially for target plants with small and narrow leaves. Research related to the pesticide application technology suggests that droplets sized 100 to 300 μm are suitable for herbicide spraying [
61], as found for the XR 11001 in the current study. However, notably, from 21 DAA, both nozzles showed satisfactory control, above 84%.
Wang et al. [
62] observed that there was no significant difference for the biological efficacy in weed control in wheat crops for different standard flat spray nozzles and air-induced flat spray nozzles. In all cases, the efficacy values were greater than 80%. The authors explained that although fine droplets tend to improve the adhesion and density of droplets on targets, they also have greater potential for drift, which may reduce deposition on targets.
In this study, after 21 DAA, the weed control was acceptable or very good, with values equal to or greater than 85% (
Table 4). The image of the control efficiency of
Urochloa decumbens (%) at 21 DAA of the herbicide mixture (glyphosate) can be seen in
Figure 2 for the different treatments studied. At 14 DAA, there was possibly not enough time for the activity of the herbicide molecule to be complete based on plant metabolism. Brightenti et al. [
63] observed that glyphosate application had a control efficiency of 85% at 25 DAA for
Urochloa decumbens. Silveira et al. [
64] found 100% control of
Urochloa decumbens at 15 DAA, and the authors explained that in the early stages of development of these grasses, control is facilitated because the plant tissues have a greater capacity to transport herbicide molecules and have a lower capacity to recover from the injuries caused. In the aforementioned study, pesticide application was carried out when the plants had three to four formed tillers (new grass shoots) and an average height of 0.30 to 0.35 m.