The experiment was conducted in a commercial eucalyptus area in São Mateus, state of Espírito Santo, Brazil (18°34’02’’S; 40°05’10”W). The soil of the area was classified as Typic Hapludult of loamy sandy texture. The region’s climate was classified as Aw, hot and humid, with a dry season in the autumn-winter and a rainy season in the spring-summer, according to the Köppen classification (Alvares et al., 2013).

The experimental area was composed by eucalyptus sprouts (hybrid Eucalyptus grandis × Eucalyptus urophylla, commercial clone CCARAOP), with mean height of 2.80 m (measured with a tape measure from the stem base to the apex) and spacing of 3.60 × 2.50 m, totaling 1,111 plants per hectare. Figure 1 shows the location of experimental area and the distribution of eucalyptus plants.

The remotely piloted aircraft (RPA) (DJI Agras T40 model; SZ DJI Technology, Nanshan, China) used in the experiment ( Figure 1 ) was set and calibrated prior to the applications. This RPA features a coaxial dual-rotor structure, with eight rotors mounted above and below the four articulating arms. The spraying system is formed by a magnetically-driven centrifugal pump, a flow controller, and two atomized nozzles with anti-dripping centrifugal valves that have a theoretical droplet size adjustment from 50 to 500 µm.

The following operational parameters were maintained constant in the RPA for all treatments: application rate of 12.0 L ha^-1, flight speed of 5.0 m s^-1, operational flight height of 4.0 m above the eucalyptus sprout canopy; and application route parallel to the planting rows, considering the wind perpendicular to the aircraft direction.

The manual electric backpack sprayer (MEBS) used (Yamaho^®; Mogi das Cruzes, Brazil) had a solution tank capacity of 20 L and a Yamaho SR-1 flat fan nozzle with air induction. The application rate used was approximately 100 L ha^-1. The spraying quality was standardized with applications at an operational speed of approximately 1.0 m s^-1. The working pressure was 200 kPa, resulting in a flow rate of 0.65 L min^-1 and an application range of 1.05 m.

The experiment was divided into two phases. The first phase consisted of obtaining data on application quality and droplet deposition on eucalyptus sprouts, whereas the second phase consisted of determining the risk of occupational exposure for applicators during the experimental applications.

The experiment was conducted in a randomized block design with four replications, with treatments implemented in a 3×3 + 1 factorial arrangement, consisting of three droplet sizes (150, 300, and 450µm); three application ranges (7, 9, and 11 m), defined in the RPA control settings; and a control treatment (using the MEBS). After the repetitions of each experimental treatment, the adjustments of the operational parameters (application range and droplet size) were manually made in the RPA control system. Table 1 shows the experimental treatments.

The experimental area was 6,400 m² (80 × 80 m). All treatments were carried out in total area and in the same block (application day). A 15.0 m border area was disregarded at the edges to allow some distance for the RPA sprayer to be activated.

The experimental unit (central area) was 2,400 m² (80 × 35 m), where the eucalyptus sprouts had a greater uniformity in relation to the canopy shape. The data collection points in each treatment were four target plants, spaced approximately 3.60 m apart, that together composed the experimental unit. For the second phase, two volunteer applicators stayed at 15.0 m from the end of the RPA application route in the experimental area, spaced 60.0 m apart. The treatment using the backpack sprayer was performed by a volunteer applicator passing between the planting rows to apply the solution on eucalyptus sprouts. Figure 2 shows the experimental area design.

The solution sprayed in all treatments in both experimental phases was composed of water, brilliant blue dye at the rate of 4.0 g L^-1, and non-silicone adjuvant based on balanced polymers specific for aerial applications with low solution volume (0.2% v v^-1) (Helper Air^®, ICL, São Paulo, SP, Brazil).

Application quality was characterized using water-sensitive paper cards (76 × 26 mm), and droplet deposition was evaluated using rectangular flexible polyvinyl chloride cards (76 × 26 mm). Water-sensitive paper and flexible polyvinyl chloride cards were attached to each target plant with the aid of metal clips, positioning them at the same height in relation to the canopy, following the angle and alternate phyllotaxy of leaves in the inner and outer sections of the upper, middle, and lower layers of eucalyptus sprout canopies ( Figure 3 ). The positioning of both card types was determined focusing on the overlap of droplets in each experimental treatment.

Both card types were collected five minutes after application of each treatment, using powder-free nitrile gloves, to allow the solution to evaporate, maintaining only the dye. They were then placed in kraft paper bags (water-sensitive paper cards) or labeled plastic bags (flexible polyvinyl chloride cards), which were labeled. These samples were then placed in expanded polystyrene boxes and taken to the Laboratory of Mechanization and Agricultural Defensives (LMDA) of the Northern Espírito Santo University Center of the Federal University of Espírito Santo, in São Mateus, ES, Brazil for analyses.

A wireless system (DropScope^®; SprayX, São Carlos, Brazil) was used to scan the water-sensitive paper cards for data acquisition and analysis. This system consists of an application software and a wireless digital microscope equipped with a digital image sensor with a resolution of over 2,500 dpi, allowing for the detection of partially overlapping droplets with a diameter of approximately 25µm. Recent research studies have confirmed the reliability of data obtained using this system for assessing spectra of droplets sprayed on shrubby crops by RPAs (Vitória et al., 2023; Lopes et al., 2023; Santos et al., 2024; Ribeiro and Vitória, 2024).

The following parameters were obtained through water-sensitive paper cards: droplet coverage (%); droplet density (droplets cm^-2); volume distribution by droplet size class (Dv_0.5 or VMD, Dv_0.1, and Dv_0.9, i.e., diameters of droplets composing 50%, 10%, and 90% of the applied volume, respectively); relative amplitude, a coefficient that determines droplet uniformity (dimensionless); and potential drift risk (%), which is the percentage of the applied volume composed of droplets smaller than 100µm.

Droplet deposition was estimated by removing the dye in the laboratory, washing the flexible polyvinyl chloride cards with 50 mL of deionized water per sample. The samples were manually shaken for approximately 60 seconds; the resulting solutions were subjected to absorbance readings at 630 ηm on a digital UV-VIS spectrophotometer with a holographic flashing monochromator (IL-226-NM-BI; Kasuaki^®, China).

The development of the calibration curve was based on absorbance of seven solutions with standard concentrations (0, 2.5, 5, 10, 20, 25, and 100 ppm). The analysis of the linear relationship between the concentration and the absorbance of standard solutions resulted in a coefficient of correlation of R² = 0.999. The equation that described this correlation was y = 0.0008x + 0.0015. Droplet deposition per unit area (µL cm^-2) was calculated considering the spectrophotometer readings, the calibration curve data, and area of flexible polyvinyl chloride cards, as described by Vitória et al. (2022); Crause et al. (2023), and Ribeiro et al. (2023a).

The solution described in section 2.4. was applied as a substitute of pesticides to measure and compare the occupational exposure of applicators to droplets applied by an RPA and an MEBS, using operational conditions identical to those described in section 2.2. The experimental treatments were similar to those described in section 2.3.

The test with the RPA was conducted with two volunteer applicators (mean height of 1.70 m and mean weight of 70 kg), who stayed 60.0 m distant from each other, at 15.0 m from the RPA flight path in the experimental area, as shown in the experimental setup in section 2.3 (Experimental design). The test with the backpack sprayer was carried out with the selection of an applicator with technical experience in spraying, as the applications of herbicides in forest areas; the applicator with a lance held in the right hand carried out up and down movements of 0.50 m over the canopy of plants, moving forward continuously through the area to be sprayed.

The potential for dermal occupational exposure of applicators was determined using whole-body dosimetry, as described in previous studies (Tsakirakis et al., 2014; Cao et al., 2015; Yan et al., 2021; Lake et al., 2021). Prior to each experimental treatment, the volunteer applicators (RPA and MEBS) were dressed on their everyday clothing, disposable coveralls with micropore laminated tissue hood (Steelflex^®, São Paulo, Brazil), a pair of disposable powder-free nitrile gloves with texturized fingertips, a disposable valveless respirator with synthetic fiber filters, and safety transparent glasses produced according to the ANSI Z87.1-2003 standard ( Figure 4 ). The two volunteer applicators corresponded to two replications per treatment for each experimental RPA treatment, as they were using new items (disposable coveralls, gloves, and respirators) in each experimental treatment to determine their individual dermal exposure.

Approximately 15 minutes after the application of each treatment, to allow for the evaporation and drying of spray drift droplets, the disposable coveralls, gloves, and respirators were removed carefully by two assistants using nitrile gloves to avoid cross contamination. The disposable coveralls were cut with scissors, sectioning them nine parts corresponding to different parts of the applicator body: head, left arm, right arm, chest, back, right thigh, left thigh, left leg, and right leg ( Figure 5 ). Samples of sectioned parts of disposable coveralls, gloves, and respirators, were placed in labeled plastic bags, individually sealed, taken to the laboratory, and refrigerated until residual dye extraction.

In the laboratory, sections of disposable coveralls, gloves, and respirators were subjected to the same procedure described in section 2.4. (Determination of application quality) for determination of dye deposition, using spectrophotometry. However, the area of each part of disposable coveralls was measured with a ruler, resulting in the following measurements: 1,080 cm² (legs), 960 cm² (thigh), 3,998 cm² (chest), 1,518 cm² (arms), 705 cm² (head), 330 cm² (respirator), and 680 cm² (hands). Each part was carefully withdrawn from the plastic bags using laboratory tweezers and individually placed in a 2,0 L glass backer. The coverall parts 1-9 were washed using 1,0 L of deionized water, and each glove and respirator was washed using 400 mL of deionized water. The bakers were shaken every minute for ten minutes with aid glass rod to ensure complete extraction of deposited droplets; the solutions were then subjected to spectrophotometric readings.

The applications were carried out in the mornings. Weather data were recorded by a meteorological station (5500AG; Kestrel^®, Boothwyn, USA). Air temperature during the experiment varied between 25.7 and 29.1°C, with a mean relative air humidity exceeding 55%, and mean wind speed between 1.3 and 5.8 km h^-1. The methodology described in the ISO Standard 22866 of the International Organization for Standardization (ISO, 2005) establishes that air temperature during applications should be between 5 and 35°C, with a maximum of 10% of wind speed measurements below 1.0 m s⁻¹, and a wind direction limit of 90 ± 30° in relation to the application line.

The Shapiro-Wilk test was applied to evaluate the homogeneity and normality of the data residuals. The results of application quality and droplet deposition were then subjected to the Tukey’s test for pairwise comparisons, and the Dunnett’s test was used for comparisons with the control treatment (MEBS) when needed. All statistical analyses were conducted in the Rbio statistical software at a 5% significance level (Bhering, 2017).

Droplet size and application range significantly affected droplet density and potential drift risk when using RPA for all ESCLs. Droplet deposition on all canopy layers was not significantly affected by the factors, confirming the null hypothesis of no difference between treatments ( Table 2 ).

The 9.0 m application range resulted in greater mean droplet coverage for all ESCLs, presenting means 15.09% and 30.18% higher than those found for application ranges of 7.0 and 11.0 m, respectively ( Table 3 ). However, these differences were not statistically significant. Droplet size was significantly affected by the application range only for the lower ESCL, in which 300µm droplet size had the highest mean coverage. Despite the variations in mean droplet coverage among the ESCLs, according to the different droplet sizes, the overall means for 150, 300, and 450µm were similar ( Table 3 ).

The application ranges of 7.0 and 9.0 m presented higher mean droplet density for all ESCLs, with a significant mean increase of approximately 2.0-fold in the quantity of droplets cm^-2 compared to the 11.0 m application range ( Table 3 ). However, the 9.0 m application range had a more uniform distribution in all canopy layers, but not differing statistically from the 7.0 m application range. Droplet size and droplet density were inversely correlated in all layers. Droplet sizes of 150 and 300µm presented higher means compared to 450µm, with increases of 6.29 droplets cm^-2, equivalent to 67.84% and 47.36% respectively ( Table 3 ).

Using the manual electric backpack sprayer (MEBS) resulted in a significantly higher mean droplet coverage, which was 65.0% higher than that found using the RPA ( Table 3 ). Although the RPA treatments showed a lower mean droplet density, statistical analyses did not show a significant difference compared to the control treatment.

The application ranges of 7.0 and 9.0 m presented a higher droplet deposition on all ESCLs, whereas the 11.0 m application range resulted in the lowest droplet deposition, accumulating primarily on the upper ESCL (0.007µL cm^-2) ( Figure 6 ). The smallest droplet size (150µm) resulted in the highest droplet deposition on the upper (0.009µL cm^-2), middle (0.006µL cm^-2), and lower (0.003µL cm^-2) ESCL compared to those found for 300 and 450µm. However, no significant difference within application ranges and droplet sizes was found for RPA, as shown by Anova (p ≥ 0.05).

Droplet deposition showed a consistent trend with droplet coverage and droplet density, regardless of the factors analyzed for the RPA treatments (application range and droplet size). The upper and middle ESCL presented higher droplet deposition compared to the lower layer ( Figure 7 ). MEBS resulted in higher droplet deposition on the upper (0.161µL cm^-2), middle (0.095µL cm^-2), and lower (0.074µL cm^-2) ESCL ( Figure 7 ). For example, the highest deposition of 150µm droplets found using RPA for the upper (0.009µL cm^-2), middle (0.006µL cm^-2), and lower (0.003µL cm^-2) ESCL were significantly different from those found using the MEBS (control), which was 94.40%, 93.68%, and 95.94% higher, respectively.

The 11.0 m application range resulted in the highest potential drift risk (PRD) for all ESCLs when using RPA, with significant differences of 38.82% and 15.33% higher compared to application ranges of 7.0 and 9.0 m, respectively. Contrastingly with the application ranges, which showed a directly proportional relationship with PRD, droplet size presented an inversely proportional relationship. The smallest droplet (150µm) resulted in a higher PRD for all ESCLs, whereas the largest droplet (450µm) showed a PRD lower than 13% for all layers ( Figure 7 ).

The control treatment (MEBS) showed the highest mean PRD for all ESCLs compared to treatments using RPA. The overall mean PRD for MEBS in all ESCLs was 51.65%, denoting a significant PDR during the applications, significantly different from treatments using RPA.

The application range significantly affected relative amplitude, but only in the middle and lower ESCL. The Anova indicated significant differences (p ≤ 0.001 and p ≤ 0.05) only for droplet size, volume distribution by droplet size class, and relative amplitude among ESCLs ( Table 4 ).

The application range showed no significant effect on variables related to volume distribution by droplet size class. DV_0.5, DV_0.1, and DV_0.9 were significantly and directly proportional to droplet size. Thus, individually, the larger the droplet size set on the RPA control (represented by DV_0.5) during the experimental treatments, the greater the mean droplet sizes (150, 300, and 450µm). The 9.0 m application range and 150µm droplet size combination presented the highest mean relative amplitude for all ESCLs, which directly affected the uniformity of droplet distribution.

Using the MEBS showed significant differences in variables related to volume distribution by droplet size class compared to treatments using RPA, exceeding 50% increase in these variables. However, the differences in mean relative amplitude were not significantly different in the different ESCLs when using RPA.

Considering the droplet size of DV_0.5, or volume median diameter (VMD), the means of canopy layers for the smallest droplet size set on the RPA control (150µm) was the closest to the actual droplet size (179.02µm) ( Table 5 ). However, the mean DV_0.5 of 300 and 450µm droplets differed from those set on the RPA during the experimental treatments. These results were confirmed by the analysis of droplet distribution using water-sensitive papers, considering the different treatments applied to the different eucalyptus canopy layers using RPA and MEBS ( Figure 8 ).

Regarding the risk of occupational exposure for applicators when using RPA, treatments with 150µm droplets (T1, T2, and T3) showed higher accumulation of droplets on upper limbs, regardless of application ranges ( Figure 9 ). This is confirmed by the highest total mean accumulated volume in the different body parts of applicators in T1, T2, and T3. For example, the 150µm droplet size with application ranges of 7.0 m (T1) and 9.0 m (T2) resulted in a higher volume deposited on the gloves, respirator, and legs, respectively. In the 11.0 m application range (T3), the deposits were limited to legs and gloves. This result was also found for treatment T4, which consisted of using 300µm droplets and 7.0 m application range.

Droplet depositions on different body parts are below the spectrophotometer’s detection limit, as evidenced in T4, indicating that the null absorbance readings were due to the low accumulation of droplets on different parts of applicators’ bodies when using RPA. This was found in all treatments using the largest droplets on the RPA (T7, T8, and T9).

Using an MEBS (control) resulted in a higher accumulation of droplets on all body parts compared to treatments using RPA ( Figure 10 ). The highest deposited volumes were found in gloves, respirator, legs, thighs, and back. Comparing the total highest volume accumulated in all body parts found for the control treatment and the RPA treatment 1 (0.0034µL cm^-2), MEBS presented deposits 0.544µL cm^-2 (99.37%) higher, corresponding to approximately 160-fold higher deposits of droplets on the applicators’ bodies, thus increasing the risk of occupational exposure for applicators.