Impact of Application Rate and Spray Nozzle on Droplet Distribution on Watermelon Crops Using an Unmanned Aerial Vehicle

Abstract

Watermelon is one of the most commonly grown vegetable crops worldwide due to the economic and nutritional importance of its fruits. The yield and quality of watermelon fruits are affected by constant attacks from pests, diseases, and weeds throughout all phenological stages of the crop. Labor shortages and unevenness of pesticide applications using backpack and tractor sprayers are significant challenges. The objective of this study was to evaluate the effect of different spray nozzles (XR110015 and MGA60015) and application rates (8, 12, and 16 L ha⁻¹) on droplet distribution on different targets in watermelon plants using an unmanned aerial vehicle. Water-sensitive papers were used as targets to analyze the droplet coverage, deposition, density, and volume median diameter. Data were collected from targets placed on the leaf adaxial and abaxial sides, fruit, apical bud, and stem of each plant. The mean droplet coverage and density increased as the application rate was increased, with no significant interaction between the factors or statistical difference between spray nozzles, except for the leaf abaxial side. No significant differences were found for the variables analyzed at application rates of 12 and 16 L ha⁻¹, whereas significant differences were observed at 8 L ha⁻¹. The use of unmanned aerial vehicles in watermelon crops is efficient; however, further studies should be conducted to evaluate their effectiveness in pest control and compare them with other application methods.

1. Introduction

Watermelon (Citrullus lanatus L.) is one of the most commonly grown vegetables in the world due to its economic importance and nutritional benefits. The fruit contains more than 90% water, considerable levels of vitamins A and C, a high lycopene content, and is widely consumed fresh [1]. More than 100 million Mg of watermelon are produced annually in the world, with an export value of EUR 2.02 billion in 2021 [2]. The largest watermelon-producing countries are China, Turkey, Iran, and Brazil [3].

Watermelon yield and quality are significantly impacted by constant attacks of pests, diseases, and weeds in all phenological stages of the crop [4,5]. Major pests affecting watermelon crops include underground pests such as the cutworm caterpillar (Agrotis ipsilon) and aerial pests such as aphids (Aphis gossypii), whitefly (Bemisia tabaci), and thrips (Frankliniella schultzei). Major diseases include anthracnose (Colletotrichum orbiculare), powdery mildew (Sphaerotheca fuliginea), downy mildew (Pseudoperonospora cubensis), and fusarium wilt (Fusarium oxysporum f. niveum).

Conventional methods for pest and disease control in watermelon crops are based on ground pesticide applications using backpack or tractor sprayers. However, these operations are sometimes inefficient and costly since watermelon has a creeping growth with countless herbaceous branches and large leaf areas covering the crop area, making it impossible for machinery and agricultural equipment to move between crop rows.

The use of new technologies to spray droplets with active ingredients on targets with greater efficiency is essential to achieve application efficiency. Despite unmanned aerial vehicles (UAVs) having been a suitable alternative for applications of pesticides and foliar fertilizers, there are some gaps or little information on droplet deposition quality for some crops, in which this technology is little explored. UAVs are used with pre-programmed flight plans and controlled by an operator at a ground station [6]. UAVs have advantages over conventional sprayers, including a low application rate, greater water savings, and a lower risk of applicator contamination; in addition, they are not limited by specific take-off and landing sites, topography, turning space, crop pattern, and canopy architecture, making them suitable for mountainous areas, muddy fields, and creeping crops with high leaf area indexes [7,8,9,10,11].

The spray nozzle, application rate, application route, application range, flight height, and flight speed are significant factors for UAV spraying [12]. They affect droplet deposition and in-flight planning, as the droplet spectrum and uniformity vary according to the canopy of plants. Research on UAV application technology is common for perennial (apple trees [13], peach [14], chestnut [15], arabica coffee [16], conilon coffee [17], vine [18], palm [19]) and annual (rice [20], corn [21], cotton [22], and weeds [23]) crops. Contrastingly, the droplet distribution and spray penetration in cucurbits vary, and some operational parameters used in the aforementioned studies are not applicable due to the irregular plant foliage and unpredictable branch growth direction.

Previous studies have investigated the impact of UAV application parameters on different crops. Spray nozzles and application rates significantly affected droplet deposition on different layers of papaya [24] and wheat [25] plants. The optimum spraying parameters for sugarcane crops were 15.0 L ha⁻¹, with a 3.0 m flight height and 4.0 m s⁻¹ flight speed [26]. An application route parallel to the planting line maximized droplet deposition on grapevine plants while minimizing endo-drift losses [18]. A flight height of 2.5 m and a flight speed of 3.0 m s⁻¹ increased droplet density on the leaf abaxial side of pumpkin plants [27]. Despite these findings, research on UAVs for creeping crops remains limited.

Therefore, the objective of this work was to fill theoretical gaps in the technical feasibility of using UAVs in watermelon crops. Three main gaps justify the present study:

(i) Research advances related to UAV operational parameters in the application of pesticides and fertilizers for annual (wheat, rice, corn, and cotton) and perennial (fruit trees and shrubs) crops are lacking for vegetable crops.

(ii) Fruit ripening is the most important stage of watermelon crops, when the control of the incidence of pests and diseases that occur in different plant parts (leaf, fruit, flower, and apical bud) hindering the final quality of the fruit is difficult. Therefore, despite studies that have shown the effects of different operational parameters (operational flight height and speed, application rate, spray nozzle, and flight route), they are not applicable to watermelon crops and should be investigated using different application rates and spray nozzles to test the application efficiency.

(iii) Watermelon crops present important specific characteristics that are not found in other studied crops, including the growing area, which is covered by primary and secondary herbaceous branches; plants covered with trichomes; fruit weights ranging from less than 1 kg to more than 20 kg (depending on the crop); and fruits initially developed below the leaf layer, which hinders the penetration of droplets to lower targets using conventional equipment.

Despite the economic importance of watermelon crops and the need for pest control during the crop cycle and the lack of information on UAV-based application technology in vegetable crops, droplet-deposition processes using UAVs in watermelon crops have never been investigated. Thus, the following hypotheses were tested to address these knowledge gaps: (a) the interaction between application rate and spray nozzles is significant for the optimal application configuration in watermelon crops; (b) droplet distribution by a UAV on watermelon crops has high potential to control fungal diseases and pests, depending on the target to be reached; (c) high application rates result in more uniform applications on watermelon crops.

Consequently, the objective of this study was to evaluate the effect of different spray nozzles and application rates on droplet distribution in different parts of watermelon plants using a UAV.

2. Materials and Methods

2.1. Characterization of the Area and Crop

The experiment was implemented, monitored, and evaluated at the Experimental Farm of the Northern University Center of Espírito Santo, Federal University of Espírito Santo, in São Mateus, ES, Brazil (18°40′25″ S, 40°51′23″ W) (Figure 1). The soil of the experimental area was classified as a Typic Hapludult of sandy loam texture. The climate of the region was classified as Aw, hot and humid, with a dry season in autumn–winter and a rainy season in spring–summer, according to the Köppen classification [28].

The experiment was conducted ins an experimental area of 2500 m² (50 × 50 m), grown with watermelon (Liverpool hybrid; Syngenta^®, São Paulo, Brazil), planted with spacing of 3.0 × 0.8 m, totaling a stand of 1042 plants in the experimental area.

2.2. Characterization of the Unmanned Aerial Vehicle

The unmanned aerial vehicle (UAV) was a DJI Agras, model T10 (DJI, SZ DJI Technology, Shenzhen, China) (Figure 2).

The UAV has a solution tank capacity of 10 L, which was previously adjusted and calibrated before application. In addition to the product tank, the UAV is equipped with a water pump, piping circuit for liquid circulation, electronic control, and valves. The spray nozzles of the UAV were placed equidistantly and perpendicularly to the aircraft axis, 1.0 m apart. The main specifications of the UAV are listed in

Table 1. Specifications of the DJI Agras UAV model T10.

Operating efficiency per hour	15 acres
Number of rotors	4
Maximum operational flight speed	0 to 7 m s−1
Maximum level flight speed	4 to 10 m s−1 (with strong GNSS signals)
Maximum bearable wind speed	0 to 8 m s−1
Tank capacity	10 L
Maximum effective spray width	3 to 5.5 m
Stationary flight duration	0 to 17 min
Maximum spraying flow	1.81 L/min−1
Number of nozzles	4

.

Before the implementation of the experiment, tests were carried out at different moments to define optimal flight operating parameters to be used as reference, since there is no scientific data on these parameters in the literature for cucurbits. In this sense, four plants were randomly selected within the experimental area, and five water-sensitive papers were placed on the plants using metal clips. Preliminary application tests were carried out at different flight heights (2, 3, and 4 m), speeds (3 and 5 m s⁻¹), and application ranges (3.5 and 5.0 m). Water-sensitive papers were then digitized in a wireless system and carefully analyzed to establish optimal parameters based on the visual distribution of the water-sensitive paper and the lowest risk of dripping into the soil. In this sense, the application rate and spray nozzles were fixed as experimental treatments, according to the proposed objective of the present study. After the preliminary tests, the defined parameters were average flight height of 3.0 m above the canopy of watermelon plants, flight speed of 5.0 m s⁻¹, average application range of 5.0 m, and application route perpendicular to the planting line. These parameters were kept constant and used for all treatments.

2.3. Experimental Design

The experiment was conducted in a randomized block design in a 3 × 2 factorial arrangement. The treatments consisted of different application rates (8.0, 12.0, and 16.0 L ha⁻¹) and spray nozzles (XR110015 flat jet nozzle (Teejet^®, Cotia, Brazil) and MGA60015 empty cone nozzle (Magnojet^®, Ibaiti, Brazil) (

Table 2. Experimental treatments.

Treatments	Spray Nozzle	Application Rate (L ha)−1	Flow Rate (L min)−1 a	Droplet Classification b
T1	MGA60015	8.0	1.18	Very thin
T2		12.0	1.78
T3		16.0	2.38
T4	XR110015	8.0	1.18	Thin
T5		12.0	1.78
T6		16.0	2.38

^a Flow rate corresponding to the application rate at 3 bar pressure; ^b droplet classification according to the nozzle manufacturer.

). The treatments were applied four times; each replication corresponded to a day of application, i.e., the variation of the blocks was the different days of application.

The experimental area was 2500 m². The treatments were applied in the total area, and all treatments were applied on the same day (block). The border of the area, 10.0 m from the edges, was disregarded to avoid the edge effect and allow some distance for the UAV sprayer to be activated.

The experimental unit for data sampling was a central area of 1500 m² (50 × 30 m), where most vigorous plants developed. Each treatment was applied to four target plants that jointly composed the experimental unit. A strip of 10.0 m was left between targeted plants. Each treatment had four plots, totaling 24 experimental units. The applications were carried out during the fruit-formation stage, at maximum crop development, and when the experimental area had full canopy cover. Figure 3 illustrates the experimental design used.

2.4. Variables Related to Application Efficiency

In all treatments, the sprayed solutions were composed of water mixed with brilliant blue dye (MarcAzul^®, Porto Alegre, Brazil) at a rate of 1.25 mL L⁻¹ and non-siliconized adjuvant based on balanced polymers, specific for low-volume aerial applications (0.2% v v⁻¹) (Helper Air^®, ICL, São Paulo, Brazil).

Water-sensitive paper with dimensions of 76 × 26 mm was used for collecting data on droplet coverage, deposition, density, and volume median diameter. The targets were placed on different parts of each plant used for data collection: leaf adaxial and abaxial sides, fruit, apical bud, and stem. The productive branch with the highest vigor of each target plant was previously selected. Water-sensitive papers were fixed using wooden clips on the leaf adaxial (Figure 4a) and abaxial (Figure 4b) sides and using double-sided tape on fruits (Figure 4c), apical buds (Figure 4d), and stems (Figure 4e); therefore, 5 water-sensitive papers were placed on each target plant, totaling 20 water-sensitive papers per experimental plot and 80 water-sensitive papers per treatment. Despite its limitations in detecting very fine droplets, water-sensitive paper is still employed because of its ease of use in obtaining droplet spectrum data.

The water-sensitive papers were removed five minutes after application of each treatment to allow the solution to evaporate, maintaining only the dye. They were then packed in labeled kraft paper bags and analyzed for droplet characterization on the same day. The analyses were performed at the Laboratory of Mechanization and Agricultural Defensives (LMDA) of the Northern University Center of Espírito Santo, Federal University of Espírito Santo, São Mateus, ES, Brazil.

A wireless system (DropScope^®; SprayX, São Carlos, Brazil) was used to scan the water-sensitive papers for data acquisition and analysis. This system is composed of application programs and a wireless digital microscope with a digital image sensor of more than 2500 dpi, which allows the estimation of partially overlapping droplets of approximately 35 µm (Figure 5). Recent studies have demonstrated the reliability of the data obtained by this system in analyses of spectrum of droplets sprayed by UAV [24,29,30]. After scanning the water-sensitive papers, data were generated, and the following parameters were evaluated: droplet coverage (%), density (droplets cm⁻²), deposition (µL cm⁻²), and volume median diameter (µm).

2.5. Climate Conditions

The applications were carried out in late afternoon. Wind speed, air humidity, and temperature were recorded during the applications (

Table 3. Climate conditions at the time of applications.

Application Date	Time	Temperature (°C)		Relative Humidity (%)		Wind Speed (m s)−1		Wind Direction
		Minimum	Maximum	Minimum	Maximum	Minimum	Maximum
30 May 2023	16:00	26.0	27.3	60.0	65.0	3.1	6.7	East-northeast.
	17:00	24.8	26.0	63.0	69.0	2.7	6.3
	18:00	24.0	24.8	69.0	75.0	1.8	4.5
	19:00	23.0	24.0	75.0	85.0	0.9	3.6
1 June 2023	16:00	26.8	28.8	68.0	77.0	1.2	3.2
	17:00	25.5	26.8	77.0	82.0	1.1	3.9
	18:00	24.2	25.6	82.0	87.0	1.9	4.6
	19:00	23.3	24.2	87.0	91.0	0.8	3.7
5 June 2023	16:00	25.9	27.1	55.0	61.0	1.30	3.6
	17:00	24.1	25.9	60.0	70.0	1.7	3.1
	18:00	21.8	24.1	70.0	79.0	1.5	1.4
	19:00	21.1	21.8	79.0	82.0	0.4	2.3
7 June 2023	16:00	25.3	26.6	63.0	72.0	1.3	5.4
	17:00	23.9	25.4	71.0	79.0	2.1	3.1
	18:00	22.5	23.9	79.0	85.0	1.8	1.2
	19:00	21.2	22.5	84.0	89.0	0.4	2.1

) using an automatic weather station Davis^®; Vantage Pro2 wireless K6152 (Davis Instruments, Hayward, CA, USA) installed at approximately 100 m from the experimental area. The applications were carried out considering the methodology described by the International Organization for Standardization (Standard 22866) [31], which recommends temperature during applications between 5 and 35 °C, a maximum of 10% of wind speed measurements below 1.0 m s⁻¹, and wind direction within 90° ± 30° in relation to the application line.

2.6. Statistical Analysis

The Shapiro–Wilk test was used to assess the normality of residuals. Data related to application efficiency were subjected to analysis of variance. When the effects of factors or their interactions were significant, means were compared using Tukey’s test.

3. Results

The results of the analysis of variance for variables related to application efficiency indicated statistical differences in the application rates (

Table 4. Analysis of variance (Anova) and normality of residuals for droplet coverage on different watermelon plant parts.

Droplet coverage (%)
Analysis of variance
	Leaf adaxial side		Leaf abaxial side		Fruit		Apical Bud		Stem
Factor	p-value	CV (%)	p-value	CV (%)	p-value	CV (%)	p-value	CV (%)	p-value	CV (%)
T	<0.01 **	39.78	0.80 ns	117.68	<0.001 ***	23.43	<0.01 **	30.11	<0.001 ***	38.08
P	0.36 ns		0.12 ns		0.21 ns		0.66 ns		0.054 ns
T × P	0.80 ns		0.80 ns		0.44 ns		0.98 ns		0.91 ns
Normality of residuals
W=	0.753 ns		0.897 ns		0.667 ns		0.990 ns		0.986 ns

P = spray nozzles; T = application rate; T × P = interaction between spray nozzle and application rate; ** = significant at p < 0.01; *** = significant at p < 0.001; ^ns = not significant; CV (%) = coefficient of variation of Anova; W = normality of residuals by the Shapiro–Wilk test.

). The interaction between the factors (spray nozzle and application rate) was not significant for any of the variables. Thus, each factor was analyzed separately within the proposed targets.

3.1. Droplet Coverage

The application rate was significant for droplet coverage on all watermelon plant targets, except for the leaf abaxial side (Table 4). The spray nozzle effect and the interaction effect between factors were not significant for any target, indicating that all treatments were statistically similar. The normality of residuals was confirmed by the Shapiro–Wilk test, with values ranging from 0.667 to 0.990. Normality tests yield important assumptions before analyses for greater accuracy of the data obtained.

The means found for the spray nozzle factor within the targets were significantly similar to Tukey’s test and confirmed by an Anova (

Table 5. Droplet coverage (%) on watermelon plant parts (leaf adaxial side—LAD, leaf abaxial side—LAB, fruit—FT, apical bud—AB, and stem—ST) using different spray nozzles and application rates. Mean values are followed by standard deviation.

Targets	Spray Nozzles
	MGA60015		XR110015
LAD	1.48 ± 0.70 a		1.73 ± 0.86 a
LAB	0.46 ± 0.42 a		0.20 ± 0.25 a
FT	0.72 ± 0.43 a		0.82 ± 0.34 a
AB	1.44 ± 0.72 a		1.52 ± 0.50 a
ST	0.32 ± 0.20 a		0.45 ± 0.21 a
Targets	Application rates (L ha)−1
	8.0	12.0	16.0
LAD	0.95 ± 0.44 b	1.64 ± 0.50 ab	2.22 ± 0.78 a
LAB	0.33 ± 0.47 a	0.38 ± 0.35 a	0.25 ± 0.29 a
FT	0.50 ± 0.20 b	0.83 ±0.30 a	1.00 ± 0.47 a
AB	0.92 ± 0.36 b	1.66 ± 0.49 a	1.87 ± 0.53 a
ST	0.18 ± 0.07 b	0.40 ± 0.18 a	0.58 ± 0.15 a

Means followed by different letters in the rows are significantly different from each other according to Tukey’s test at p ≤ 0.05.

). However, the XR10015 nozzle presented the highest means in the leaf adaxial side (LAD), fruit (FT), and stem (ST), with a difference in droplet coverage of 0.25, 0.10, and 0.13 percentual points, equivalent to 14.45%, 12.20%, and 28.90%, respectively, compared to the MGA60015 nozzle. On the leaf abaxial side, the MGA60015 nozzle presented the highest mean value, with a difference equivalent to 56.52%, 2.3-fold higher compared to the XR110015 nozzle.

Droplet coverage on LAD, FT, apical bud (AB), and ST increased as the application rate was increased. The application rate of 8.0 L ha⁻¹ presented lower droplet coverage, and 12 and 16.0 L ha⁻¹ were statistically similar, except for the leaf abaxial side (LAB), which did not differ statistically between application rates. However, the intermediate application rate (12.0 L ha⁻¹) presented the highest droplet coverage on LAB.

The application rate of 16.0 L ha⁻¹ presented the highest droplet coverage on LAD, being 26.12% and 57.20% higher compared to the rates of 12 and 8.0 L ha⁻¹, respectively. Similar results were found for the application rate of 16.0 L ha⁻¹ in the other targets. The highest application rate was 17.0% and 50.0% higher on FT compared to the rates of 12 and 8.0 L ha⁻¹; 11.22% and 50.80% higher on AB compared to the rates of 12 and 8.0 L ha⁻¹; and 31.03% and 69.00% higher on ST compared to the rates of 12 and 8.0 L ha⁻¹. The differences between 16.0 and 12.0 L ha⁻¹ did not reach 50.0%, presenting statistical similarity; in contrast, the application rate of 8.0 L ha⁻¹ resulted in differences greater than 50.0% in droplet coverage compared to the other application rates.

3.2. Droplet Density

The application rate factor significantly affected (p < 0.05) droplet density on the LAD, AB, and ST (

Table 6. Analysis of variance (Anova) and normality of residuals for droplet density on different watermelon plant parts.

Droplet Density (Droplet cm−2)
Analysis of variance
	Leaf adaxial side		Leaf abaxial side		Fruit		Apical bud		Stem
Factor	p-value	CV (%)	p-value	CV (%)	p-value	CV (%)	p-value	CV (%)	p-value	CV (%)
T	<0.05 *	42.33	0.95 ns	121.21	0.79 ns	77.40	<0.05 *	38.29	<0.05 *	55.23
P	0.36 ns		0.12 ns		0.50 ns		0.76 ns		0.95 ns
T × P	0.85 ns		0.85 ns		0.65 ns		0.63 ns		0.51 ns
Normality of residuals
W=	0.252 ns		0.808 ns		0.215 ns		0.530 ns		0.043 ns

P = spray nozzles; T = application rate; T × P = interaction between spray nozzle and application rate; * = significant at p < 0.05; ^ns = not significant; CV (%) = coefficient of variation of Anova; W = normality of residuals according to the Shapiro–Wilk test.

). All residuals were normally distributed according to the Shapiro–Wilk test. The interaction between the application rate and the spray nozzle was not significant.

The MGA60015 spray nozzle presented the highest mean droplet density on LAB (

Table 7. Droplet density (droplets cm⁻²) on watermelon plant parts (leaf adaxial side—LAD; leaf abaxial side—LAB; fruit—FT; apical bud—AB; and stem—ST) using different spray nozzles and application rates. Mean values are followed by standard deviation.

Targets	Spray nozzles
	MGA60015		XR110015
LAD	33.04 ± 15.96 a		38.91 ± 23.40 a
LAB	11.95 ± 10.15 a		4.92 ± 6.57 a
FT	17.00 ± 8.26 a		21.24 ± 19.47 a
AB	32.53 ± 13.00 a		36.28 ± 21.42 a
ST	8.37 ± 4.01 a		8.49 ± 6.78 a
Targets	Application rates (L ha)−1
	8.0	12.0	16.0
LAD	23.91 ± 10.24 b	34.50 ± 15.46 ab	49.52 ± 23.83 a
LAB	8.18 ± 9.59 a	9.30 ± 9.21 a	7.80 ± 9.70 a
FT	17.26 ± 7.47	18.13 ± 22.75	21.97 ± 11.41
AB	24.71 ±11.18 b	33.91 ±12.51 ab	47.59 ± 20.13 a
ST	4.41 ± 1.70 b	9.52 ± 6.77 ab	11.35 ± 4.39 a

Means followed by different letters in the rows are significantly different from each other according to Tukey’s test at p ≤ 0.05.

), with a difference of 7.02 droplets cm⁻², equivalent to 58.80%, 2.43-fold higher compared to the XR110015 nozzle. The XR110015 nozzle resulted in the highest number of droplets on the other targets, However, it presented means close to those of the MGA60015, not differing statistically.

The number of droplets per square centimeter increased on all watermelon plant parts as the application rates increased, except for LAB. The application rates of 16.0 and 12.0 L ha⁻¹ were statistically similar for LAD, AB, and ST.

The mean values and the impacts of droplets on water-sensitive papers (Figure 6) showed increases in the mean droplet coverage and density, according to the accessibility of the targets. The highest mean values were found for LAD and AB on the upper part of watermelon plants. The highest application rate (16.0 L ha⁻¹) resulted in a higher droplet density on LAD, with a difference of 15.02 and 25.61 droplet cm⁻² compared to the application rates of 12.0 and 8.0 L ha⁻¹, respectively. The highest application rate resulted in a 28.74% and 48.07% higher droplet density on AB compared to the application rates of 12.0 and 8.0 L ha⁻¹, respectively.

Watermelon plant parts with more obstacles (FT, LAB, and ST) presented lower mean droplet densities. The application rate of 16.0 L ha⁻¹ showed a 17.47% and 21.43% higher droplet density on FT compared to the application rates of 12.0 and 8.0 L ha⁻¹, respectively. The differences between application rates on FT were less than 50% and not significant according to the Anova. The differences between the application rates of 12.0 and 8.0 L ha⁻¹ on ST were 1.83 and 6.94 droplet cm⁻², respectively. The application rate of 12.0 L ha⁻¹ resulted in higher droplet density on LAB, which was 12.04% and 16.12% higher compared to the rates of 8 and 16.0 L ha⁻¹, respectively. The application rates had a similar effect on droplets on water-sensitive papers on LAB (Figure 6), with no difference between application rates, regardless of the spray nozzle type, as indicated by the Anova and Tukey test.

3.3. Droplet Deposition

The application rate was significant for droplet deposition on all targets, except for the leaf abaxial side (

Table 8. Analysis of variance (Anova) and normality of residuals for droplet deposition on different watermelon plant parts.

Droplet deposition (µL cm)−2
Analysis of variance
Factor	Leaf adaxial side		Leaf abaxial side		Fruit		Apical bud		Stem
	p-value	CV (%)	p-value	CV (%)	p-value	CV (%)	p-value	CV (%)	p-value	CV (%)
T	<0.05 *	43.07	0.58 ns	110.97	<0.01 **	27.96	<0.01 **	28.92	<0.01 **	49.37
P	0.26 ns		0.15 ns		0.21 ns		0.31 ns		0.09 ns
T × P	0.74 ns		0.76 ns		0.37 ns		0.90 ns		0.81 ns
Normality of residuals
W=	0.580 ns		0.809 ns		0.353 ns		0.694 ns		0.448 ns

P = spray nozzles; T = application rate; T × P = interaction between spray nozzle and application rate; * = significant at p < 0.05; ** = significant at p < 0.01; ^ns = not significant; CV (%) = coefficient of variation of Anova; W = normality of residuals according to the Shapiro–Wilk test.

). The spray nozzle factor and the interaction between application rate and spray nozzle were not significant, rejecting the hypothesis of difference between treatments. All residuals were normally distributed according to the Shapiro–Wilk test. The coefficients of variation of the Anova for the targets ranged from 27.96% to 110.97%, indicating a possible effect of the UAV operational conditions connected to external environmental factors but not invalidating the statistical analyses.

Similarly to droplet coverage and density, the mean droplet deposition increased as the application rate increased, with no significant interaction between the factors or statistical difference between spray nozzles, except for the leaf abaxial side. However, application rates of 12.0 and 16.0 L ha⁻¹ resulted in a statistically similar droplet deposition, significantly differing from the rate of 8.0 L ha⁻¹.

The application rate of 16.0 L ha⁻¹ resulted in the highest mean droplet deposition on LAD (0.065 µL cm⁻²) (Figure 7a), being 23.07% and 53.85% higher compared to the 12.0 and 16.0 L ha⁻¹, respectively. The XR110015 nozzle presented an increase of 0.010 µL cm⁻² in droplet deposition on LAD compared to the MGA60015 nozzle. The application rate of 12.0 L ha⁻¹ was 25.0% and 50.0% higher on LAB compared to 8.0 and 16.0 L ha⁻¹, respectively (Figure 7b). The MGA60015 nozzle resulted in a 2.0-fold increase (0.006 µL cm⁻²) in droplet deposition on LAB compared to the XR110015 nozzle.

Droplet deposition on FT (Figure 7c) and AB (Figure 7d) showed similar results to those on LAD, with 16.0 L ha⁻¹, and the XR110015 spray nozzle presented the highest means. The stem is the plant part that experiences great interference by plant parts and is located close to the ground level; however, the application rate of 16.0 L ha⁻¹ also presented an increase in droplet deposition, which was 0.005 and 0.0013 µL cm⁻² higher compared to those of 12.0 and 8.0 L ha⁻¹, respectively (Figure 7e).

3.4. Volume Median Diameter

The application rate and spray nozzle factors were significant for the volume median diameter (VMD) on the leaf adaxial side, but with no significant interaction effect between factors (

Table 9. Analysis of variance (Anova) and normality of residuals for volume median diameter on different watermelon plant targets.

Volume median diameter (µm)
Analysis of variance
Factor	Leaf adaxial side		Leaf abaxial side		Fruit		Apical bud		Stem
	p-value	CV (%)	p-value	CV (%)	p-value	CV (%)	p-value	CV (%)	p-value	CV (%)
T	<0.01 **	6.90	0.56 ns	37.60	0.44 ns	9.38	0.109 ns	9.11	0.67 ns	33.34
P	<0.05 *		0.65 ns		<0.05 *		0.37 ns		0.90 ns
T × P	0.95 ns		0.42 ns		0.79 ns		0.08 ns		0.33 ns
Normality of residuals
W=	0.170 ns		0.401 ns		0.300 ns		0.101 ns		0.015 ns

P = spray nozzles; T = application rate; T × P = interaction between spray nozzle and application rate; * = significant at p < 0.05; ** = significant at p < 0.01; ^ns = not significant; CV (%) = coefficient of variation of Anova; W = normality of residuals according to the Shapiro–Wilk test.

). Only the spray nozzle was significant for the fruit. The factors and their interaction had no significant effect on VMD for LAB, AB, or ST.

The coefficients of variation (Anova) for the targets ranged from 6.90% to 37.60%, indicating the accuracy of the experimental design in obtaining the data. The residuals from the ST data were not normally distributed based on the Shapiro–Wilk test.

The XR110015 spray nozzle resulted in the largest droplet diameters on all watermelon plant parts, ranging from 171.83 to 206.38 µm, and was significant for LAD and ST, as indicated by the Anova. VMD on the watermelon plant parts varied among application rates, with the application rate of 12.0 L ha⁻¹ presenting higher means compared to 8.0 and 16.0 L ha⁻¹ (

Table 10. Volume median diameter (µm) on watermelon plant parts (leaf adaxial side—LAD, leaf abaxial side—LAB, fruit—FT, apical bud—AB, and stem—ST) using different spray nozzles and application rates. Mean values are followed by standard deviation.

Targets	Spray nozzles
	MGA60015		XR110015
LAD	188.43 ± 17.54 b		202.24 ± 20.45 a
LAB	160.21 ± 59.02 a		171.83 ± 69.01 a
FT	186.74 ± 18,24 b		206.38 ± 23.43 a
AB	190.91 ± 21.45 a		197.66 ± 18.65 a
ST	210.08 ± 21.98 a		213.72 ± 20.35 a
Targets	Application rates (L ha)−1
	8.0	12.0	16.0
LAD	181.93 ± 12.73 b	208.88 ± 23.86 a	195.20 ± 12.45 ab
LAB	170.00 ± 78.65 a	180.85 ± 41.48 a	147.22 ± 67.02 a
FT	200.00 ± 27.51 a	200.17 ± 24.81 a	190.00 ± 16.04 a
AB	188.13 ± 25.74 a	206.50 ± 10.35 a	188.23 ± 16.84 a
ST	193.31 ± 24.00 a	212.37 ± 41.03 a	195.71 ± 24.06 a

Means followed by different letters in the rows are significantly different from each other according to Tukey’s test at p ≤ 0.05.

).

There was little variation in VMD among the different targets. The largest interquartile differences (∆Q) were found for LAB (Q₁ = 146.70 µL; Q₃ = 196.50 µL). The difference in the third quartile (Q₃) between the stem (219.00 µm) and leaf abaxial side (196.50 µm) was 22.50 µL, equivalent to 10.27% (Figure 8).

The maximum and minimum droplet diameters in the targets were 158.30 and 243.70 µm (LAD), 116.36 and 268.10 µm (LAB), 163.50 and 238.70 µm (FT), 153.20 and 243.70 µm (AB), and 163.50 and 277.70 µm (ST), respectively. Overall, no outliers were found (VMD < 300.00 µm), regardless of the target, i.e., there was no value outside of the interquartile detection range.

4. Discussion

The results indicate that application rates had a significant effect on droplet coverage, density, and deposition on the leaf adaxial side, fruit, apical bud, and stem (ST) of watermelon plants, with increasing means as the application rate was increased. Similar results were found for large [32,33] and small [34] shrub crops and perennial grass [26].

Watermelon plants have a creeping growth habit with no defined canopy layers; thus, droplet distribution on watermelon plants is variable when using UAVs, unlike perennial and annual plants [17,30,35,36]. This explains the absence of significant interaction between the application rate and spray nozzle for droplet coverage, density, deposition, and volume median diameter (VMD), as recently found for papaya plants with the use of different application rates and spray nozzles (MGA60015 and XR110015) [24].

Although the highest application rate (16.0 L ha⁻¹) increased droplet density and deposition, the current investigation evaluated only the efficiency of the application using a dye tracer since the droplets sprayed by UAVs are highly concentrated. Therefore, the highest application rates may not always provide the best control efficacy for the pesticide. In the control of diseases in Coffea arabica plants [37] and pests in wheat plants [25], the rates applied using a UAV achieved the same control efficacy as high rates applied using conventional sprayers.

Considering the limitations of battery life and load capacity (8–40 L) [38,39], the UAV operational capacity tends to increase by reducing application rates without losing gains in pest control. The application rate of 12.0 L ha⁻¹ did not differ from 16.0 L ha⁻¹ for most variables, regardless of the spray nozzle, presenting satisfactory results on the leaf adaxial side, fruit, and apical bud, which experienced no interference from watermelon ramifications and the overlapping of leaves. This indicates the potential for controlling diseases that grow on the leaf adaxial side, such as Pseudoperonospora cubensis and Cercospora citrullina, and attacks from chewing and sucking insects on fruits with contact insecticides.

The apical bud is one of the main targets covered in watermelon plants, as it hosts sucking insects in its tissues covered by trichomes, including Frankliniella schultzei, which is found in the apical bud and inside flowers. The droplet density and deposition resulting from the application rates of 12.0 and 16.0 L ha⁻¹ using the XR110015 spray nozzle were promising for the apical bud and stem (where flowers grow). Higher values were found when using the fan nozzle on these plant parts due to the transverse distribution of the spray mixture, symmetrical to the nozzle center, reaching the targets with larger droplets [40]. When using flat jet/fan nozzles and UAVs for perennial crops, droplet distribution was greater in the upper layer and lower in the lower plant layer [41,42,43].

The leaf abaxial side (LAB) has a greater number of stomata and thinner cuticles; thus, it is one of the main pathogen infection sites. Additionally, the LAB of watermelon plants hosts sucking pests such as Aphis gossypi and Bemisia tabaci. The higher droplet coverage, density, and deposition on LAB by the MGA60015 nozzle at a rate of 12.0 L ha⁻¹ was due to the greater turbulence in droplet distribution. The fine droplet distribution at the cone’s outer edge, caused by the turbulence chamber within the nozzle [44] and combined with the downwash effect, facilitates droplet penetration into the lower plant layers [45]. The use of a UAV equipped with conical nozzles for pumpkin crops showed that the operational flight height and speed were significant in increasing deposition and droplet density on LAB [27]. The application rates had no significant effect on the leaf abaxial side using a UAV on watermelon plants, as observed in a laboratory study [46]. Moreover, the angle between the leaf and the airflow is an important factor for droplet deposition on LAB [47].

The VMD found for the XR110015 spray nozzle was higher, as the production of larger droplets is a typical characteristic of flat jet nozzles. The MGA60015 spray nozzle resulted in droplets with smaller diameters, a desirable characteristic of conical nozzles because droplets easily penetrate the canopy of plants with high leaf density [48]. This results in higher values of variables related to the application efficiency for the LAB.

The mean VMD found on the leaf abaxial side, fruits, and stems, which have greater plant barriers (branching and overlapping leaves), indicated that these plant parts should be sprayed with fine droplets to enhance droplet deposition, regardless of the application rate and spray nozzle. However, the risks related to primary and secondary drift and evaporation of droplets under adverse climate conditions increase as the droplet size is reduced [49]. Overall, the VMD values found are consistent with the manufacturers’ catalog, which describes fine to medium droplets [50,51].

Although the results denoted the variability of droplet distribution on different watermelon plant parts, further studies should be carried out to confirm the viability of using UAVs in watermelon crops. Nevertheless, several operational parameters (operational flight height and speed, application rates, application range, and spray points) should be tested at different stages of watermelon plants to improve the results related to application efficiency. In addition, the effectiveness of pesticides in pest and disease control and foliar fertilizers should be considered to enhance and leverage the use of UAVs in vegetable crops. Moreover, the importance of evaluating the efficacy of applied products as a parameter to complement assessments of the quality of droplet deposition on target crops has been highlighted [17,37,52,53,54].

5. Conclusions

The use of unmanned aerial vehicles (UAVs) for spraying on agricultural crops is expanding due to their numerous advantages compared to conventional equipment. However, there are no operational parameters in the technical–scientific literature that elucidate the feasibility of using UAVs for applications to watermelon crops. Nevertheless, the main objective of this study was to evaluate the effect of spray nozzles and application rates on droplet distribution on different targets in watermelon plants using a UAV. The results found after applying the methodology, data collection, and analysis of the results obtained showed the following:

• Application rates significantly affect droplet distribution on the leaf adaxial and abaxial sides, fruit, apical bud, and stem of watermelon plants, with variable droplet distribution depending on the target.
• Droplet density and deposition on the leaf adaxial side, fruit, apical bud, and stem were higher when using the XR110015 spray nozzle. Droplet density and deposition on the leaf abaxial side were 58.80% and 50.0% higher, respectively, when using the MGA60015 spray nozzle.
• Application rates of 12.0 and 16.0 L ha⁻¹ can be used as a reference for evaluations of the efficacy of pesticides, as they resulted in statistically similar means regarding application efficiency.
• The results demonstrated that the use of UAVs for applications to watermelon crops is efficient. However, after understanding the droplet-deposition process through the application efficiency, further studies should be conducted to evaluate the efficacy of this technology for controlling pests and diseases in watermelon crops, with comparisons between applications using UAV and conventional equipment at different phenological stages of watermelon plants.