Precision management of Spodoptera frugiperda infestation hotspots in Bt and non-Bt corn crops with bioproduct applied and monitored via drone

Manejo de precisão de hotspots de infestação de Spodoptera frugiperda na cultura do milho Bt e não Bt com bioinsumo aplicado e monitorado via drone

Abstract

Spodoptera frugiperda (J.E. Smith) (Lepidoptera: Noctuidae) is a polyphagous pest widely distributed in tropical and subtropical regions, with reports of occurrence in various areas of the world. It encompasses 357 host plants, with emphasis on soybean (Glycine max L.), maize (Zea mays L.), and cotton (Gossypium hirsutum L.). The damage caused by this species severely compromises agricultural productivity and generates significant economic losses. Traditional management, based on intensive use of synthetic chemical insecticides, has had negative impacts on agroecosystems and on selected resistant populations. In this scenario, the use of drones and bioproducts emerges as a valuable tool for adequate sampling and sustainable management. Considering that NDVI versus NDRE comparison under real field conditions is still necessary, and the field-level validation of these spectral indices associated with bioproduct-based management in Bt and non-Bt corn remains limited, the present study aimed to detect, at the field level, the spectral reflectance of Bt and non-Bt corn plants infested with S. frugiperda larvae using the NDVI and NDRE vegetation indices derived from multispectral imaging acquired by remotely piloted aircraft (RPA). The experiment was conducted at the Chã de Jardim Experimental Area, UFPB, in Areia-PB, under a randomized block design (RBD). The varieties B2433PWU, Robusto, and Jabatão were evaluated. The treatments consisted of the application of Fitoneem® (active ingredients Azadirachtin A and B) and control (water) via DJI T40 spray drone. The results showed that leaf injuries can be efficiently monitored using remote sensing, with the NDVI index being the most effective. Infestation was influenced by both the cultivar and the applied treatment. In the Jabatão variety, applications of Azadirachtin reduced the expression of damage by up to 95%. Regarding production variables, only the number of grains per row and the number of grains per ear were significantly affected by the treatments. In practice, our research shows that the spectral reflectance of Bt and non-Bt corn plants infested with S. frugiperda larvae, as measured by the NDVI, is effective for validating the spectral response associated with bioproduct-based management in Bt and non-Bt corn fields.

Keywords:
Azadirachtin; remote sensing; multispectral imaging; vegetation index; IPM

Resumo

A Spodoptera frugiperda (J.E. Smith) (Lepidoptera: Noctuidae) é uma praga polífaga amplamente distribuída em regiões tropicais e subtropicais, com registros de ocorrência em diversas regiões do mundo e um espectro de 357 plantas hospedeiras, com destaque para as culturas de soja (Glycine max L.), milho (Zea mays L.) e algodão (Gossypium hirsutum L.). Os danos causados por essa espécie comprometem severamente a produtividade agrícola e geram perdas econômicas significativas. O manejo tradicional, baseado no uso intensivo de inseticidas químicos sintéticos, tem tido impactos negativos nos agroecossistemas e em populações resistentes selecionadas. Nesse cenário, o uso de drones e bioprodutos surge como uma ferramenta valiosa para amostragem adequada e manejo sustentável. Considerando que a comparação entre NDVI e NDRE em condições reais de campo ainda é necessária, e que a validação em campo desses índices espectrais associados ao manejo baseado em bioprodutos em milho Bt e não-Bt permanece limitada, o presente estudo teve como objetivo detectar, em campo, a reflectância espectral de plantas de milho Bt e não-Bt infestadas com larvas de S. frugiperda utilizando os índices de vegetação NDVI e NDRE derivados de imagens multiespectrais adquiridas por drones. O experimento foi conduzido na Área Experimental Chã de Jardim, UFPB, em Areia-PB, em delineamento de blocos casualizados (DBC). As variedades avaliadas foram B2433PWU, Robusto e Jabatão. Os tratamentos consistiram na aplicação de Fitoneem® (ingredientes ativos Azadiractina A e B) e controle (água) via drone pulverizador DJI T40. Os resultados mostraram que os danos foliares podem ser monitorados eficientemente por sensoriamento remoto, sendo o índice NDVI o mais eficaz. A infestação foi influenciada tanto pela cultivar quanto pelo tratamento aplicado. Na variedade Jabatão, as aplicações de Azadiractina reduziram a expressão dos danos em até 95%. Em relação às variáveis ​​de produção, apenas o número de grãos por fileira e o número de grãos por espiga foram significativamente afetados pelos tratamentos. Na prática, nossa pesquisa mostra que a reflectância espectral de plantas de milho Bt e não-Bt infestadas com larvas de S. frugiperda, medida pelo NDVI, é eficaz para validar a resposta espectral associada ao manejo baseado em bioprodutos em lavouras de milho Bt e não-Bt.

Palavras-chave:
Azadiractina; sensoriamento remoto; imagem multiespectral; índice de vegetação; MIP

1. Introduction

Corn (Zea mays L.) is a crop of global importance, ranking among the principal agricultural products. It has high nutritional value for both animal and human diets, making it a food of considerable importance (Sologuren, 2015; Maximiano, 2017). The 2024/2025 harvest in Brazil ends with an estimated corn production of 119.7 million tons, an increase of 3.5% from the previous cycle (CONAB, 2024). Multiple pests are found in corn production in Brazil, such as the fall armyworm (Spodoptera frugiperda) (JE Smith) (Lepidoptera: Noctuidae).

Synthetic insecticides are often used to manage S. frugiperda and corn varieties expressing the bacterium Bacillus thuringiensis (Bt). However, there is a growing demand for biological or natural pesticides (Nascimento et al., 2025). Sisay et al. (2019) highlight that botanical insecticides are a promising alternative for managing S. frugiperda, mainly because they combine biological efficacy with lower environmental impact than synthetic compounds. In their experiments, they found that extracts of Azadirachta indica, Schinus molle, and Phytolacca dodecandra exhibited vigorous insecticidal activity, resulting in mortality exceeding 95% within the first 72 hours of exposure and significantly reducing leaf damage in corn plants. These findings reinforce the role of plant-based insecticides as complementary tools in Integrated Pest Management (IPM) programs, expanding control options and reducing dependence on the continuous use of conventional chemical agents.

The intensification of corn cultivation in regions under high pressure from S. frugiperda has favored rapid selection for resistant genotypes, compromising the performance of Bt events and increasing management costs. Recent studies show that changes in receptor genes—such as variants in SfABCC2—are directly associated with reduced sensitivity to Cry proteins, leading to control failures even in systems with intensive insecticide use. Genomic research and population monitoring indicate that these alleles can arise locally and disperse, reinforcing the need for continuous molecular surveillance. In addition, recent reviews highlight that biological agents, as well as the proper integration of Integrated Pest Management tactics, are viable alternatives for slowing the evolution of resistance and sustaining productivity in critical areas (Banerjee et al., 2022; Abbas et al., 2022; Tandy et al., 2023). In addition, remote sensing is a valuable tool for obtaining information about the Earth's surface and atmosphere. In this context, there are several methods, including suborbital remote sensing, an innovative and effective approach that delivers high-resolution images enabled by multispectral sensors (Radoglou-Grammatikis et al., 2020; Wolff et al., 2022).

The Normalized Difference Vegetation Index (NDVI) is one of the most widely used and established spectral indicators in vegetation studies. This index is calculated from the normalized ratio of the reflectance recorded in the near-infrared (NIR) band, which is associated with the scattering of radiation by the internal structures of the leaf mesophyll, to the reflectance of the red band, which is predominantly absorbed by chlorophyll pigments (Bhattarai et al., 2019). Meanwhile, the RedEdge band Normalized Difference Index (NDRE) has been gaining prominence for its greater sensitivity in detecting physiological stresses in plants, particularly at advanced phenological stages, where the NDVI tends to saturate (Boris and Hideo, 2019).

In the context of Agriculture 5.0/6.0, both remote-piloted aircraft and spray drones are valuable tools. Spray drones are a technology that has been gaining market share and visibility in the Brazilian agricultural sector. It is an activity that has been gaining ground in the market and continually improving its procedures to better serve rural producers, enabling the efficient application of various products, such as fertilizers, granular pesticides, herbicides, and other agricultural inputs (Silva et al., 2025a, Silva et al., 2025b). This innovation seeks to improve the precision, efficiency, and sustainability of modern agriculture. However, some obstacles must be overcome to make this technology more accessible, namely, equipment costs and the shortage of skilled labor.

Despite advances in biological control and remote sensing, field-level integration of bioproduct application via drones with spectral validation in Bt and non-Bt maize remains underexplored. Based on the current contextualization, this research tested the central hypothesis that the neem-oil-based bioproduct tested will be sufficient to promote population reduction; in addition, NDVI is expected to decrease with increasing S. frugiperda infestations, which, in turn, will be influenced by the corn cultivar and the application of bioproducts via drone spraying. The overall objective of this study was to detect, at the field level, the spectral reflectance of Bt and non-Bt corn plants infested by S. frugiperda larvae, using the NDVI and NDRE vegetation indices generated by multispectral images from remotely piloted aircraft, to evaluate the action of bioproducts that was applied to different corn varieties using a spray drone. We also verified the relationship between pest infestation levels and the NDVI and NDRE vegetation indices to assess whether these indices vary with bioinput and corn cultivar.

2. Material and Methods

The experiment was conducted on an area of 2,760 m² in Chã de Jardim, in the municipality of Areia-PB, belonging to the Center for Agricultural Sciences of the Federal University of Paraíba. The area is located at the geographical coordinates S 6° 58’ 11.6” and W 35° 43’ 56.8”. The experiment was conducted from April to August 2024 and from April to August 2025.

2.1. Implementation, design, and application of bioinput

Three varieties of corn were used in the planting: an early commercial transgenic hybrid variety, B2433PWU from Brevant® seeds, which has high yield, early cycle, summer and off-season, and is intended for grain and silage; a native variety called Jabatão, which has a long cycle, ranging from 100 to 120 days, is tall, has very uneven and large grains, is very susceptible to attack by pests, such as fall armyworm, corn earworms and corn borers, but is well accepted by family farmers in the state of Paraíba; and the last, the commercial hybrid Robusto® from the company Selegrãos, which also has high productivity and a long cycle, around 120 days, and, like the others, has a high adaptive capacity to different types of soil and climate. The area was prepared mechanically and manually. Liming and fertilization were based on soil analysis of the area. Sowing occurred in April, as the activity was rain-fed and climatic conditions were favorable in the region.

In this study, one corn seed was sown per hole, at a depth of 2-3 cm, with a spacing of 0.3 meters between plants and 0.8 meters between rows. The B2433PWU variety was sown in three alternating strips, each 20 meters wide by 15 meters long. The Jabatão and Robusto varieties were arranged in four alternating strips, each 4 meters wide by 15 meters long. We chose this planting arrangement because we simulated the refuge in strips.

We used a strip-plot experimental design with four replicates per variety. For each experimental plot, 10 corn plants were marked, corresponding to an area of 1.6 m by 3.3 m. Among these plants, five were previously identified and monitored from the beginning to the end of the crop cycle, being used to obtain the biometric and productive variables evaluated in the experiment. After sowing and initial establishment of the plants, evaluations began with the first detection of S. frugiperda larvae, recorded at phenological stage V5. Monitoring was performed weekly until the transition from the final vegetative to the initial reproductive stages (FV/R1) to quantify the extent of damage. For this assessment, the rating scale proposed by Davis et al. (1992) was used, which allows the degree of leaf injury to be classified in a standardized, comparable manner across treatments.

The morphological variables of corn evaluated in the experiment included characteristics essential to vegetative and reproductive development, enabling a comprehensive analysis of the effects of treatments and varieties. The following were measured: (i) ear length (cm), defined as the distance between the basal and apical ends of the unshucked ear; (ii) ear diameter (mm), obtained in the central region using a digital caliper; (iii) number of kernels per row; (iv) total number of kernels per ear; and (v) kernel weight per ear (g). All variables were measured after harvesting and subsequent manual threshing of the ears.

The experiment was conducted with two treatments: Treatment 1 (the control; water application) and Treatment 2 (the product Fitoneem®). The application was carried out in continuous strips, arranged perpendicular to the planting line, called blocks, which were separated by different corn varieties. Thus, all plots received the treatment uniformly.

The bioproducts were applied using a DJI T40 spray drone at an average speed of 15 km/h, a flight height of 4.5 meters, a wind speed of 6-10 km/h, and a flow rate of 20 l/ha. In Treatment 1 (T1), only water was applied as a control. In Treatment 2 (T2), 500 ml/ha of Fitoneem® was used, with 150 ml/ha of Santara® adjuvant, for a total spray volume of 20 liters. The first application occurred 30 days after planting (DAP), and the same procedure was repeated at 60 DAP for all treatments. Although the applications were submitted at different times, the data were combined and jointly analyzed.

The insect monitored in this study was the fall armyworm. Throughout the incidence period, weekly surveys were conducted to assess the insect's population density at the field level. The assessments were performed manually to determine the presence of the caterpillar, plant development, and damage caused by S. frugiperda to corn plants. In addition, flights were conducted over the experimental area using multispectral drones to identify potential damage characteristics resulting from the pest attack and to evaluate the effectiveness of the applied product.

2.2. Flights with RPA

The flights were carried out using DJI's Mavic 3 multispectral remotely piloted aircraft. The RPA comprises five cameras, one of which is an RGB camera with a 20 MP 4/3 CMOS sensor. The others are multispectral cameras with four 1/2.8-inch single-band CMOS sensors, capable of taking 5 MP photos with a focal length equivalent to 25 mm and an aperture of f/2.0, obtaining images in the following bands: Green (G): 560 ± 16 nm, Red (R): 650 ± 16 nm, Red Edge (RE): 730 ± 16 nm, Near Infrared (NIR): 860 ± 26 nm (DJI, 2023).

This equipment also includes a spectral sunlight sensor mounted on its upper surface, which detects solar radiation in real time. When combined with image data from each band of the multispectral camera, we obtain more accurate reflectance values, thereby improving the consistency of data collected across different regions, climatic conditions, and times. Flights were conducted weekly, and flight parameters, including height and front and side overlaps, were recorded.

2.3. Image processing

The multispectral images were processed using Pix4D software. The multispectral images were loaded into the software. Then processing began, which basically consists of three steps: Initial processing, where the program identifies specific features in the images, such as key points, and which images have the same key points to combine them, as well as calibrating internal parameters, such as focal length, and external parameters, such as camera orientation. In the second step, the point cloud and mesh were generated. Finally, in the third and last stage, the digital elevation model (DEM), orthomosaic, and indices were obtained.

From image processing, the following indices were obtained: the Normalized Difference Vegetation Index (NDVI) and the Normalized Difference Vegetation Index in the Red Edge (NDRE). In general, these indices enable the observation and evaluation of plant photosynthetic activity through arithmetic calculations that use the near-infrared (NIR) and red (Red) bands for NDVI and the red edge (RedEdge) and red (Red) bands for NDRE. For both indices, this calculation was performed per pixel, yielding values ranging from -1 to 1. With these two indices, it will also be possible to compare their effectiveness in identifying insect pests in corn crops.

2.4. Statistical analysis

As there was no difference between the seasons in the Analysis of Variance using the Aligned Rank Transform (P > 0.12), we pooled the data from both seasons for a single analysis. Given that the pest did not occur throughout the cycle, we used the most pest-expressive stage (FV). We performed a correlation analysis between the vegetation index and S. frugiperda damage using this single evaluation.

Multiple comparisons among treatments were performed using contrasts from the nonparametric ART factorial model, with means followed by the same letter considered statistically similar. To investigate the multivariate structure of the data, a non-hierarchical cluster analysis was performed using all phenotypic variables evaluated in Bt (PWU) and non-Bt (Robusto and Jabatão) corn varieties subjected to Fitoneem® (Fito) and control (Control) treatments. All statistical analyses and graph preparation were performed using R.

3. Results

The field results revealed that the population variation of S. frugiperda was influenced by the corn variety and by the aerial application of Fitoneem®. For example, the variety that showed the least increase in Damage Variation (DV) was the Jabatão variety [DV: 0.05 (95% CI = -0.0023-0.1023). However, the Robusto variety with Fitoneem® application did not have a significant impact on the population reduction of S. frugiperda [DV: 0.6500 (95% CI = 0.4161-0.8839] and did not differ significantly from that of the control treatment [DV: 0.4500 (95% CI = 0.1654-0.7346]. The variation in damage to the Bt variety was very low, not exceeding 0.04, or 4% (Table 1).

Table 1
Matrix of means and confidence intervals for absolute Damage Variation (DV) and NDVI and NDRE indices.

	Damage Variation (DV)		NDVI		NDRE
	Control	Fitoneem®	Control	Fitoneem®	Control	Fitoneem®
Jabatão	Mean = 0.40	Mean = 0.05	Mean = 0.83	Mean = 0.80	Mean = 0.46	Mean = 0.46
	LCI = 0.30	LCI = -0.002	LCI = 0.82	LCI = 0.77	LCI = 0.44	LCI = 0.44
	UCI = 0.76	UCI = 0.10	UCI = 0.84	UCI = 0.83	UCI = 0.48	UCI = 0.47
Robusto	Mean = 0.45	Mean = 0.65	Mean = 0.81	Mean = 0.83	Mean = 0.45	Mean = 0.47
	LCI = 0.16	LCI = 0.41	LCI = 0.80	LCI = 0.82	LCI = 0.43	LCI = 0.45
	UCI = 0.73	UCI = 0.88	UCI = 0.83	UCI = 0.84	UCI = 0.47	UCI = 0.49
PWU	Mean = 0.01	Mean = 0.01	Mean = 0.85	Mean = 0.85	Mean = 0.46	Mean = 0.49
	ICI = 0.00	LCI = 0.00	LCI = 0.84	LCI = 0.84	LCI = 0.44	LCI = 0.48
	UCI = 0.04	UCI = 0.02	UCI = 0.86	UCI = 0.86	UCI = 0.48	UCI = 0.50

• LCI = Lower Confidence Interval; UCI = Upper Confidence Interval.

Our results showed that it is feasible to monitor S. frugiperda using remote NDVI detection. Using the original data from Table 1, we could find that the relationship between NDVI and S. frugiperda damage (r = -0.630; P = 0.0050; sample size= 40) was significant and stronger than that between NDRE and damage (r = -0.4620; P = 0.053; sample size= 40). Therefore, higher S. frugiperda damage will result in lower NDVI. The red spots in the figures 1 -4 indicate the highest level of infestation of S. frugiperda, the images approaching the NDVI (Figures 1-2) were more representative than the NDRE (Figures 3-4).

Figure 1
Representation of Spodoptera frugiperda infestation hotspots, with Bt and non-Bt corn crops – areas with yellow and red dots and Normalized Difference Vegetation Index (NDVI) (image captured on June 2, 2024 season). The red spots also indicate the highest level of Spodoptera frugiperda infestation.

Figure 2
Representation of Spodoptera frugiperda infestation hotspots, with Bt and non-Bt corn crops – areas with yellow and red dots and Normalized Difference Vegetation Index (NDVI) (image captured on April 25, 2025 season). The red spots also indicate the highest level of S. frugiperda infestation.

Figure 3
Representation of the evaluation blocks used in the study with Spodoptera frugiperda infestation hotspots, with Bt and non-Bt corn crops – areas with yellow and red dots and Normalized Difference Red Edge (NDRE) (image captured on June 2, 2024 season). The red spots also indicate the highest level of S. frugiperda infestation.

Figure 4
Representation of Spodoptera frugiperda infestation hotspots, with Bt and non-Bt corn crops – areas with yellow and red dots and Normalized Difference Red Edge (NDRE) (image captured on April 25, 2025 season). The red spots also indicate the highest level of S. frugiperda infestation.

The results of the analysis of variance are presented in Table 2. For the variables Ear Length (EL) and Ear Diameter (ED), there was a difference between the Varieties (F_{EL variety}=14.3411, P_{EL variety}< 0.00001; F_{ED variety}=14.9927, P_{ED variety}< 0.00001), with no effect of Treatment (F_{EL treatment}= 0.2978, P_{EL treatment}= 0.5862; F_{EL treatment}= 0.2710, P_{EL treatment}= 0.8858), and of the interaction Treatment versus Variety (F_{EL treatment vs variety}=0.0601; P_{EL treatment vs variety}= 0.9416; F_{ED treatment vs variety}= 14.9927, P_{ED treatment vs variety}= 0.8858).

Table 2
Agronomic variables (Mean±Standard Error) related to the production of Bt and non-Bt corn under the influence of bioinput application via drone sprayer.

	Ear Length (EL) (cm)
	Jabatão	PWU	Robusto	Median
Control	14.28±0.50	12.00±0.62	11.92±0.41	12.73±0.32 A
Fitoneem®	14.46±0.40	12.52±0.53	11.94±0.39	12.97±0.29 A
Median	14.37±0.32 A	12.26±0.41 B	11.93±0.27 B
	Ear Diameter (ED) (mm)
	Jabatão	PWU	Robusto	Median
Control	43.80±0.79	49.60±1.05	46.96±1.01	46.79±0.62 A
Fitoneem®	45.31±0.82	49.45±0.86	45.42±0.87	46.73±0.54 A
Median	44.56±0.57 B	49.53±0.67 A	46.19±0.67AB
	Number of Grains per Row (NGPF)
	Jabatão	PWU	Robusto
Control	29.02±1.29	28.56±1.52	26.31±1.33	27.96±0.80 B
Fitoneem®	31.80±1.15	29.25±1.09	31.30±1.28	30.78±0.68 A
Median	30.41±0.88 A	28.90±0.92 A	28.80±0.99 A
	Number of Grains per Ear (NGPE)
	Jabatão	PWU	Robusto	Median
Control	313.24±16.30	498.51±28.40	394.67±27.66	402.14±34.35B
Fitoneem®	349.05±18.25	516.90±25.95	456.20±20.73	440.71±15.35ª
Median	331.14±12.41B	507.70±19.04ª	425.43±17.76B
	Ear Weight (EW) (g)
	Jabatão	PWU	Robusto	Median
Control	144.75±11.14	143.57±13.74	132.55±10.46	140.29±13.55
Fitoneem®	141.55±7.51	135.60±8.92	118.70±7.45	131.95±4.50
Median	143.15±13.42	139.58±8.11	125.62±6.43

• EL: Ear Length. ED: Ear Diameter. NGPF: Number of Grains per Row. NGPE: Number of Grains per Ear. EW: Ear Weight. Means followed by the same letters do not differ from each other according to the contrasts generated by the Nonparametric Analysis of Variance in the Aligned Rank Transform (ART) Factorial Model. Sample size = 10.

There was a significative effect of the application of Fitoneem® on the variables Number of Grains per Row (NGPR) and non-significative effect on Number of Grains per Ear (NGPE) (F_{NGPR treatment}= 5.0260, P_{NGPR treatment}= 0.0269; F_{NGPE treatment}= 1.7390, P_{NGPE treatment}= 0.1006), on the other hand, there was no effect of Variety (F_{NGPR variety}= 0.8358, P_{NGPR variety}= 0.4361; F_{NGPE variety}= 0.3025, P_{NGPE variety}= 0.7395) or interaction (F_{NGPR treatment vs variety}= 1.3528, P_{NGPR treatment vs variety}= 0.2626; F_{NGPE treatment vs variety}= 29.6917, P_{NGPE treatment vs variety}< 0.0001). For the Ear Weight (EW) variable, there was no effect of Treatment (F_EW= 0.3381, P_EW= 0.5620), Variety (F_EW= 1.6597, P_EW= 0.1947), or interaction (F_EW= 0.1321, P_EW= 0.8763).

Based on the multiple variables addressed in the study, the non-hierarchical cluster (Figure 5) showed greater distance between the non-Bt varieties, Robusto and Jabatão, after the Fitoneem® application than in the control. This reveals the contribution of this biological product as a potential bio-reducing agent of the intensity of damage by S. frugiperda, as well as a biostimulator of the morphophysiology of the non-Bt corn varieties addressed in this study.

Figure 5
Non-hierarchical cluster of multiple variables collected in Bt (PWU) and non-Bt corn varieties, Robusto (ROB) and Jabatão (JAB) under the influence of Fitoneem^® (Fito) and Control (Control).

4. Discussion

The aerial application of Fitoneem® in the corn production field significantly affected the population of S. frugiperda, depending on the corn variety used, resulting in less damage to the Jabatão variety [DV: 0.05 (95% CI = -0.0023-0.1023), and also interfering with the number of grains per ear and number of grains per row, consequently affecting the vegetative indices of these plants and providing a greater multivariate Euclidean distance in relation to the control. This may have occurred due to the product's efficiency or to the specific characteristics of each corn variety, such as genetic resistance and physiological traits that affect interactions with S. frugiperda, even when bio-inputs are used.

The greater severity of damage caused by S. frugiperda in the Robusto variety is related to structural factors of the plant, since this variety has a smaller architecture and more limited growth, both in terms of height and leaf length, when compared to Jabatão, as these factors may favor the exposure of the fall armyworm to Fitoneem® when sprayed on the Jabatão variety, which may increase the residual effect of this product on the leaves. The variety of corn chosen affects pest control efficiency. The Robusto variety is a hybrid, and in this study, it showed less resistance to insect herbivory than Jabatão, a regional variety from Paraíba. Another point to highlight is the strategic combinations of mechanical defense, such as increased rigidity due to lignification and trichome thickness, which influence the feeding capacity of chewing insects (Chávez-Arias et al., 2022). They may also be related to chemical strategies, such as a higher abundance of terpenes, a group of secondary metabolites that function as defense toxins and deter herbivory (Jan et al., 2021). Therefore, if barriers prevent the fall webworm from feeding effectively, feeding will be reduced, thereby increasing the pest's susceptibility to the bioproduct used, as is the case for caterpillars in the Jabatão variety.

In the control treatment, the Robusto and Jabatão varieties were observed, and there was a significant increase in S. frugiperda population size without intervention, thereby reinforcing its high polyphagous and adaptive capacity (Murad et al., 2021). After aerial application of Fitoneem® via drone, a significant decrease in S. frugiperda damage was observed, reducing damage by up to 95% in the Jabatão corn variety, indicating that the caterpillar population had a greater impact on the Jabatão variety than on the Robusto variety.

The results demonstrate the feasibility of remotely monitoring S. frugiperda using multispectral drones, with the NDVI index as a reference. This is due to a stronger relationship between NDVI and S. frugiperda damage than between NDRE and S. frugiperda damage. The highest NDVI values were observed in treatments with Fitoneem®, highlighting its effectiveness in reducing damage caused by S. frugiperda. However, these results can vary across corn varieties. Silva et al. (2025b) found that the NDRE, in particular, showed greater discriminatory capacity at more advanced stages of the crop, as revealed by machine learning metrics such as Accuracy and Sensitivity greater than 80.00%. They used the Bt corn varieties NK 501 VIP3 and NK 509 VIP3.

This result reinforces the importance of integrated pest management practices using remote sensing to assess the crop's phytosanitary condition and the efficiency of application. These morphological and physiological effects on plants can cause considerable changes in spectral properties, especially in the red and yellow bands of the electromagnetic spectrum, which are used to calculate NDVI. These bands, which received these applications, can simulate variations in plant chlorophyll and moisture levels caused by pest damage (Yamanura and Patil, 2021).

The use of remote sensing to monitor S. frugiperda has proven to be a promising approach and, in addition to its growing application in various regions of the world, it has been encouraged, offering an effective alternative for large-scale monitoring, especially in extensive agricultural areas, enabling early detection and more efficient management of this pest (Wang et al., 2023), which reinforces the findings in this study, which identified higher NDVI values associated with the levels of damage caused by S. frugiperda. The convergence of results from current research strengthens the evidence base for the use of multispectral drones in this context. It is important to note that NDVI can be influenced by factors beyond pest damage, including the presence of phytopathogens, invasive plants, variations in water availability (whether due to deficit or excess), and nutritional deficiencies (Pei et al., 2019; Achille et al., 2021).

The results presented demonstrate experimental consistency and practical applicability, although some limitations should be considered. Environmental variability across experimental periods, including temperature, humidity, solar radiation, and field conditions, may have partially influenced the plants' spectral response. In addition, genetic, morphological, and physiological differences among the evaluated corn varieties may have contributed to distinct responses to S. frugiperda damage and to the application of bioinputs. The evaluations were carried out within specific time windows of the crop cycle, limiting inferences about spectral dynamics throughout the entire corn cycle. It should be noted, however, that multispectral image acquisition was standardized and always carried out between 12:00 and 14:00 to minimize interferences caused by clouds and variations in solar radiation incidence during flights.

Another important point is that the study demonstrated consistent associations between spectral indices and pest-induced damage through correlation analyses, highlighting the potential of NDVI and NDRE as phytosanitary monitoring tools under field conditions. However, because these are correlational analyses, it is not possible to establish direct causal relationships between the variables. Therefore, future studies incorporating more detailed physiological, biochemical, structural, and temporal analyses could broaden the understanding of the mechanisms underlying maize's spectral response to S. frugiperda infestation.

5. Final Considerations

We concluded that NDVI shows a stronger relationship with S. frugiperda damage than NDRE and was therefore the most appropriate index for assessing the severity of field infestations by this pest. In addition, it was observed that infestation intensity is affected by both the corn cultivar and the bioproduct used. We could show that the management with the Jabatão variety, combined with applications of Azadirachtin, results in less damage, reduces S. frugiperda attacks by up to 95%, and increases the number of grains per row and per corn ear. This study highlights the potential of neem seed oil, which contains the active ingredients Azadirachtin A and B, as a tool for integrated pest management and the preservation of agroecosystems, thereby contributing significantly to sustainability.

Acknowledgements

This research was funded by the Brazilian National Council for Scientific and Technological Development (CNPq) (Process numbers: 420064/2023-0 and 308296/2025-7).

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding author.

References

• ABBAS, A., ULLAH, F., HAFEEZ, M., HAN, X., DARA, M.Z.N., GUL, H. and ZHAO, C.R., 2022. Biological control of fall armyworm, Spodoptera frugiperda. Agronomy, vol. 12, no. 11, pp. 2704. https://doi.org/10.3390/agronomy12112704
  » https://doi.org/10.3390/agronomy12112704
• ACHILLE, L.S., ZHANG, K. and ANOMA, C.J.K., 2021. Analysis of climate variability and relationship with vegetation in Garamba National Park from 1990–2020. Open Journal of Ecology, vol. 11, no. 10, pp. 700-723. https://doi.org/10.4236/oje.2021.1110044
  » https://doi.org/10.4236/oje.2021.1110044
• BANERJEE, R., HASAN, M. and JURAT-FUENTES, J.L., 2022. Mechanisms and spread of resistance to Bacillus thuringiensis toxins in Spodoptera frugiperda: insights from genomic and population studies. Pest Management Science, vol. 78, no. 6, pp. 2282-2294. https://doi.org/10.1002/ps.6823
  » https://doi.org/10.1002/ps.6823
• BHATTARAI, G.P., SCHMID, R.B. and MCCORNACK, B.P., 2019. Remote sensing data to detect hessian fly infestation in commercial wheat fields. Scientific Reports, vol. 9, no. 1, pp. 6109. https://doi.org/10.1038/s41598-019-42620-0 PMid:30992554.
  » https://doi.org/10.1038/s41598-019-42620-0
• BORIS, B. and HIDEO, H., 2019. Comparison of NDVI and NDRE indices to detect differences in vegetation and chlorophyll content. J. Mech. Cont. & Math. Sci, vol. spl1, no. 4, pp. 20-29. https://doi.org/10.26782/jmcms.spl.4/2019.11.00003
  » https://doi.org/10.26782/jmcms.spl.4/2019.11.00003
• COMPANHIA NACIONAL DE ABASTECIMENTO – CONAB, 2024 [viewed 10 December 2024]. Brazilian grain harvest [online]. Available from: https://www.conab.gov.br/info-agro/safras/graos
  » https://www.conab.gov.br/info-agro/safras/graos
• CHÁVEZ-ARIAS, C.C., RAMÍREZ-GODOY, A. and RESTREPO-DÍAZ, H., 2022. Influence of drought, high temperatures, and/or defense against arthropod herbivory on the production of secondary metabolites in maize plants: a review. Current Plant Biology, vol. 32, pp. 100268. https://doi.org/10.1016/j.cpb.2022.100268
  » https://doi.org/10.1016/j.cpb.2022.100268
• DAVIS, F.M., NG, S. and WILLIAMS, W.P., 1992. Visual rating scales for screening whorl-stage corn resistance to fall armyworm Mississippi: Mississippi State University, 9 p. Technical Bulletin, no. 186.
• DJI, 2023 [viewed 4 April 2025]. DJI Mavic 3 Enterprise Series – User Manual [online]. Available from: https://www.dji.com/downloads
  » https://www.dji.com/downloads
• JAN, R., ASAF, S., NUMAN, M., LUBNA. and KIM, K.-M., 2021. Plant secondary metabolite biosynthesis and transcriptional regulation in response to biotic and abiotic stress conditions. Agronomy, vol. 11, no. 5, pp. 968. https://doi.org/10.3390/agronomy11050968
  » https://doi.org/10.3390/agronomy11050968
• MAXIMIANO, C.V., 2017. Preconditioning of corn seeds in water with different concentrations of ozone in the initial development of seedlings and in the control of Fusarium spp Brasília: University of Brasília, 55 p. Dissertation Master’s in Agronomy.
• MURAD, N.F., SILVA-BRANDAO, K.L. and BRANDAO, M.M., 2021. Mechanisms behind polyphagia in a pest insect: responses of Spodoptera frugiperda (JE Smith) strains to preferential and alternative larval host plants assessed with gene regulatory networks. Biochimica et Biophysica Acta. Gene Regulatory Mechanisms, vol. 1864, no. 3, pp. 194687. https://doi.org/10.1016/j.bbagrm.2021.194687 PMid:33561559.
  » https://doi.org/10.1016/j.bbagrm.2021.194687
• NASCIMENTO, B.M.C., NUNES, A.D.M., SOUZA JUNIOR, S.L.D., SANTOS, L.F.D.S., MIELEZRSKI, F., BRITO, C.H.D., CÂNDIDO, B.A.P., MEDEIROS, I.L., PRAXEDES JÚNIOR, W.G., REZENDE, J.M., RAMALHO, F.S., SILVA, R.M.M.F. and MALAQUIAS, J.B., 2025. Beyond the mortality effect: spodoptera frugiperda multiple nucleopolyhedrovirus promotes changes in feeding and inhibits larval growth and weight gain in fall armyworm. Microorganisms, vol. 14, no. 1, pp. 1. https://doi.org/10.3390/microorganisms14010001 PMid:41597521.
  » https://doi.org/10.3390/microorganisms14010001
• PEI, Z., FANG, S., YANG, W., WANG, L., WU, M., ZHANG, Q., HAN, W. and KHOI, D.N., 2019. The relationship between NDVI and climate factors at different monthly timescales: a case study of grassland in Inner Mongolia, China (1982-2015). Sustainability, vol. 11, no. 24, pp. 7243. https://doi.org/10.3390/su11247243
  » https://doi.org/10.3390/su11247243
• RADOGLOU-GRAMMATIKIS, P., SARIGIANNIDIS, P., LAGKAS, T. and MOSCHOLIOS, I., 2020. A compilation of UAV applications for precision agriculture. Computer Networks, vol. 172, pp. 107148. https://doi.org/10.1016/j.comnet.2020.107148
  » https://doi.org/10.1016/j.comnet.2020.107148
• SILVA, A.S., OLIVEIRA, R., NETO, J.S.P., BORGES, V.P., GUERRERO, M.F.C., FILHO, A.A.C.D.A., SOUSA, R.G.C., NASCIMENTO, A.R.B. and MALAQUIAS, J.B., 2025a. High-resolution suborbital remote sensing for detecting injuries caused by Saccharicoccus sacchari (Cockerell, 1895) (Hemiptera: Pseudococcidae) in sugarcane. Frontiers in Agronomy, vol. 7, pp. 1630354. https://doi.org/10.3389/fagro.2025.1630354
  » https://doi.org/10.3389/fagro.2025.1630354
• SILVA, A.S., SANTOS, L.F.S., SILVA, M.T., COSTA, M.M.L., SALUSTINO, A.S., ROLIM, R.R., SILVA, L.B., GONZAGA, K.S., DOURADO, W.S., GONÇALVES, N., MIELEZRSKI, F., NASCIMENTO, N.F.F. and MALAQUIAS, J.B., 2025b. Multispectral imaging to validate the effectiveness of the combination of biological and chemical management of Spodoptera frugiperda in a region with high selection pressure. Brazilian Journal of Biology, vol. 85, pp. e297263. https://doi.org/10.1590/1519-6984.297263 PMid:41779531.
  » https://doi.org/10.1590/1519-6984.297263
• SISAY, B., TEFERA, T., WAKGARI, M., AYALEW, G. and MENDESIL, E., 2019. The efficacy of selected synthetic insecticides and botanicals against fall armyworm, Spodoptera frugiperda, in maize. Insects, vol. 10, no. 2, pp. 45. https://doi.org/10.3390/insects10020045 PMid:30717302.
  » https://doi.org/10.3390/insects10020045
• SOLOGUREN, L., 2015. Global demand grows and Brazil has room to expand production. Visão Agrícola, Piracicaba, no. 13, pp. 8-11.
• TANDY, P., EMRICH, S.J. and JURAT-FUENTES, J.L., 2023. Screening for resistance alleles to Cry1 proteins through targeted sequencing in native and invasive populations of Spodoptera frugiperda. Journal of Economic Entomology, vol. 116, no. 3, pp. 935-944. https://doi.org/10.1093/jee/toad061 PMid:37311017.
  » https://doi.org/10.1093/jee/toad061
• WANG, J., HUANG, Y., HUANG, L., DONG, Y., HUANG, W., MA, H., ZHANG, H., ZHANG, X., CHEN, X. and XU, Y., 2023. Migration risk of fall armyworm (Spodoptera frugiperda) from North Africa to Southern Europe. Frontiers in Plant Science, vol. 14, pp. 1141470. https://doi.org/10.3389/fpls.2023.1141470 PMid:37077648.
  » https://doi.org/10.3389/fpls.2023.1141470
• WOLFF, W., FRANCISCO, J.P., FLUMIGNAN, D.L., MARIN, F.R. and FOLEGATTI, M.V., 2022. Optimized algorithm for evapotranspiration retrieval via remote sensing. Agricultural Water Management, vol. 262, pp. 107390. https://doi.org/10.1016/j.agwat.2021.107390
  » https://doi.org/10.1016/j.agwat.2021.107390
• YAMANURA, R.M. and PATIL, B., 2021. NDVI-derived LAI model: a new tool for crop monitoring. Journal of Crop and Weed, vol. 17, no. 1, pp. 76-85. https://doi.org/10.22271/09746315.2021.v17.i1.1408
  » https://doi.org/10.22271/09746315.2021.v17.i1.1408

Edited by

• Editor:
  Takako Matsumura Tundisi

Publication Dates

• Publication in this collection
  03 Aug 2026
• Date of issue
  2026

History

• Received
  09 Jan 2026
• Accepted
  17 June 2026