Evaluating the Effect of Pulse Width Modulation-Controlled Spray Duty Cycles on Cotton Fiber Quality Using Principal Component Analysis

Abstract

The optimization of cotton defoliant application is critical for enhancing fiber quality and yield. This study aims to assess the impact of different defoliant duty cycles on cotton fiber quality by applying Principal Component Analysis (PCA) to High-Volume Instrument (HVI) data from two fields. Three duty cycles—20%, 40%, and 60%—along with a control treatment were evaluated. PCA was used to identify the key factors influencing cotton quality, with a focus on parameters such as fiber length, strength, and uniformity. The results revealed that the 40% duty cycle consistently produced the most stable and uniform cotton fiber quality across both fields, minimizing variability in critical parameters. In contrast, the 20% and 60% duty cycles, as well as the control, introduced greater variability, with the control treatment showing the most significant outliers. These findings suggest that a 40% duty cycle is optimal for balancing effective defoliation with high-quality cotton production. Future research should explore the robustness of the 40% duty cycle across different environmental conditions and investigate the integration of advanced technologies to further optimize defoliant applications. This study provides valuable insights for improving cotton production practices and ensuring consistent fiber quality.

1. Introduction

Cotton production is important in the United States, making substantial contributions to the economy and agricultural sector. Texas, the leading cotton-producing state in the U.S., plays a crucial role in the nation’s cotton production landscape. According to the USDA-ERS [1], Texas accounted for approximately 40% of U.S. cotton production, with a total production of 4.17 million bales of cotton in 2013 [2]. This statistic highlights the substantial contribution of Texas to the overall cotton output of the U.S. The economic significance of cotton extends beyond these production figures, as emphasized in a study focusing on the impact of insect-resistant cotton. Cotton is cultivated on 7% of the U.S. cropland, with the cotton textile industry contributing 4% to the GDP and 14% to the total industrial product [3]. Additionally, the cotton industry provides significant employment to the U.S. workforce, with 20% engaged in the sector. Cotton exports also play a crucial role, accounting for 17% of the total export earnings [3]. These figures underscore the multifaceted economic importance of cotton production in the U.S., spanning agricultural activities, industrial output, and international trade.

Genetic variation among cotton cultivars and their response to supplemental irrigation is crucial for optimizing cotton production, especially in irrigated systems. Understanding how different germplasms react to water availability is essential to maximizing yield and fiber quality [4]. This highlights the interplay between genetic diversity, irrigation, and production outcomes, emphasizing the need for continued research and innovation. Sustainable agriculture initiatives like Better Cotton promote environmentally responsible practices, encouraging farmers to adopt improved agricultural principles to support sustainability and community well-being [5]. Moreover, the quality of cotton seeds, a key byproduct of lint production, is economically important as they are used for oil livestock feed, impacting the cotton production value chain [6]. Efficient nitrogen fertilization also plays a critical role in optimizing yield and fiber quality, emphasizing the need for effective nutrient management [7]. Overall, U.S. cotton production is a major economic driver, contributing to employment, industrial output, and trade, with genetic diversity, sustainable practices, and seed and nutrient management being key to its long-term sustainability.

Automation and robotics have been increasingly integrated into cotton production in the United States to address various challenges and enhance agricultural efficiency [8]. Technological advancements in agronomy, such as new cotton varieties, improved irrigation, and precision agriculture, have introduced autonomous multi-purpose robotic platforms [9]. These platforms streamline operations and optimize resource use. One key application is autonomous weed control, which reduces labor, minimizes herbicide use, and improves productivity, contributing to sustainable farming by reducing chemical inputs. Robotics has also advanced cotton harvesting through mobile platforms equipped with computer vision and AI, which autonomously identify and harvest cotton bolls, addressing labor shortages and reducing costs [10]. In pest and disease management, robotics improves crop monitoring by integrating drones and NDVI for precise detection and treatment strategies, enhancing resilience [11]. Robotics further assists in tasks like row detection, boll location estimation, and autonomous navigation, with technologies such as stereo-vision and lidar improving harvesting and field operations [12,13]. Integrating robotics in U.S. cotton farming represents a transformative shift toward greater efficiency, sustainability, and technological advancement.

Cotton defoliation is a vital process in preparing cotton plants for mechanical harvesting. This practice involves removing leaves from the plants, which is crucial for improving harvest efficiency and fiber quality [14]. Effective defoliation allows harvesting machines to access cotton bolls more easily and enhances the quality of the harvested fiber. Chemical defoliants are widely used in cotton production to accelerate leaf abscission and boll maturation, making mechanical harvesting more efficient [15]. Successful defoliation minimizes the impurity content in cotton, prevents plant lodging, and shortens the time from defoliation to harvest, all of which contribute to a more productive harvesting process [16,17]. Additionally, proper defoliation reduces lint staining and debris during harvesting, further improving fiber quality [18,19]. By employing chemical defoliants and other defoliation techniques, farmers can optimize the harvesting process, enhance fiber quality, and boost the overall productivity of cotton cultivation [20,21].

While PCA has been widely used in various studies focusing on cotton, including analyses of genetic variation, yield traits, and stress responses, no previous studies have applied PCA in conjunction with PWM to control defoliant application in cotton production. The lack of direct references is due to the novelty of PWM-controlled defoliant application, with this research being one of the first to explore this method. In response to the gap in the existing literature, PWM was selected for its precision in controlling defoliant delivery. Additionally, related studies utilizing other methods of defoliant application and fiber quality analysis were reviewed to provide a broader context for the novel approach used in this study. Similar studies have employed PCA to assess genetic variation in physiological parameters of cotton [22], investigate genetic divergence in yield-related traits in cotton germplasm [23], identify domestic and foreign cotton fibers using near-infrared spectroscopy [24], enhance the similarity or dissimilarity of cotton fibers harvested at different locations [25], estimate genetic diversity in cotton genotypes for earliness, yield, and fiber quality traits [22], classify cotton fiber maturity phenotypes [26], differentiate different types of raw cotton [27], classify cotton cultivars for heat tolerance [28], assess heterosis and genetic components in cotton yield and fiber traits [29], correlate cotton biomass accumulation, growth traits, and yield [30], evaluate salt stress response in upland cotton [31], evaluate drought resistance in cotton [32], assess heat tolerance in upland cotton [33], and evaluate drought stress responses in cotton genotypes [34]. These studies collectively demonstrate the versatility of PCA in analyzing various aspects of cotton, ranging from genetic diversity and fiber quality to stress responses. PCA has proven valuable in unraveling complex relationships and patterns within cotton datasets, providing researchers with insights into key traits and characteristics essential for cotton improvement and production. The present study builds on these methodologies by integrating PWM control for defoliant applications, offering a more precise approach to managing defoliant delivery and assessing its impact on fiber quality.

The overall objective was to optimize cotton defoliant application to enhance fiber quality and yield. Specifically, the study aimed to determine the most effective duty cycle for defoliant application that consistently produces high-quality cotton fiber while minimizing variability across key parameters such as fiber length, strength, and uniformity. Three different duty cycles—20%, 40%, and 60%—and a control treatment were evaluated to achieve this. PCA was applied to High-Volume Instrument (HVI) data from two fields to identify the key factors contributing to cotton quality and to understand how these different duty cycles influence those factors. Although both HVI and Advanced Fiber Information System (AFIS) data were collected, the results presented in this study focus exclusively on the HVI data. The study sought to determine which duty cycle produced the most consistent and high-quality cotton fiber across varying field conditions. Ultimately, the research aimed to provide practical recommendations for cotton producers on the most effective defoliant application strategies, contributing to improved cotton production practices that optimize both fiber quality and yield.

2. Materials and Methods

2.1. Location

The research was conducted at the Edisto Research and Education Center (EREC) in Blackville, SC, USA, across two selected locations on the research farm (Field 1: 33.34736° N, 81.31925° W; Field 2: 33.35398° N, 81.31024° W). Both locations followed a completely randomized experimental design, with 4 replications in Field 1 and 2 replications in Field 2 (Figure 1). In Field 1, cotton was planted in 6 rows, organized into two main groups, G1 and G2. Each group was subdivided into 3 smaller rows, each measuring 9.8 m in length (Figure 2). This subdivision facilitated a more detailed and controlled study of the experimental treatments. The layout ensured that each subgroup could be independently analyzed, allowing for a more precise comparison of the effects across different field sections. The arrangement also helped minimize variability and enhance repeatability.

In Field 2, cotton was planted in late May of 2022. The plants in this field are smaller than those in Field 1 (Figure 2). Six rows were selected for the study, with lengths varying between 21 and 27 m. A skip-row planting method was employed in both fields to accommodate the movement of the mobile robot. This approach involved leaving every other row unplanted, creating wider gaps between the planted rows, which facilitated the robot’s navigation and ensured consistent application of treatments across the designated areas (Figure 3).

2.2. Equipment

This study employed the Husky A200 mobile platform (Clearpathrobotics, Kitchener, ON, Canada), previously used in other research projects at the Sensor and Automation Laboratory (see Figure 4a). The Husky A200 is a robust yet lightweight platform, capable of pulling or carrying payloads of up to 75 kg, making it well-suited for various research applications. For navigation, the platform is equipped with an advanced suite of sensors and components, including an Inertial Measuring Unit (UM7, CH Robotics, Victoria, Australia), a Global Positioning System (Swiftnav, Swift Navigation, San Francisco, CA, USA), motors, encoders, and a laser scanner (UST-10LX, Hokuyo, Osaka, Japan). These features collectively enable precise and reliable movement and operation in the field, ensuring the accurate execution of tasks, such as cotton defoliant application.

The sprayer unit used in this study (Model #1598042, County Line, Austin, TX, USA) is a 94 L, 2-nozzle trailer sprayer, originally equipped with a built-in 12 V diaphragm pump (Figure 4b) capable of delivering 9.5 L per minute at a rated pressure of 482 KPa. To better suit the research needs, the sprayer was retrofitted with six nozzles (Model #625147-001, Capstan Ag Systems Inc., Topeka, KS, USA) and various components, such as valves, O-rings, fly nuts, and other sprayer parts were supplied by Wilger Inc. (Wilger Industries, Saskatoon, SK, Canada). The configuration included three nozzles on each side, positioned at 38 cm, 84 cm, and 145 cm from the ground, respectively, as illustrated in Figure 4c. These nozzles were designated as low, mid, and top nozzles.

The top nozzle was mounted on an aluminum extrusion and angled at 40 degrees, with the extrusion being adjustable via three hex screws to accommodate variations in crop height during field testing. The positioning of the bottom and middle nozzles was based on the spray pattern spread of the tip used, while the top nozzle’s position was determined by crop height. Each nozzle was equipped with an ER110-06 spray tip (Wilger Industries, Saskatoon, SK, Canada), a conventional flat fan nozzle known for producing a fine droplet spray with a consistent pattern.

2.3. Sprayer Controller

The sprayer controller’s electronics are divided into two components: the main controller (Figure 5a) and the sprayer board (Figure 5b). The main controller is an ARM Cortex-M4 (MK66FX1M0VMD18, NXP, Eindhoven, The Netherlands) with 256 Kb of SRAM, 1280 Kb of Flash RAM, 4 Kb of EEPROM, six UARTs, three SPI interfaces, four I2C interfaces, two USB controllers, and an Ethernet port. It features 100 programmable GPIO pins, twenty-five 16-bit timers, and four 32-bit timers. Designed for versatile lab use, most GPIO and special function pins (serial, CAN, I2C, SPI) are connected to a header on the side. The controller also includes a coin cell battery for the Real-Time Clock, a micro-SD card for storage, and two USB ports for programming and host mode.

The block diagram of the sprayer system is illustrated in Figure 6. The system comprises two stacked electronic boards, with the main controller positioned beneath the sprayer board. Three 3-cell LiPo batteries supply power, each with a capacity of 5 Ah, connected in parallel to provide a total of 15 Ah. This power is routed through a 12 V relay to operate the diaphragm pump and sprayer nozzles. The 12 V supply is then lowered to 5 V via a DC/DC converter to power the transceiver unit. In addition, a low dropout voltage regulator (LM1117, TI, Dallas, TX, USA) reduces the 5 V to the required voltage levels for the microcontroller, GPS module, and microSD card. The GPS module (Copernicus II, Trimble, Westminster, CO, USA) geolocates the sprayer unit, while the microSD card serves as a backup for post-debugging of data transmitted by the transceiver. Additionally, the sprayer board includes a temperature sensor (TMP-36, Analog Devices, Wilmington, MA, USA) that monitors the board’s temperature. However, it is not currently used in the firmware but may be incorporated in future updates.

2.4. Cotton Cultivars and Defoliants

Delta Pine cultivars were selected for the study, with DP 2038B3XF planted in the first field and D10 DP 2055 in the second. These cultivars are known for their high yield potential and strong fiber quality, making them suitable for the study’s objectives. The first field was planted on 12 May 2022, and the second on 25 May 2022, allowing for a comparing growth and defoliation effects under slightly different planting conditions. Both fields were managed according to the best practices outlined in the South Carolina cotton growers’ guide, ensuring consistency in cultivation and care.

A carefully selected mixture of two defoliants and one boll opener was used for defoliation, reflecting the standard practices at the EREC. The chemicals, commonly applied by the EREC farm crew, were chosen for their effectiveness in facilitating the uniform removal of cotton leaves and promoting the opening of cotton bolls, essential for harvesting. The specific chemicals and their compositions are detailed in

Table 1. Chemical formulations of the defoliants and boll opener used, as commonly applied by the EREC farm crew for cotton defoliation.

Product Formulation	Active Ingredient	Concentration	Remarks
Folex 6 EC	Tribufos	454 g/38 L	Cotton defoliant
Free fall SC	Thidiazuron	91 g/38 L	Cotton defoliant
Super boll	Ethephon	907 g/38 L	Plant regulator/Boll opener

, clearly understanding the inputs used in the defoliation process. This combination of chemicals was applied to ensure the defoliation process was efficient and effective, aligning with the study’s goal of optimizing cotton quality and yield.

2.5. Data Analysis

A total of 120 plants were randomly selected for the study, with 80 plants from Field 1 and 40 from Field 2. Each selected plant was tagged with red thread before treatment to ensure consistent data collection across multiple time points. The field test involved three different duty cycles as treatments: 20% (13.0 mL/s), 40% (22.8 mL/s), and 60% (31.9 mL/s). For the control treatment, a conventional tractor-mounted sprayer was used on the opposite side of the research field, applying the same chemical defoliant as in the research plots. Spraying was conducted approximately 20 days before the harvest in both fields.

To evaluate cotton yield and fiber quality, cotton bolls were manually harvested from a 3 m row segment (Figure 2). Around 300 g of seed cotton per sample was selected for ginning using a 10 Saw Eagle Cotton Gin machine. This machine, powered by a 1-1/2 horsepower electric motor, features 25.4 cm diameter saws with top and bottom mounted cast ribs and a rotating doffer brush to remove lint from the saw and deliver it to the lint storage section. Data on lint turnout and seed yield were recorded. Approximately 65 g of cotton fiber per sample was sent to the Cotton Incorporated Laboratory for HVI and AFIS analysis using the USTER AFIS PRO 2. The USTER AFIS PRO 2 is an advanced fiber testing instrument used in the textile industry to analyze key cotton fiber properties, including fineness, maturity, short fiber content, nep content, and trash levels. It provides detailed insights into fiber quality, helping optimize processing and ensure high-quality textile production.

2.6. Principal Component Analysis

Line quality was assessed using PCA with a focus on fiber length across the three duty cycle treatment levels. This statistical technique simplifies the complexity of large datasets by transforming them into a smaller set of principal components (PCs). These PCs capture the most significant sources of variation within the data, enabling clearer interpretation and more effective analysis.

In this study, PCA was crucial for identifying subtle variations in fiber length distributions that could result from the different defoliant application rates used. The fiber length data, along with other quality parameters, were obtained using both the HVI and AFIS. However, this paper focuses on the analysis of HVI data. The decision to concentrate on HVI data was made to streamline the analysis and emphasize the key cotton quality parameters measured by HVI, such as fiber length, strength, and uniformity. These parameters are critical for understanding the impact of different defoliant duty cycles on overall cotton quality. While AFIS provides detailed insights into fiber fineness, maturity, and other specific attributes, the scope of this paper is limited to the broader quality metrics captured by HVI. Future studies may delve into the AFIS data to complement the findings presented here, offering a more comprehensive understanding of how different defoliant treatments affect all aspects of cotton fiber quality.

The PCA applied in this study was mean-centered, with the algorithm based on Singular Value Decomposition (SVD). To ensure the robustness of the model, cross-validation was employed using a full cross-validation method with 28 segments. The analysis generated a total of seven components, but the model suggested five components as the most relevant. The optimal number of components was determined to be five, effectively capturing the key variations in the cotton fiber quality data.

HVI provided key bulk fiber properties such as staple length, length uniformity, micronaire (a measure of fineness and maturity), fiber strength, elongation, color grade, and trash content. By reducing the dimensionality of this comprehensive dataset from HVI measurements, PCA facilitated the detection of patterns and differences that might not be readily observable through traditional analysis methods. This capability is particularly valuable in agricultural research for uncovering hidden trends and relationships within complex datasets, providing insights that inform better decision-making and optimization of practices. In this study, using PCA enabled a more accurate and comprehensive assessment of how the different duty cycle treatments influenced cotton fiber quality, guiding the refinement of defoliation strategies.

3. Results

This study investigated the impact of different defoliant duty cycles on cotton fiber quality, focusing on key parameters measured by the HVI, such as fiber length, strength, and uniformity. By applying PCA to data collected from two distinct fields, we aimed to identify the most effective defoliant application rate that consistently produces high-quality cotton. This analysis centers on HVI results to highlight the broader quality metrics influenced by the various treatments. The following sections present the findings from PCA, illustrating the effects of 20%, 40%, and 60% duty cycles, as well as a control treatment, on cotton quality across both fields.

3.1. PCA Results for HVI Based on Correlation Loading, Score, Influence, and Explained Variance Plots of Field 1

3.1.1. HVI—Correlation Loading Plot

The Correlation Loading Plot (Figure 7) for Field 1 visualizes the relationships among various cotton fiber quality parameters measured by the HVI. Key quality parameters like Upper Half Mean Length (UHML), Strength (Str), and Elongation (Elo) were clustered closely, indicating strong positive correlations, particularly contributing to the variance captured by PC-1 (80%). In contrast, Short Fiber Content and Trash Content were negatively correlated with these quality traits and contributed more significantly to PC-2 (9%).

Key Observations

The following section presents the critical findings from the analysis of defoliant duty cycles in Field 1. Key fiber quality parameters, including fiber length, strength, and uniformity, were assessed to determine the effects of the 20%, 40%, and 60% duty cycles and the control treatment. These observations provide a clear understanding of how each duty cycle impacts cotton fiber quality in this field, with particular emphasis on the performance of the 40% duty cycle.

• Fiber Length and Strength Parameters. The Upper Half Mean Length (UHML), Strength (Str), Elongation (Elo), and Length Uniformity Index (UI) are closely clustered together in the top right quadrant of the plot. Their proximity suggests a strong positive correlation among these parameters, which is consistent with the expectation that longer, more uniform fibers also tend to exhibit higher strength and elongation. These parameters are also heavily loaded on PC-1, implying that they are major contributors to the overall variability in fiber quality.
• Color and Micronaire. The Micronaire (Mic), Reflectance (Rd), and Yellowness (+b) values are grouped together in the lower right quadrant. These parameters appear to be positively correlated but show a negative correlation with short fiber content and trash content. This clustering suggests that higher micronaire values, which indicate coarser fibers, are associated with higher reflectance and less yellowness. This relationship could be due to the influence of fiber maturity on color measurements, where mature fibers (higher Mic) tend to have less variability in color characteristics.
• Short Fiber Content and Trash Content. The Short Fiber Content and Trash Content are located in the lower and left quadrants, respectively, and are negatively correlated with most of the other quality parameters, particularly those related to fiber length and strength. This inverse relationship underscores the detrimental impact of short fibers and trash on overall cotton quality. The positioning of these contributes suggests that they are less influential on PC-1 but still contribute to the variability captured by PC-2.
• Area %. The Area %, which is often related to the coverage of fibers on the test sample, appears somewhat isolated but still correlates positively with Trash Content. This might indicate that as trash content increases, the apparent area covered by the fibers also increases, which could be due to the inclusion of non-lint materials.

Implications

The following section explores the broader implications of the findings from Field 1, focusing on how the different defoliant duty cycles influence cotton production practices. The analysis provides insights into the practical applications of the 40%, 20%, and 60% duty cycles, with an emphasis on optimizing cotton fiber quality and yield. These implications offer guidance for improving defoliant application strategies and enhancing overall cotton harvesting efficiency in similar agricultural settings.

The PCA results indicate that fiber length, strength, and uniformity are the most significant factors influencing the overall quality of cotton as measured by HVI, as these parameters heavily load on the first principal component. This suggests that any variation in these properties will have the largest impact on the perceived quality of the cotton. On the other hand, while still relevant, short fiber content and trash content contribute more to secondary sources of variability (PC-2).

The negative correlation between high-quality fiber attributes (length, strength) and undesirable traits (short fibers, trash) further emphasizes the importance of minimizing the latter during cotton processing to maintain high-quality output. The distinct separation of these variables in the PCA plot allows for a clear understanding of how different factors contribute to the overall cotton quality profile.

3.1.2. HVI—Score Plot

The Score Plot (Figure 8) shows the distribution of treatment groups across the principal components, reflecting the effects of the different duty cycles (20%, 40%, 60%) and the control (100%) on cotton quality:

• 20% Duty Cycle (Red Circles). The points are dispersed across the plot, particularly along PC-1, indicating higher variability in fiber quality with this lower duty cycle.
• 40% Duty Cycle (Green Triangles). This group displayed more clustering near the center of the plot, suggesting more consistent and stable cotton quality with this duty cycle.
• 60% Duty Cycle (Blue Diamonds). Points for this treatment are moderately spread, with some variability along PC-2, yet showing relatively stable outcomes.
• Control (100%, Blue Squares). The control points include a significant outlier, indicating a distinct and potentially less favorable cotton quality profile when no modulation is applied.

3.1.3. HVI—Influence Plot

The Influence Plot (Figure 9) provided further insights by highlighting the influence of each sample on the PCA model, as indicated by Hotelling’s T² and F-residuals:

• The most notable outlier is from the control group (100%), which lies far to the right of the plot, beyond Hotelling’s T² limit. This suggests that the control group significantly influences the model, potentially distorting the overall analysis.
• The 20%, 40%, and 60% duty cycles show varying levels of influence, but none exceed critical limits, indicating that their contributions to the model are within acceptable bounds.

3.1.4. HVI—Explained Variance Plot

The Explained Variance Plot (Figure 10) adds a crucial dimension to the analysis by showing how much variance each principal component (PC) explains. The plot indicates that PC-1 captures the majority of the variance, approximately 80%, while PC-2 captures an additional 9%. Beyond PC-2, the additional components (PC-3 to PC-7) contribute minimal additional variance, with the total variance leveling off around 95–100%.

This plot confirms that the first two principal components are the most significant, validating the focus on PC-1 and PC-2 in the previous analyses. The sharp rise in explained variance from PC-0 to PC-1 highlights the dominant role of the primary quality parameters in determining cotton fiber quality. The diminishing returns from additional components suggest that further PCs contribute little to explaining the variability in the dataset, reinforcing the decision to consider the first two components primarily.

Combining insights from the Correlation Loading, Score, Influence, and Explained Variance Plots provides a comprehensive understanding of the impact of different defoliant duty cycles on cotton quality. PC-1 emerges as the dominant source of variability, heavily influenced by key quality parameters such as fiber length and strength, with the Explained Variance Plot showing that PC-1 accounts for 80% of the total variance. In contrast, PC-2 captures additional variability, particularly associated with less desirable traits like short fiber content and trash content but contributes less significantly to the overall variance.

The Influence Plot reveals a significant outlier in the control group, indicating that traditional spraying methods may lead to more extreme variations in cotton quality. Among the treatments, the 40% duty cycle emerges as the most consistent in maintaining stable cotton quality, while the 20% and 60% cycles introduce more variability, though still within acceptable limits. These results underscore the importance of selecting an appropriate duty cycle to optimize cotton quality, with the 40% duty cycle appearing to be the most effective in balancing consistency and quality. The PCA analysis, supported by the explained variance and influence assessments, provides a robust framework for understanding the impact of defoliant application rates on cotton fiber properties.

3.2. PCA Results for HVI Based on Correlation Loading, Score, Influence, and Explained Variance Plots of Field 2

3.2.1. HVI—Correlation Loading Plot

The Correlation Loading Plot (Figure 11) for Field 2 illustrates the relationships among the key cotton fiber quality parameters measured by the High-Volume Instrument (HVI). The first principal component (PC-1) explains 79% of the total variance, while the second principal component (PC-2) accounts for an additional 11%. This variance distribution indicates that most of the variability in fiber quality can be explained by the factors represented by PC-1, with PC-2 capturing secondary, yet still relevant, sources of variation.

Key Observations

In Field 2, the analysis focused on the same fiber quality parameters under varying defoliant duty cycles. This section highlights how different application rates influenced fiber length, strength, and uniformity and provides a comparative analysis of the 20%, 40%, and 60% duty cycles, along with the control treatment. The observations offer insights into this specific field’s optimal defoliant application strategies.

• Fiber Length and Strength Parameters. The Upper Half Mean Length (UHML), Strength (Str), Elongation (Elo), and Length Uniformity Index (UI) are closely grouped in the lower left quadrant of the plot. These parameters show strong positive correlations with each other and are heavily loaded on PC-1. This suggests that longer, stronger, and more uniform fibers are the primary contributors to the overall cotton quality in Field 2, similar to the findings in Field 1.
• Micronaire and Elongation. The Micronaire (Mic) and Elongation (Elo) values are positioned closer to the center of the plot, showing moderate loading on PC-1 and some influence on PC-2. This positioning indicates that these factors contribute to the overall variability but are not as dominant as fiber length and strength parameters.
• Trash Content and Short Fiber Content. Trash Content and Short Fiber Content are located in the upper right quadrant, indicating a strong positive correlation with each other and a negative correlation with the key quality traits (length, strength, uniformity) found in the lower left quadrant. These undesirable traits are more closely associated with PC-2, suggesting that while they contribute to the variability in cotton quality, their impact is secondary compared to the more critical parameters.
• Color and Area %. The Reflectance (Rd), Yellowness (+b), and Area % are dispersed across the plot, with Rd and +b leaning towards the upper left and Area % towards the upper right. These parameters show varying correlations with both PC-1 and PC-2, indicating that they contribute to the overall fiber quality but do not dominate the variance explained by either component.

Implications

This section delves into the implications of the results from Field 2, examining how the defoliant duty cycles impact cotton fiber quality under varying conditions. By analyzing the outcomes of the 20%, 40%, and 60% duty cycles, along with the control treatment, the discussion will outline the practical takeaways for cotton producers, offering recommendations for refining defoliant use to maximize both quality and productivity in different field environments.

The PCA results for Field 2 reinforce the importance of fiber length, strength, and uniformity as the dominant factors influencing cotton quality, as these parameters are heavily loaded on PC-1. The negative correlation between these high-quality traits and undesirable traits like trash content and short fiber content underscores the need to minimize these less desirable elements to maintain high-quality cotton. The moderate influence of micronaire, elongation, and color-related parameters suggests that these factors are relevant but less critical than the primary quality indicators.

Overall, the Correlation Loading Plot for Field 2 highlights the consistent importance of fiber length, strength, and uniformity across different fields while acknowledging the role of secondary factors like trash content and color in shaping the overall cotton quality profile.

3.2.2. HVI—Score Plot

The Score Plot (Figure 12) for Field 2 provides a visual representation of how the different treatment levels—20%, 40%, 60% duty cycles, and the control (100%)—affect the cotton quality parameters, as captured by the first two principal components (PC-1 and PC-2). In this plot, PC-1 explains 79% of the total variance, while PC-2 accounts for an additional 11%, reflecting the major sources of variability in the data.

Key Observations

• 20% Duty Cycle (Red Circles). The 20% duty cycle points are spread across the plot, particularly along PC-1, indicating significant variability in the cotton quality parameters at this lower duty cycle. Some points are positioned far to the right of the plot, suggesting that this treatment can lead to distinct outcomes, possibly reflecting variations in fiber length, strength, and other critical parameters.
• 40% Duty Cycle (Green Triangles). The points corresponding to the 40% duty cycle are more tightly clustered near the center of the plot, particularly along PC-1. This clustering suggests that the 40% duty cycle produces more consistent and stable cotton quality, with less variability across the key parameters. This result mirrors the findings from Field 1, indicating that the 40% duty cycle may be optimal for achieving consistent fiber quality.
• 60% Duty Cycle (Blue Diamonds). The 60% duty cycle points are more dispersed, especially along PC-2. This dispersion indicates some variability in cotton quality, particularly in the secondary parameters captured by PC-2, such as short fiber content and trash content. However, the points remain relatively close to the center, suggesting that the overall impact on fiber quality is moderate.
• Control (100%, Blue Squares). The control points (100%) are also somewhat spread out but show a tendency to cluster near the center-left of the plot. However, there are outliers, particularly on the left side, indicating that traditional spraying methods without modulation can lead to more variability in cotton quality. These outliers suggest that the control treatment may result in less consistent fiber quality outcomes compared to the modulated treatments.

Implications

The Score Plot for Field 2 supports the findings from Field 1, indicating that the 40% duty cycle consistently produces the most stable and uniform cotton quality across different treatment conditions. The wider spread of points for the 20% and 60% duty cycles suggests that these treatments introduce more variability, particularly in critical fiber quality parameters like length and strength, as captured by PC-1. The control treatment shows some outliers, indicating that cotton quality can be more variable and potentially less predictable without modulation.

Overall, the Score Plot reinforces the idea that the 40% duty cycle is the most effective for maintaining consistent cotton quality, while the 20% and 60% duty cycles, as well as the control, introduce varying degrees of variability that could affect the overall quality of the harvested cotton.

3.2.3. HVI—Influence Plot

The Influence Plot (Figure 13) for Field 2 provides insights into the influence of individual samples on the PCA model, particularly focusing on Hotelling’s T² and F-residuals, which help identify outliers and influential data points. The x-axis represents Hotelling’s T², which measures the overall influence of each sample on the PCA model, while the y-axis shows the F-residuals, indicating the degree of deviation of each sample from the PCA model.

Key Observations

• Control (100%, Blue Squares). The control samples (100%) are relatively widespread along Hotelling’s T² axis, with one sample nearing the critical limit but not exceeding it. This suggests that while the control treatment does have a noticeable influence on the PCA model, it does not introduce extreme outliers. However, the spread indicates variability in the cotton quality parameters under the control condition, which is consistent with the Score Plot findings.
• 20% Duty Cycle (Red Circles). The 20% duty cycle samples are primarily clustered around moderate Hotelling’s T² values, with some variability in F-residuals. These samples show a moderate influence on the PCA model, suggesting that the 20% duty cycle introduces variability but not to an extreme degree. The positioning indicates that this treatment affects cotton quality in a relatively consistent but varied manner.
• 40% Duty Cycle (Green Triangles). The 40% duty cycle samples are tightly clustered with low Hotelling’s T² values, indicating minimal influence on the PCA model. This clustering suggests that the 40% duty cycle produces the most consistent cotton quality with the least variability, reinforcing the findings from both the Correlation Loading and Score Plots. The low influence of these samples on the PCA model highlights the stability and predictability of this treatment.
• 60% Duty Cycle (Blue Diamonds). The 60% duty cycle samples show a wider spread, particularly in the higher range of Hotelling’s T² values, with a few points approaching the critical limit. This suggests that the 60% duty cycle introduces more variability in cotton quality parameters, making these samples more influential on the PCA model. The positioning also indicates that the 60% duty cycle may produce some outliers, though none exceed the critical limits.

Implications

The Influence Plot for Field 2 supports the conclusion that the 40% duty cycle is the most consistent and least variable treatment, showing minimal influence on the PCA model. While still within acceptable limits, the 20% and 60% duty cycles exhibit more variability, as indicated by their spread along the Hotelling’s T² axis. The control treatment shows some influence, with a broader spread of samples, indicating variability in cotton quality outcomes under traditional spraying methods.

These findings underscore the importance of selecting an appropriate duty cycle for defoliant application. The 40% duty cycle consistently emerges as the most reliable choice for maintaining stable and predictable cotton quality, while the 20% and 60% cycles, along with the control, introduce varying degrees of influence that could lead to more variability in the final fiber quality. This analysis, combined with the results from the Correlation Loading and Score Plots, provides a comprehensive understanding of how different treatments impact cotton quality in Field 2.

3.2.4. HVI—Explained Variance Plot

The Explained Variance Plot (Figure 14) for Field 2 visually represents how much variance each principal component (PC) captures in the dataset. This plot is essential for understanding the significance of each component in explaining the variability in cotton fiber quality parameters as measured by the HVI.

Key Observations

• Dominance of PC-1. The plot shows that PC-1 captures the majority of the variance, accounting for approximately 79% of the total variability in the dataset. This confirms that PC-1 is the most critical component, heavily influenced by key quality parameters such as fiber length, strength, and uniformity. These are the primary factors driving the differences in cotton quality observed in Field 2.
• Contribution of PC-2. PC-2 explains an additional 11% of the variance, making it the second most significant component. While its contribution is much smaller than that of PC-1, PC-2 still plays a crucial role in capturing secondary variability, particularly related to less desirable traits such as short fiber content and trash content.
• Diminishing Returns from Subsequent PCs. The variance explained by subsequent principal components (PC-3 to PC-7) diminishes rapidly, with each additional component contributing only a small percentage of the total variance. By the time PC-3 is included, the cumulative variance explained approaches 90%, and beyond PC-4, the additional components add very little to the overall understanding of the data.

Implications

The Explained Variance Plot for Field 2 highlights the critical importance of PC-1 in explaining the majority of the variability in cotton fiber quality. This aligns with the findings from the Correlation Loading and Score Plots, where fiber length, strength, and uniformity were identified as the dominant factors influencing cotton quality. The significant contribution of PC-2 indicates that while secondary factors like short fiber content and trash content are less influential, they still play a role in the overall quality assessment.

The rapid decline in explained variance beyond PC-2 suggests that the first two components are sufficient to capture the most relevant information in the dataset. This reinforces the focus on PC-1 and PC-2 in previous analyses, as these components provide the most meaningful insights into how different defoliant duty cycles impact cotton quality in Field 2.

Overall, the Explained Variance Plot supports the conclusion that the key drivers of cotton quality are well captured by the first two principal components, with PC-1 being the most critical. This analysis will be integral when combining the results from Field 2 with those from Field 1 to develop a comprehensive understanding of the impact of defoliant application rates on cotton fiber properties.

4. Discussions on the Summary of Results for Field 1 and Field 2

The PCA conducted on the cotton fiber quality parameters for both Field 1 and Field 2 provides a comprehensive understanding of the impact of different defoliant duty cycles on cotton quality.

Key Findings

The following section summarizes the key findings from the comparative analysis of Field 1 and Field 2. Both fields were assessed using different defoliant duty cycles—20%, 40%, and 60%—as well as a control treatment. The analysis highlights the influence of each duty cycle on cotton fiber quality, focusing on critical parameters such as fiber length, strength, and uniformity. These findings provide important insights into how different application rates affect cotton production outcomes, offering practical guidance for optimizing defoliant use in varying field conditions. The discussion details the performance of each duty cycle and its implications for cotton quality across both fields.

• Dominance of Key Quality Parameters. In both fields, PC-1 emerged as the dominant source of variability, explaining approximately 80% of the total variance in Field 1 and 79% in Field 2. This component is heavily influenced by critical cotton quality parameters such as Upper Half Mean Length (UHML), Strength (Str), Elongation (Elo), and Length Uniformity Index (UI). These parameters consistently emerged as the primary drivers of cotton quality across both fields.
• Role of PC-2. PC-2, accounting for 9% of the variance in Field 1 and 11% in Field 2, captured secondary variability related to less desirable traits such as Short Fiber Content and Trash Content. While these factors are less influential than those captured by PC-1, they still play a significant role in shaping the overall cotton quality profile.
• Consistency of the 40% Duty Cycle. The Score Plots for both fields consistently showed that the 40% duty cycle produced the most stable and uniform cotton quality, with minimal variability across key parameters. This suggests that the 40% duty cycle is the most effective in balancing consistency and fiber quality across different field conditions.
• Variability in 20% and 60% Duty Cycles. Both the 20% and 60% duty cycles introduced more variability in cotton quality, as evidenced by the wider spread of data points in the Score and Influence Plots for both fields. The 20% duty cycle, in particular, showed greater variability in Field 2, indicating that lower application rates may lead to inconsistent fiber quality outcomes.
• Influence of Control Treatments. The Influence Plots revealed that the control treatments (100% duty cycle) had a noticeable impact on the PCA model, with significant variability and outliers observed in both fields. This suggests that traditional spraying methods without modulation can lead to less predictable cotton quality, potentially compromising the uniformity and overall quality of the harvested cotton.
• Explained Variance. The Explained Variance Plots for both fields confirmed that the majority of the variability in cotton quality is captured by the first two principal components (PC-1 and PC-2). Subsequent components contributed minimal additional variance, reinforcing the focus on these primary components in the analysis.

5. Conclusions

The PCA across Field 1 and Field 2 revealed that the 40% duty cycle consistently produced the most stable and uniform cotton quality, minimizing variability across key parameters such as fiber length, strength, and uniformity. This trend was consistently observed in both fields, confirming the 40% duty cycle as the most effective balance between defoliant application and cotton quality. The stable clustering of data points associated with this duty cycle suggests a predictable outcome crucial for optimizing cotton harvesting and maintaining high fiber quality standards.

In contrast, the 20% and 60% duty cycles and the control treatment introduced greater variability. The 20% duty cycle, particularly in Field 2, showed significant variability, indicating that lower application rates can lead to inconsistent quality. While more stable, the 60% duty cycle still showed variability in secondary parameters such as short fiber content and trash content. The control treatment exhibited the highest variability and outliers, indicating that traditional spraying methods are less reliable for producing consistent fiber quality.

While this study highlights the 40% duty cycle as optimal, further research should explore its robustness across different environmental conditions and soil types. Investigating how environmental factors such as temperature, soil moisture, and weather conditions might influence cotton fiber quality in conjunction with defoliant application is important. Additionally, incorporating AFIS data in future studies will offer another insight into fiber fineness and maturity, complementing the HVI findings. Furthermore, investigating advanced technologies like remote sensing and machine learning for real-time defoliant adjustments and exploring a broader range of defoliant chemicals will further enhance cotton quality and production practices.