Variations in Pesticide Practices Among Indonesian Farmers Across Different Commodities

2.1 Survey design and data collection

The study was conducted from June to December 2020 across five provinces in Indonesia: West Java, Banten, Central Java, North Sumatra, and Lampung (Figure 2), collectively accounting for 14.7% of all provinces in the country. These provinces were selected due to their significance as major agricultural production centres representing a diverse range of commodities, including food crops (rice, maize), horticultural crops (shallot, red chili, potato), and estate crops (oil palm). The selection was based on their unique agroecological conditions and the variety of cultivated commodities, which are critical to understanding the diverse pesticide use practices among Indonesian farmers. This purposive selection helps capture a broad spectrum of farming practices and socio-economic contexts that influence pesticide use.

2.jpg

Figure 2. Research sites and production ranking in the country

A quantitative approach was employed, using a survey method and individual interviews with a structured questionnaire to gather data from 354 randomly selected respondents across 33 sub-districts and 73 villages within these five provinces. Before commencing the study, the farmers were informed about the purpose of the survey by the local extension agents and sought their consent to be interviewees. The survey was designed to gather comprehensive information on various aspects of pesticide use practices. To ensure validity and reliability, the survey questions were constructed based on an extensive review of existing literature on pesticide use, farmer behavior, and agricultural practices, as well as consultations with experts in the field. The structured questionnaire included both closed and open-ended questions covering the following key themes:

1. Demographic information: Age, education level, farming experience, land ownership status, and land size.
2. Pest and disease management: Types of pests and diseases encountered, estimated level of crop damage from pests and diseases if pesticides are not applied, perceived effectiveness of pesticides used, pesticide purchasing patterns, types and number of pesticide brands used, frequency and methods of pesticide application, mixing practices, and the rotation of pesticide brands used.
3. Economic factors: Costs associated with pesticide purchases and financial considerations influencing pesticide use decisions.
4. Knowledge and skills: Participation in pesticide training.
5. Information sources: Access to and use of information sources, such as extension services, fellow farmers, pesticide retailers, and mass media, to obtain pesticide information.

The questionnaire was designed to gather both quantitative data (e.g., the number of pesticide brands used and the cost of pesticides) and qualitative data (e.g., farmers' motivations, perceptions, and decision-making processes regarding pesticide use). To enhance reliability, the survey was pre-tested with a small group of farmers to refine question clarity and relevance.

A total of 354 respondents, classified based on six commodity cohorts, participated in this study, as detailed by regency and commodity in Table 2. They were rice (83 farmers), maize (60 farmers), shallot (63 farmers), red chili (60 farmers), potato (58 farmers), and oil palm (30 farmers).

Table 2. Total samples and commodities from the research site

2.2 Data analysis

Descriptive and inferential statistics were used to analyze the survey results, providing a comprehensive understanding of the variations in pesticide use and the factors driving these differences. The study's variables are summarized in Table 3.

Table 3. The classification of variables used in the study is based on the methods of analysis

The non-parametric methods employed in this study are specifically the Kruskal-Wallis test and Post-Hoc analysis. Using the Kruskal-Wallis test and subsequent Mann-Whitney U tests allowed for a robust analysis of differences in pesticide use practices across multiple commodity groups, aligning with the study's aim to explore the variations in farmers' behavior. These methods were chosen due to their suitability for comparing multiple groups with non-normally distributed data. The Kruskal-Wallis test, a non-parametric method, was selected to determine whether there are statistically significant differences in pesticide use practices among different commodity groups (food crops, horticulture, and estate crops). This test is appropriate for comparing three or more independent groups when the data do not meet the assumptions required for parametric tests, such as normality and homogeneity of variance. In this study, key variables like the number of pesticide brands used, pesticide costs, types of pests and diseases, and seasonal purchasing patterns were measured on an ordinal or non-normally distributed scale, making the Kruskal-Wallis test suitable for analysing these differences across multiple groups. Following the Kruskal-Wallis test, post-hoc tests using the Mann-Whitney U test were conducted to identify specific group differences. The Mann-Whitney U test, also a non-parametric test, was used to perform pairwise comparisons between groups to determine which specific commodity groups differ in pesticide use practices. This test is appropriate when comparing two independent samples that do not follow a normal distribution, aligning with the nature of the data in this study.

Before applying the Kruskal-Wallis and Mann-Whitney U tests, key assumptions were checked:

• Normality: The Kolmogorov-Smirnov test was applied to test the normality of the data for key variables. The results indicated that the data were not normally distributed (p < 0.05), justifying the use of non-parametric methods.
• Independence of Observations: Each respondent's data were treated as independent, fulfilling the assumption of independence required for both the Kruskal-Wallis and Mann-Whitney U tests.
• Homogeneity of Variances: While the Kruskal-Wallis test does not require the assumption of equal variances, the assumption was still checked to better understand the data's characteristics.

Table 4. Output of the test of normality, Kolmogorov-Smirnov

In the statistical analysis, the Kolmogorov-Smirnov normality test was applied to test the normal distribution of the data for four variables. The null hypothesis (H0) for the normality test implies that the population was normally distributed, while the alternative hypothesis (H1) means the opposite. This test showed that the data were not normally distributed, and all four variables had significant values less than the reference alpha value (0.05), which means rejecting the null hypothesis (Table 4).

The Kolmogorov-Smirnov normality test can be used for a single sample to determine whether the distribution of scores in the sample matches the distribution of scores in a given theoretical or empirical population. Thus, the null hypothesis was expected to be rejected. Compared to the other types of inferential statistical tests, the goodness-of-fit test is, to some extent, unique because researchers mostly expect to retain the null hypothesis or to prove that the sample comes from a specific type of distribution (for example, a normal distribution). Alternatively, by using most other inferential tests, the researcher expects to reject the null hypothesis or demonstrate that one or more samples are not from a specific or similar population. It is important to highlight that the alternative hypotheses for a goodness-of-fit generally do not specify the alternative distribution. It would be the most likely distribution for the data if the null hypothesis is rejected. The Kolmogorov-Smirnov normality test is used at least on continuous and ordinal data types [23].

Since the data were not normally distributed, the Kruskal-Wallis test was employed. The Kruskal-Wallis test is a statistical analysis for non-parametric data [23, 24]. The comparison between the groups can be solved using this test since the data did not meet the assumptions of the variance analysis [24]. The sample groups were divided into different commodities. Therefore, the hypotheses of the study were symbolized with the equations below:

1. $H_1: \theta_1=\theta_2=\theta_3=\theta_4=\theta_5$, as the null hypothesis, which means that the median of the population grouped by commodities is equal.
2. $H_1: \theta_1 \neq \theta_2 \neq \theta_3 \neq \theta_4 \neq \theta_5$, as the alternative hypothesis, which means there is a difference between at least two of the k = 5 population’s medians.

The Kruskal-Wallis test result showed a significant difference, so further testing was required. A post-hoc test using the Mann-Whitney U test was employed to analyse the difference across the commodity groups. All statistical tests were conducted using SPSS Version. 28.

Based on the descriptive analysis of farmers’ characteristics (Table 5), all farmers are classified as working age groups [25], indicating the ability to run farms in the long term. The use of production inputs is differentiated by age variations, and age has a significant impact on agricultural output [26]. One of the most critical factors influencing farm productivity is education. Farmers who respond pass the minimum requirement of basic formal education, demonstrating that they are literate farmers (have reading and writing skills). Achieving a certain level of education will help farmers manage information and implement it in their farming [27]. However, the study finds a slightly different result in oil palm farmers with the highest average formal education.

Longer farming experience leads to accumulated knowledge and skills, which allegedly can influence farming productivity. Study [28] stated that farming experience over time has a significant impact on technology adoption. However, in this study, having a long farming experience does not necessarily indicate that farmers are willing to use pesticides in the best way possible. Farmers with an average of 18 years of experience, for example, do not fully practice proper pesticide mixing and rotation. Table 7 corroborates that farmer age does not significantly influence rotation and mixing behavior.

Farmers' decisions to use more inputs may be influenced by their land status. Study [29] underlined the significant and positive impact of land ownership and land area on increasing technical efficiency and affecting output value. Land ownership and land area have a significant and positive impact on increasing technical efficiency and influencing output value. Other studies found that increasing farm size has the opposite effect on pesticide use, with the quantity of pesticides decreasing [8, 30]. Meanwhile, the farmers' status in this study demonstrates that not all farmers manage their own land. Landowners may apply more or fewer pesticides depending on short-term land productivity [4]. Farmers acting as owners make the decisions about how to manage their farms, including whether to buy and use more pesticides or less. Some cultivators receive information about pesticide use from landowners.

The use of pesticides corresponds to the control of pests and diseases. Pest and disease attacks have occurred in all the respondents' farms. On the other hand, the average damage caused by pests and diseases is nearly identical. This value is predicted based on farmers' perceptions of what would happen if they did not use the current type and amount of pesticide that they are using. Each pesticide has a different level of efficacy. The efficacy of pesticides based on farmers' perceptions is also investigated because it can help explain why they buy a specific brand or type of pesticide. The positive view of farmers toward pesticide efficacy is one of the factors that affected the perception of the adoption of practices of pesticide safety [31]. However, because there is a perception gap between extension agents and farmers on efficacy, farmers' perceptions may lead to a false notion of pesticide efficacy [32].

Aside from analysing farmers' idiosyncratic, two related issues were also collected and analysed: pesticide training attended by farmers and farmers' attitudes toward pesticide use (Table 5). Training is a non-formal education that aims to improve farmers' knowledge and skills in applying pesticides safely and correctly. This study's training information excludes unintentional informal consultations or meetings. Knowledge, behavior, and beliefs differ significantly between pesticide training participants and non-participants [33], and the attendance of horticulture farmers in training has a positive impact on aspects of pesticide use [30]. Study [34] revealed that only about 29 percent of farmers in Iran have ever received pesticide management training, and the training intervention has helped to improve farmers' attitudes toward pesticide management.

Ideally, training would educate farmers on the best practices for using pesticides safely, wisely, and correctly. This study discovered that more than half of farmers had attended training on pesticides; however, they continued to use pesticides in an unsafe, unwise, and improper manner, as revealed by three defined attitudes in Table 6. The first attitude investigated in this study is about purchasing for control or prevention. This preventive action implies that farmers applied pesticides before the attack. Farmers concerned about pest and disease attacks in previous seasons or by the experiences of fellow farmers, usually practice this habit. They typically apply a lower dose of pesticide, which is then increased in response to an increase in the number of pest and disease attacks. Farmers plan to avoid the loss in advance because it will be more difficult to control after the outbreak, or result in a higher risk of loss. In other words, actions for prevention will result in a greater use of pesticides than actions for control. This precaution will eventually become a farmer's habit, providing them with a sense of security in ensuring their harvest yield.

The second attitude concerns pesticide mixing (Tables 6 and 7). Pesticide mixing is motivated mainly by the perceived need to enhance pest and disease control and reduce labor costs. However, it is associated with potential risks. Inappropriate mixing can reduce pesticide efficacy [35, 36] and lead to excessive residues and environmental pollution risks [37]. In addition, pesticide mixing is sometimes associated with potential hazards when farmers mix pesticides. Farmers do not always wear standard personal protective equipment, which can endanger their health due to pesticide exposure [10, 38-40]. Nonetheless, the main concern is mixing pesticides according to the instructions to maintain the efficacy level of active ingredients in pesticide products and avoid pest and disease resistance. As a result, farmers' knowledge and skills in pesticide handling are critical, as well as promoting training programs that emphasize the environmental and health consequences of unsafe mixing practices. Strengthening farmers’ understanding of the linkage between mixing behavior, residues, and environmental risks is crucial for promoting safer and more sustainable pest management.

The last attitude discussed in Table 6 and Table 7 is regarding pesticide rotation. However, pesticide rotation was less common, with only 31% of farmers rotating different pesticides across multiple planting seasons. This study suggests that one reason for the limited practice of pesticide rotation is that farmers often use the same pesticides for different commodities planted in the previous season. Additionally, in some cases, the lack of rotation is due to farmers relying on only one type of pesticide to control pests, reducing the need or incentive to rotate pesticides. Study [41] stated that pesticide rotation is one of the effective methods of delaying pest-resistant emergence. This statement is also supported by the study [42]. The study emphasizes that rotating insecticides is an important part of resistance management. At the same time, this practice will help delay the development of resistance to one or more insecticides used. Most farmers may be unaware of the importance of pesticide rotation in preventing resistance. In this study, rotating pesticides does not refer to the mode of action but rather to pesticide brand loyalty. In other words, farmers appear to use the same brands over long periods of cultivation. Similarly, a study of rice farmers' satisfaction and loyalty to specific pesticide brands in West Java province, Indonesia, found that only 11% of farmers become switch buyers, while the rest are habitual to committed buyers [43].

The study reveals significant differences in the behavior of pesticides used among farmers across various commodity groups in Indonesia, notably in the number of pesticide brands, costs, pest and disease types, and seasonal purchasing behaviors. One of the reasons, in addition to increased pesticide use, could be the availability of pesticide brand options. Regarding differences in the number of pesticide brands used, this study reveals that potato farmers use the most pesticide brands due to a high incidence of pests and diseases. In contrast, oil palm farmers face fewer pest types like beetles and fungi (Ganoderma), and rely primarily on herbicides for weed control [44]. Weed is one of the main obstacles to increasing the potential yield of oil palm production in Indonesia [45]. According to study [46], oil palm farmers extensively use herbicides for weed control.

Further analysis reveals that rice, shallot, red chili, and potato farmers use a larger number of pesticide brands (Table 10). Farmers cultivating rice, shallot, red chili, and potato tend to purchase more pesticide brands due to the increased number of pest and disease attacks. On the other hand, oil palm and maize farmers choose fewer pesticide brands because maize and oil palm farming are not as intensively maintained as the other commodities in the research location. Pests and diseases do not attack these two plants as frequently or as frequently as they do rice, chili, shallot, and potato. Two maize commodity research sites exist in the Provinces of Lampung and Banten. Maize farmers in Lampung work more intensively in farming management than farmers in Banten, located near a forest and are not being treated optimally. As a result, the use of pesticides differs between the two locations. Furthermore, the majority of the damage to maize farming is caused by uncontrollable animal attacks (wild boars and monkeys). Meanwhile, the research locations for oil palm commodities (Lampung and North Sumatra Provinces) share similarities in farming patterns and farmers' pesticide use behavior, primarily for weed control.

Table 10. The grouping of commodities based on the four variables examined

Pesticide brands purchased by horticulturists (shallot, red chili, and potato farmers) and rice farmers are more diverse. Insecticides, fungicides, and herbicides are the three types of pesticides that all respondents commonly use. In contrast, farmers rarely use molluscicides and bactericides due to the rarity of attacks from these pests and disease types. Furthermore, according to interviews, each farmer who grows red chili and potatoes buys six pesticide brands on a regular basis, whereas those who grow rice and shallots buy five. Farmers of maize and oil palm, on average, use three and two brands, respectively. Insecticides, more than any other type, contribute significantly to the variety of pesticide brands. Even though oil palm farmers use only one insecticide brand, the other farmer groups purchase 2-3. Similarly, farmers who plant red chili and potato crops use a lot of fungicides. Many vegetable farmers spray insecticides and fungicides excessively or on a regular basis [18, 47], whereas oil palm and maize farmers use more herbicide brands than the other farmer groups. A study [16] supported these findings, indicating that 60-90% of horticultural farmers in East Java use at least 6 to 9 pesticide brands. This intensive use underscores the need for improved pesticide management to avoid resistance, environmental harm, and health risks, advocating for more efficient and sustainable practices.

A horticultural commodity such as chili has higher operational costs compared to rice and maize. The extended chili planting season, which can last for multiple harvests, often necessitates more pesticides. Additionally, chili farmers are accustomed to using various brands of pesticides. This finding is consistent with a study [15], which revealed that the longer growing season of crops leads to a higher frequency of pesticide applications per season, as observed in tomato farming in Ethiopia.

Further interviews with farmers found that horticultural farmers incur a higher proportion of pesticide costs per hectare, amounting to 43% of total farming costs during the rainy season and 39% during the dry season. This aligns with using a greater variety of pesticide brands and higher doses compared to farmers of staple crops and plantations. It was also discovered that the pesticide costs remain similar during both planting seasons. Farmers attempt to manage fluctuations in pesticide costs by adjusting dosages, mixing products, and seeking price differences for purchasing pesticides. However, these efforts may increase the risk of farmers using counterfeit pesticides, which are generally cheaper.

According to study [48], the percentage of pesticide use could increase or decrease in line with the expansion or reduction of cultivated land areas. Various previous studies conducted in different regions of Indonesia have found that land ownership status affects pesticide costs. For example, rice farmers in Jember who own their land incur higher pesticide costs than tenant farmers and sharecroppers, at 2.3%, while tenant farmers and sharecroppers spend 0.9% and 1.7% of their total farming costs on pesticides, respectively [49]. The use of superior seed varieties can also contribute to the rise in pesticide costs. For instance, the cultivation of high-yield maize varieties in Gorontalo requires 43% higher pesticide costs compared to local varieties [50]. Although farmers often complain about pesticide costs, their proportion is still lower compared to input costs for fertilizers and quality seeds. For example, highland potato farming requires pesticide costs amounting to 3.8% of the total farming costs [51], while garlic farming in highland areas of several central districts involves substantial pesticide use, consuming 5.33-12.15% of total farming costs [52]. However, chili farmers in Jember, East Java, spend 8.65% of their total costs on pesticides, which is higher than their seed costs of around 4.58% [53].

Production function analyses of rice [54-56], maize in Gorontalo [50], large red chili in East Java [53, 57], and shallots in Lampung [58] show that pesticide costs significantly and substantially affect agricultural production. However, increased pesticide costs in other studies do not always improve crop conditions. For example, the increase in pesticide costs had no significant effect on the production of horticultural crops such as shallots [59], tomatoes [60], oil palm [61], or rice in Jember [49]. This indicates that the impact of pesticide costs on the production of various commodities in Indonesia yields conflicting research findings.

In terms of differences in the type of pests and diseases, this study confirms that rice crops have the most variety of pests and diseases, while oil palm has the least. Pests and diseases vary by crop, and those that attack rice crops may not be found in horticultural crops, but the same types of pests and diseases may be found in both shallot and red chili. These phenomena necessitate a different approach to control and manage pests and diseases, such as selecting the appropriate pesticides. As a result, farmers' knowledge of identifying pests and diseases is critical, which leads to the selection of the appropriate pesticides.

The findings from Tables 9 and 10 highlighted that pest and disease management strategies vary significantly across different commodity groups due to the diversity of crops and their specific vulnerabilities. For Group Commodity 1, which includes a range of crops such as red chili, rice, shallot, and potato, farmers face a wide array of pests and diseases, necessitating a comprehensive and adaptable approach to pest management. This complexity requires tailored solutions for each crop, involving research investments, pest-resistant varieties development, integrated pest management techniques, and farmer education. In Group Commodity 2, which focuses on maize and potato, the shared pests and diseases simplify pest management to some extent. However, targeted strategies are still needed to address specific issues for each crop. Research should focus on identifying and managing these unique threats. Meanwhile, Group Commodity 3, dealing solely with oil palm, faces more specialized pest and disease challenges, allowing for the development of highly focused management strategies. This specialization means fewer types of pesticides may be used, but solutions need to be tailored specifically to the needs of oil palm cultivation. Overall, these findings suggest that resource allocation, research, and policy should be tailored to the specific needs of each commodity group to manage pest and disease challenges effectively.

This study also found differences in seasonal pesticide purchasing patterns. Rice farmers tend to buy pesticides for one season at a time, while shallot farmers prefer to buy in bulk to stock up for the next planting season. Rice farmers' seasonal purchasing habits differ significantly from the shallot and red chili farmers', implying that food crop farmers will purchase pesticides differently during each cropping season. Aside from the pests and diseases already discussed in Table 5, this behavior may be influenced by rice farmers' financial ability to purchase pesticides. Due to financial constraints, these farmers can only provide pesticides for one season. Another reason is that horticultural farmers admit that obtaining certain pesticides is difficult, particularly before planting season, due to high demand. To ensure pesticide availability for the following season, they purchase pesticides in bulk to meet the needs of multiple cropping seasons. This behavior is also associated with the intended use of pesticides, which is emphasized for prevention purposes (See Table 6).

Based on further interviews with the respondents, the primary reason for purchasing pesticides is influenced by popular pesticide brands among farmers. The main factors influencing pesticide purchase are quality and price. However, the horticultural and oil palm farmers regard price as a moderately significant factor in pesticide purchase, whereas rice and maize farmers are more price-sensitive. Quality is the primary consideration for selecting pesticides, aligning with the most frequently recognized purpose of pesticide use among farmers, which is prevention. This implies that, even in the absence of present infestations, there is a general feeling of unease over possible pest and disease dangers to crops. Farmers' experience and knowledge regarding pesticide efficacy contribute significantly, as those with positive experiences or a good understanding of pesticides are more likely to utilize them. Additionally, recommendations from agricultural extension workers or suppliers can influence farmers' decisions when selecting appropriate products. The pricing and accessibility of pesticides in the market are other important considerations; farmers are more likely to buy them if they are reasonably priced and easily available. Modern agricultural practices and the latest technologies also contribute to these decisions, as farmers who adopt new technologies often use pesticides to enhance crop yields. Economic pressures may drive farmers to purchase pesticides to mitigate the risk of crop failure. Finally, community habits and trends can impact this behavior, as farmers often follow practices prevalent among their peers.