Pesticide Use under Public Good Agricultural Practices Standard: A Comparative Study in Thailand

Abstract

The local implementation of Thailand’s public good agricultural practices (GAP) certification standard was investigated by examining its effectiveness in raising farmers’ awareness, lowering pesticide use, and thus improving food safety and quality assurance. A total of 98 certified and 100 uncertified farms were surveyed in two Thai provinces, Chanthaburi and Nakhon Si Thammarat, respectively. Considering durian (Durio zibethinus Murr.) cultivation, the economic size of farming in the former region is relatively larger and production more market-oriented, while in the latter region, the farming scale is smaller. It was observed that knowledge and understanding of GAP is much higher among the certified as compared to the uncertified farmers, although the effectiveness of GAP in lowering pesticide use remains unconfirmed. The comparison between the two regions also suggests that farmers’ pesticide use could be affected by other contextual factors to a greater extent than the GAP. Results from a multivariate linear regression analysis highlighted the critical importance of training on Q-GAP and pesticide use in reducing pesticide use among certified farmers. This suggests that GAP applicant farmers should be encouraged to participate in training on those subjects as an integral part of their Q-GAP application or renewal procedure.

1. Introduction

Since the early 2000s, countries of the Association of Southeast Asian Nations (ASEAN) have implemented national good agricultural practices (GAP) standards to improve the safety and quality of agricultural produce. All of them are public food safety standards run by the responsible state agencies, except for Thai GAP, a national private GAP standard managed by the Thai Chamber of Commerce and Kasetsart University. The introduction of these national GAP standards is largely a response to the rapidly increasing levels of agricultural pesticide use in the region, matched by the growing concerns of foreign and domestic consumers about food safety [1]. The emergence of national public GAP standards in the ASEAN region may be considered a counter movement to the global private trend toward food safety standardization of the global value chain [2]. National GAP standards created in other regions of the world are mostly private (e.g., ChinaGAP, ChileGAP, CanadaGAP), with some of them seeking harmonization with GlobalGAP, the prominent GAP standard of west European origin with a significant influence on the global fresh produce market [3].

Originally established in 1997 by a consortium of European retailers, as of May 2019, GlobalGAP had 200,279 certification-holding growers in more than 135 countries [4,5]. As a pioneering and still the most influential GAP standard in the world, GlobalGAP has attracted the bulk of academic GAP studies. Earlier studies reported GlobalGAP’s tendency to give advantage to larger farmers over smaller farmers due to the costly investments it requires from the applicant to meet its relatively stringent standards, e.g., [6,7,8]. Since the early 2010s, many GlobalGAP studies have sought to identify the socioeconomic attributes or characteristics of GlobalGAP-certified farmers via comparison with the situation of uncertified farmers, thereby demonstrating the socioeconomic advantage of GlobalGAP-certified farmers over uncertified farmers [9,10,11,12,13,14,15,16,17]. One study highlighted significant disparities that are also present among GlobalGAP-certified farmers [18].

In the case of the public GAP standards of ASEAN nations, governments largely bear the costs of audit and certification. They generally set lower certification standards than GlobalGAP and other private GAP standards, which may pose a greater potential opportunity for the inclusion of small-scale farmers in mainstream markets [2]. While sharing the goals of food safety assurance, public GAP standards of ASEAN countries involve varying levels of grower certification and differing models of policy design and implementation. Thailand’s Q-GAP has the largest number of certified farmers (146,000 in 2021), accounting for the majority of certified farmers in the ASEAN region. The number of farmers maintaining a certified status in the other ASEAN countries is much smaller (

Table A1. Adoption of national public GAP standards by farmers in ASEAN nations.

Country	Program Name	Year of Formulation	Number of Farmers with Certified Status (Year)	Scheme Owner
Malaysia	MyGAP	2002	664 (2016)	Department of Agriculture
Thailand	Q-GAP	2003	>146,000 (2021)	National Bureau of Agricultural Commodity and Food Standards
Singapore	SingaporeGAP-VF	2004	8 (2020)	Agri-Food and Veterinary Authority
Indonesia	IndoGAP	2004	13,666 (n/a)	National Food Safety Competent Authority
The Philippines	PhilGAP	2005	82 (2016)	Department of Agriculture
Vietnam	VietGAP	2008	1574 (2014)	Ministry of Agriculture and Rural Development
Myanmar	MyanmarGAP	2009	0 (2015)	Department of Agriculture
Laos	LaoGAP	2010	300 (2016)	Department of Agriculture
Cambodia	CamGAP	2010	0 (2017)	Department of Sanitary and Phytosanitary Plants Protection
Brunei	BruneiGAP	2014	4 (2019)	Agriculture and Agri-Food Department

Source: MyGAP: obtained via personal contact in the Department of Agriculture, Malaysia, in 2017, Q-GAP: [33], SingaporeGAP-VF: [34], PhilGAP, VietGAP, MyanmarGAP, and LaoGAP: [35], IndoGAP: [35,36], LaoGAP: [37], and BruneiGAP: [38]. n/a = not available.

in Appendix A). The significantly greater number of certified farmers in Thailand than the rest of the ASEAN group would be partly attributed to its large proportion of fresh fruit and vegetable (FFV) farmers. However, considering that countries such as the Philippines, Indonesia, and Vietnam also have a large proportion of FFV farmers, differences in the policy design and implementation of public GAP approaches would account much more for this discrepancy. For instance, Malaysia’s MyGAP is considered much more demanding in terms of compliance than Thailand’s Q-GAP [19]. The Philippines’ PhilGAP is much more focused on corporate farmers and rarely provides certification to household-managed small farmers [20]. Most importantly, Thailand appears to be more determined to achieve a large number of national public GAP certifications in comparison to other ASEAN nations by investing more resources [21]. Hence, Q-GAP may be seen to ensure a higher level of fairness and equity by involving small-scale farmers.

However, Q-GAP has been questioned about its role as a certification standard. The level of control over food safety risks was questioned, as the levels of farmers’ compliance and regulatory enforcement were low [1,22]. Consequently, growers’ credibility and social recognition of Q-GAP certification remain limited. Similarly, as distributors may not fully trust the level of food safety guaranteed by Q-GAP certification, the expected marketing leverage through certification and the associated traceability may not be fully functioning, especially in the export market, where exporters are considerably sensitive to overseas consumers’ concerns about food safety and environmental sustainability.

Generally, there could be a proportional relationship between the level of required strictness in compliance with a standard and the required costs of enforcement. Stricter standards would require larger enforcement costs by farmers because of more complex operating changes to prevent pollution. Conversely, more relaxed standards could be easily achieved with fewer enforcement resources [23]. Hence, socioeconomically less powerful farmers in developing countries are excluded from the mainstream market via the deployment of strict private food safety standards, such as GlobalGAP. In contrast, with the deployment of public standards, such as Q-GAP, farmers are more easily certified due to loose compliance criteria. Therefore, the Q-GAP certification standard is placed in the context of the following general concerns: how strict and satisfactory is the level of Q-GAP compliance required by the enforcing agency and is it actually achieved by the growers? Should the level of farmer participation be sacrificed by raising the required degree of compliance and enforcement? Is the current level of the standard appropriate to maintain a large social participation?

This study addresses Thailand’s Q-GAP standard, with a focus on the fresh fruit sector, by examining the effectiveness of its standard in improving food safety and quality assurance. Two comparisons were performed: first, the perceptions of certified and uncertified farmers regarding their awareness about food safety and quality assurance, as well as their appreciation of the public GAP program; second, the level of pesticide use by certified and uncertified farmers, also considering their experiences of training, auditing, and record-keeping.

Three versions of the Q-GAP code of practice have been issued since its establishment: TAS 9001–2004 was introduced in 2004, TAS 9001–2009 in 2009, and the current version, TAS 9001–2013, in 2013. Several policy evaluation studies of Q-GAP implementation have been conducted concerning the standard enforcement and producers’ compliance, but most of them are based on pre-2013 versions (see Appendix A on the certification details). Comparing 45 Q-GAP-certified and 245 uncertified farmers for a total of nine vegetable and fruit crops in a watershed of Chiang Mai Province, northern Thailand, Schreinemachers et al. (2012) revealed that there were no statistically significant differences between the two types of farmers, based on the amount of active chemical ingredients involved in the pesticides used, the methods of pest control adopted, or the handling of pesticides, such as safety protection undertaken [1]. Amekawa (2013) found that 34 of the 64 Q-GAP-certified pomelo growers from two communities in Chaiyaphum Province, northeast Thailand, did not understand the purpose of GAP. Additionally, most of those who noted a reduction in their pesticide use, taking place around the time of their certification, ascribed the reduction to the growth stage of pomelo instead of the effect of Q-GAP training [22]. Srisopaporn et al. (2014) compared the agrochemical use by 41 rice farmers who were continued Q-GAP adopters with that by 66 adopt-then-disadopters and 70 never-adopters in Ayutthaya Province, central Thailand. The continued adopters spent significantly less on total fertilizer costs than adopt-then-disadopters and never-adopters. They also applied pesticides with markedly lower frequency, except for herbicide application, which showed no significant differences. The authors attributed the observed differences from the findings of Schreinmachers et al. (2012) [1] and Amekawa (2013) [22] to the easier technical possibility in rice cultivation to reduce pesticides compared to the case of FFV production [24]. The findings of these studies are more or less outdated because the bodies of their research were all conducted before 2013. As the local Department of Agriculture (DoA) officers claim, the level of Q-GAP policy implementation (in terms of both enforcement and compliance) could have been upgraded by introducing the TAS 9001–2013 code of conduct. The study by Amekawa et al. (2021) represents an exception: field research was conducted from June 2019 to January 2020. By comparing 41 Q-GAP-certified and 90 uncertified cabbage farmers in Chiang Mai Province, northern Thailand, the research found that the certified farmers’ use of insecticides, fungicides, and herbicides, in terms of adoption rate, frequency of pesticide application, and aggregate pyrethroid residues detected, was significantly lower than the uncertified farmers. However, the targeted crop, cabbage, is a vegetable mostly aimed at domestic sales; therefore, it does not promote significant export, and there would be little or no push to intensify pesticide use to meet the aesthetic demand of the overseas market [21].

Four reasons motivated this study.

First, this work considered a study case applied to a fruit crop, namely durian (Durio zibethinus Murr.), involving large export sales.

Second, this study explored a complex and multidimensional domain, where conflicting opinions among the respondent farmers with respect to perceptions, behavior, practices, and outcomes due to GAP adoption/non-adoption are accounted for. Their training and audit experiences, conventional or innovative pest control systems applied, and the economic and environmental outcomes are also taken into consideration. This approach is different from the ASEAN public GAP studies that focus on identifying more structural factors (e.g., income, labor, and land size) affecting the adoption of a GAP standard [24,25,26,27,28,29,30,31].

Third, while most case studies of ASEAN national public GAP standards have been conducted in a single geographic area, this study involved two geographical areas showing dissimilar conditions, one region showing relatively more market-oriented farming conditions while the other relatively less market-oriented, specifically concerning the economy of size. Amekawa’s (2013) study is an exception, in that it comparatively examined the cases of both export-oriented and domestic-sales-focused pomelo farmers from two different districts in the Chaiyaphum Province, northeastern Thailand [22]. However, that study compared certified farmers between two districts in the same province. In contrast, our study compared the cases of certified and uncertified farmers across two geographically distant provinces.

Fourth, this study examined pesticide use by farmers regarding the quantity of active chemical ingredients involved. Further, it conducted a multivariate linear regression analysis to identify the main factors affecting the quantity of insecticides, fungicides, and herbicides used by growers. As far as the authors know, no such analysis has been conducted in previous GAP studies.

2. Materials and Methods

2.1. Sampling Procedures

A questionnaire-based survey was conducted from July to September 2016, involving farmers engaged in durian cultivation. Research sites were selected from two provinces in Thailand: Chanthaburi (CH) and Nakhon Si Thammarat (NST), respectively (Figure 1). The former is much more export-oriented than the latter. CH was selected because it represents the most prosperous durian production area in Thailand. In contrast, durian production in NST is largely sold through the internal market, and its export is limited. In the year of this research, CH had the largest area (27,375 ha) and amount (187,790 tons) of durian production by province in Thailand, with a yield of 6860 kg/ha. NST had the fifth largest area (6040 ha) and the fourth largest amount (30,298 tons) of durian production among the Thai provinces, with a yield of 5016 kg/ha [32]. In both provinces, the intermediaries are the main buyers of durian production from farmers; they sort and distribute the purchased product, directing it to export or domestic markets. China is the major destination of durian exports from both provinces, but detailed information on the extent of exports is not available.

A total of 3363 certified farmers in CH and 347 certified farmers in NST made up the sample universe. These farms are located in 10 districts throughout CH and 15 districts throughout NST. Three districts were selected in CH, namely Klung, Thamai, and Muang, with 941, 561, and 86 certified farmers, respectively. Four districts were selected in NST, namely Thasala, Changklang, Lansaka, and Chawang, with 183, 36, 25, and 24 certified farmers, respectively. These districts contained relatively high proportions of certified farmers in each province. Concerning CH, we subsequently selected five subdistricts (Mabpi, Tapon, Sueng, Wangsubpasod, and Troknong) from Klung district, two subdistricts (Khaobaisri and Thungbenja) from Thamai district, and Plubpla subdistrict from Muang district. Concerning NST, we chose one subdistrict for each of the four districts, namely Talingchan (in Thasala), Changklang (in Changklang), Khunthale (in Lansaka), and La-ai (in Chawang). Again, as before, these subdistricts were selected as they have a relatively high concentration of certified farmers.

In each province, we aimed to interview 50 certified and uncertified durian farmers each, randomly sampled. Unfortunately, the responses from two certified farmers were found invalid and therefore removed. The final analysis consisted of 48 interviews from CH and 50 from NST. The same number of both certified and uncertified farmers was selected in each considered village. A typical farm household interviewed involved a middle-aged couple of durian growers. Upon our request, the most experienced person in the family, the one responsible for growing durian, was interviewed. To ensure informed consent, every respondent was briefed about the purpose and content of the study before the interview. An information form about the research, main researchers, and their affiliated institutions was also given to the interviewees. Each interview lasted about 30–50 min, and it was based on a prepared questionnaire.

The questionnaire topics included the following: (1) basic farm profile, (2) economic conditions of durian farming, (3) growers’ perceptions about Q-GAP and pesticide use, (4) audit experience (certified farmers only), (5) training experience, (6) pesticide use and management, and (7) pesticide safety and handling. Questions related to farmers’ motives, benefits, and reasons associated with their participation or non-participation in Q-GAP were answered based on multiple predefined choices. For pesticide use, farmers were asked about the annual amount of pesticides applied considering insecticides, fungicides, and herbicides. According to their chemical composition, the total annual amount of active chemical ingredients was estimated in terms of milliliters per hectare. The sample included responses of those who did not use any pesticide as nil. Incomplete or uncertain answers were removed from the analysis. Questions related to integrated pest management (IPM) were also considered in the questionnaire.

At times, a few farmers did not understand the topic of question or have a precise memory about their past conduct. Those farmers were removed from the analysis.

2.2. Statistical Data Processing

Continuous response variables were statistically processed by applying a two-way ANOVA to compare the performance of the certified and uncertified farmers within each region and to check each single effect, both for farmer type and region. Nominal response variables (i.e., Yes = 1; No = 0) were statistically processed according to an R*C contingency table (province and farmer type, respectively) and a chi square test for independence. Regarding the items whose data were gathered based on multiple-choice answers, the total sample of the answers are expressed as “TC: total number of counts.” Since some farmers gave more than one answer, the total number of the samples could be more than that of the respondents.

A multivariate linear regression procedure was applied to identify significant factors influencing the quantity of pesticides used by farmers. In total, 12 regressions were conducted, according to the following factorial combination: two farmer types (certified and uncertified), three pesticide types (insecticides, fungicides, and herbicides), and two provinces (Chanthaburi and Nakhon Si Thammarat). The multiple linear regression (MLR) model is expressed as follows:

Y_i = β₀ + β₁X_i,1 + ⋯ + β_nX_i,n + ε_i

where Y_i is the amount of pesticides (mL/ha of active chemical ingredients), β₀, β₁, …, β_n are the coefficients, and ε_i are independent normally distributed random variables (residual errors) with zero mean and constant variance σ².

Variables related to farmers’ socioeconomic conditions, their perceptions of pesticide use and Q-GAP, and their experience of training and record-keeping were included as explanatory variables for the amount of pesticide used per hectare by the respondent farmers. These variables were previously examined in the existing literature [1,19,21] as important considerations for producers’ pesticide use performance under a GAP.

As the amount of pesticide use required by non-users is included as nil in the regression analysis, the number of training days of farmers who had not participated in pesticide use training or Q-GAP training is also included as nil. Meanwhile, information on the number of training days for IPM and organic fertilizer was not obtained at the data collection stage. Hence, these variables were included in the regression as nominal variables related to the presence or absence of training experience. Likewise, other variables were either nominal or continuous variables, with their unit of measurement indicated in the presentation of the results.

At the end of the model building, it was confirmed that all explanatory variables included were normally distributed and that the variables were not significantly correlated between them. All continuous explanatory variables were standardized.

3. Results

3.1. Socioeconomic Profile of the Farmers Surveyed

In CH, the certified farmers show a significant gender difference from the uncertified farmers (p < 0.01), with males making up more than three-quarters of the respondents (

Table 1. Socioeconomic profile of the respondent farmers with respect to the two provinces (CH vs. NST) and the two farm types (certified vs. uncertified) according to the results of a two-way ANOVA.

Variable Description	CH (N = 98)	NST (N = 100)	Prob	CERT (N = 98)	UNCERT (N = 100)	Prob	CH-CERT (N = 48)	CH-UNCERT (N = 50)	Prob	NST-CERT (N = 50)	NST-UNCERT (N = 50)	Prob
Gender: Man (%)	60.2	54.0	n.s.	67.3	47.0	***	77.1	44.0	***	58.0	50.0	n.s.
Age (years)	51.2 (9.2)	54.6 (11.0)	**	55.1 (10.9)	50.8 (9.1)	***	52.1 (9.7)	50.4 (8.7)	n.s.	57.9 (11.4)	51.2 (9.5)	***
Education (years)	10.1 (4.2)	9.1 (4.1)	*	10.0 (4.2)	9.2 (4.1)	n.s.	10.5 (4.2)	9.7 (4.2)	n.s.	9.4 (4.3)	8.7 (4.0)	n.s.
Total farm size (ha)	3.3 (4.5)	1.7 (1.6)	***	3.1 (4.3)	1.9 (2.2)	***	4.3 (5.7)	2.4 (2.6)	**	2.0 (1.7)	1.4 (1.6)	*
Durian farmland size (ha)	2.5 (3.4)	1.3 (1.1)	***	2.3 (3.2)	1.5 (1.8)	**	3.2 (4.3)	1.9 (2.3)	*	1.5 (1.2)	1.1 (1.1)	n.s.
Number of durian trees per ha (N/ha)	133.4 (66.5)	125.3 (80.3)	n.s.	121.8 (64.1)	136.6 (81.6)	n.s.	125.3 (49.8)	141.1 (79.0)	n.s.	118.5 (75.8)	132.2 (84.7 )	n.s.
Annual durian yield (kg/ha)	8254 (7009)	5319 (5371)	***	7396 (6163)	6128 (6597)	n.s.	9008 (6646)	7468 (7377)	n.s.	5845 (5278)	4788 (5464)	n.s.
Total annual durian sales (1000 Thai Baht)	1607 (3892)	641 (1590)	**	1643 (4039)	605 (1144)	**	2423 (5265)	823 (1453)	**	895 (2131)	388 (659)	n.s.
Annual durian sales per ha (1000 Thai Baht/ha)	592.5 (651.6)	461.0 (598.1)	n.s.	662.9 (747.1)	391.9 (445.8)	***	747.7 (809.2)	443.4 (407.7)	**	581.5 (680.5)	340.5 (479.6)	**

CH: Chanthaburi, NST: Nakhon Si Thammarat; CERT: Certified, UNCERT: Uncertified. *** p < 0.01, ** p < 0.05, * p < 0.10; n.s. = not significant. Standard deviation in parentheses.

). In NST, the certified farmers are significantly older than the uncertified farmers (p < 0.01). In both provinces, the certified farmers have a significantly larger farm landholding. In CH, the durian farmland of the certified farmers is significantly larger in comparison to that of the uncertified farmers (p < 0.10), but no significant difference is identified for this variable in NST. In CH, the certified farmers have a significantly higher total annual durian sales in comparison to their uncertified counterparts (p < 0.05), whereas there is no significant difference in NST. In both provinces, the certified farmers have a significant advantage in annual durian sales per ha. In either province, no significant difference between the two types of farmers is observed with respect to educational levels, the number of durian trees per ha, or annual durian yield.

The durian farmland size in CH is approximately twice that in NST (p < 0.01), indicating the scale advantage of farmers in CH. For both types of farmers, the annual durian yield in CH is significantly higher in comparison to that in NST (p < 0.05) (

Table A3. Regional difference in the respondents’ durian production by farmer type.

Variable Description	CERT (N = 98)		UNCERT (N = 100)
	Difference (CH Minus NST)	Prob	Difference (CH Minus NST)	Prob
Size of durian farmland (ha)	1.7	**	0.8	**
Number of durian trees (ha)	6.8	n.s.	8.9	n.s.
Annual durian yield (kg/ha)	3157	**	2680	**
Total annual durian sales (1000 Thai Baht)	1528	*	435	*
Annual durian sales per ha (Thai Baht/ha)	166.2	n.s.	102.9	*

CH: Chanthaburi, NST: Nakhon Si Thammarat; CERT: Certified, UNCERT: Uncertified. ** p < 0.05, * p < 0.10; n.s. = not significant at 0.1.

in Appendix A). Accordingly, for both types of farmers, the total annual durian sales in CH are significantly larger in comparison to those in NST (p < 0.10). Without complete data on export sales of many studied farmers due to their lack of knowledge regarding how much of their sales go for export, we cannot characterize the certified durian farmers in CH as more export-oriented in comparison to those in NST. It would be safer to simply refer to the durian farmers in CH as relatively more market-oriented in comparison to those in NST, regarding the economy of size.

3.2. Farmers’ Adoption of Q-GAP Standard

The certified farmers in CH were found to be motivated to adopt Q-GAP standard primarily because it provides “guaranteed access to export market/more income through Q-GAP certification” (35.5%) (

Table 2. Certified farmers’ motivation in applying for Q-GAP.

Variable Description (1 = Yes)	Certified Farmers (%)
	CH (TC = 51)	NST (TC = 53)	Prob > Chi-sq
Guaranteed access to export market/more income	35.3	9.4	***
Higher farm-gate price for durian sales	17.6	37.7	**
Product safety/consumer health	15.7	22.6	n.s.
Promised higher productivity	9.8	0.0	**
Reduction in pesticide use	9.8	0.0	**
Recommended by the relevant agency/individual	5.9	15.1	n.s.
Others	5.9	15.1	n.s.

CH: Chanthaburi, NST: Nakhon Si Thammarat. TC signifies the total number of counts. *** p < 0.01, ** p < 0.05; n.s. = not significant.

). Combined with the second (“higher farm-gate price for durian sales” accounting for 17.6%) and fourth motives (“higher productivity promised by the relevant agency” accounting for 9.8%), economic incentives account for 62.7%. In NST, where the economy of size for durian farming is smaller than CH, “higher farm-gate price for durian sales” accounts for the largest proportion (37.7%). Together with “guaranteed access to export market/more income” accounting for 9.4%, economic motives account for 47.2%.

The certified farmers in CH view the utmost benefit of participation in Q-GAP as “guaranteed access to export market/more income” (20.3%) (

Table 3. Certified farmers’ views about the benefits of Q-GAP.

Variable Description (1 = Yes)	Certified Farmers (%)
	CH (TC = 59)	NST (TC = 66)	Prob > Chi-sq
Guaranteed access to export market/more income	20.3	13.6	n.s.
Product safety/consumer health	15.3	6.1	*
Improved product quality	11.9	15.2	n.s.
Improved pest/pesticide management	8.5	12.1	n.s.
Increased confidence as a certified grower	8.5	6.1	n.s.
Higher farm-gate price for durian sales	5.1	33.3	***
No benefits	18.6	9.1	n.s.
Others	11.9	4.5	n.s.

CH: Chanthaburi, NST: Nakhon Si Thammarat. TC signifies the total number of counts. *** p < 0.01, * p < 0.10; n.s. = not significant.

), although it falls short of the original motive by 15%. Coupled with “higher farm-gate price for durian sales” accounting for 5.1%, the perceived economic benefits account for only 25.4%—a significant gap from the observed economic incentive of 62.7%. Meanwhile, 18.6% of the farmers claimed they had gained no benefits from their participation in the Q-GAP program. In NST, “higher farm-gate price for durian sales” accounts for the largest proportion (33.3%). Together with “guaranteed access to export market/more income” (13.6%), economic benefits account for 46.9%, which is close to their economic motivations to apply for the certification program (47.2%).

Regarding the reasons for the uncertified farmers in the two provinces not applying for Q-GAP, in both provinces, the reason with the largest share is related to the fact that they did not know much about Q-GAP, if anything at all (

Table 4. Uncertified farmers’ reasons for not having applied for Q-GAP.

Variable Description (1 = Yes)	Certified Farmers (%)
	CH (TC = 54)	NST (TC = 51)	Prob > Chi-sq
Did not know much about Q-GAP	42.6	78.4	***
Q-GAP is not useful/offers no benefits	25.9	2.0	***
Q-GAP requires too many things to do	11.1	7.8	n.s.
Record-keeping is cumbersome	5.6	0.0	*
DoA audit is too loose	5.6	0.0	*
No time available to spare for Q-GAP	0.0	9.8	**
Others	9.3	2.0	n.s.

CH: Chanthaburi, NST: Nakhon Si Thammarat. TC signifies the number of counts. *** p < 0.01, ** p < 0.05, * p < 0.10; n.s. = not significant.

). However, its share is significantly different between the two areas, with CH accounting for 42.6% and NST accounting for as much as 78.4% (p < 0.01). In CH, the second reason for not having applied for Q-GAP, which accounts for as much as 25.9%, is that the farmers who pointed to this knew about Q-GAP but did not consider it useful for their needs as durian farmers. However, this accounted for only 2% in NST, with there being a significant difference (p < 0.01).

3.3. Farmers’ Perceptions of GAP Policy and Pesticide Use

The certified farmers were checked whether they could correctly relate the goal of the policy to food safety assurance. It was found that 89.6% of the certified farmers in CH and 88.0% of the certified farmers in NST evidenced their understanding of the policy goal, with there being significant gaps from the uncertified farmers in both areas (p < 0.01) (

Table 5. Respondent farmers’ perceptions of GAP policy and pesticide use.

Variable Description (1 = Yes)	CH (N = 98) (%)	NST (N = 100) (%)	Prob	CERT (N = 98) (%)	UNCERT (N = 100) (%)	Prob	CH-CERT (N = 48) (%)	CH-UNCERT (N = 50) (%)	Prob	NST-CERT (N = 50) (%)	NST-UNCERT (N = 50) (%)	Prob
Can relate the goal of the Q-GAP policy to food safety	64.3	55.0	n.s.	88.8	31.0	***	89.6	40.0	***	88.0	22.0	***
2. Know of integrated pest management (IPM) and can explain what it is	45.9	23.0	***	39.8	29.0	n.s.	54.2	38.0	n.s.	26.0	20.0	n.s.
3. Think that Q-GAP-certified farmers are economically more advantageous than uncertified farmers due to certification	31.6	75 (n = 96)	***	59.2	46.9 (n = 96)	*	35.4	28.0	n.s.	82.0	67.4 (n = 46)	n.s.
4. Think that pesticides are not very harmful to the health of users when appropriately managed	40.8	56.0	**	46.9	50.0	n.s.	50.0	32.0	*	44.0	68.0	**
5. Think that pesticides are not very harmful to the health of consumers when appropriately managed	60.2	90.0	***	74.5	76.0	n.s.	60.4	60.0	n.s.	88.0	92.0	n.s.
6. Think that pesticides are not very harmful to the environment when appropriately managed	36.7	71.0	***	43.9	64.0	***	33.3	40.0	n.s.	54.0	88.0	***
7. Think that sufficient assistance has been received from local government agencies to obtain good agricultural technologies and practices	67.3	78.0	*	78.6	67.0	*	72.9	62.0	n.s.	84.0	72.0	n.s.

CH: Chanthaburi, NST: Nakhon Si Thammarat; CERT: Certified, UNCERT: Uncertified. *** p < 0.01, ** p < 0.05, * p < 0.10; n.s. = not significant at 0.1. The number of samples is indicated as n in parentheses for items that fall short of the complete sample N.

). The farmers were asked whether they knew the concept of IPM and could explain it concisely. In this regard, no significant difference was found between the two types of farmers in either province.

The farmers surveyed were asked whether they think the Q-GAP certificate would be economically profitable. No significant difference between the two types of farmers was identified in either area.

The farmers were also asked for their views about the possible adverse impact of pesticide use when it is managed appropriately in relation to the following three classes of affected subject: users’ health, consumers’ health, and the environment. In CH, there is a significantly higher proportion of certified farmers, in comparison to their uncertified counterparts, who recognize that pesticides would not be significantly harmful to the health of users when appropriately used (p < 0.10). In NST, there is a significantly higher proportion of uncertified farmers who subscribe to the view that pesticides would not be very harmful to the users’ health (p < 0.05) and the environment (p < 0.01) when they are appropriately used.

Further, farmers were questioned whether they thought they had received sufficient support from local governments for their access to available agricultural technologies and practices. No significant differences were found in either region.

3.4. Farmers’ Training Experiences

The proportion of the certified farmers in NST who have experience of receiving training from the government on pesticide use is significantly higher in comparison to their uncertified counterparts (p < 0.01) (

Table 6. Respondent farmers’ public training experiences related to GAP and the use of pesticides and organic fertilizer.

Variable Description	CH (N = 98)	NST (N = 100)	Prob	CERT (N = 98)	UNCERT (N = 100)	Prob	CH-CERT (N = 48)	CH-UNCERT (N = 50)	Prob	NST-CERT (N = 50)	NST-UNCERT (N = 50)	Prob
Ever received government training on pesticide use (1 = yes) (%)	67.3	78.0	*	83.7	62.0	***	70.8	64.0	n.s.	96.0	60.0	***
2. Number of days taken for participation in government training on agricultural pesticides	1.45 (1.76) (n = 66)	4.05 (4.64) (n = 78)	***	2.16 (2.69) (n = 82)	3.79 (4.82) (n = 62)	***	1.21 (0.41) (n = 34)	1.72 (2.39) (n = 32)	n.s.	2.83 (3.35) (n = 48)	6.00 (5.71) (n = 30)	*
3. Ever received government training on Q-GAP (1 = yes) (%)	45.9	43.0	n.s.	63.3	26.0	***	62.5	30.0	***	62.0	22.0	***
4. Number of days taken for participation in government training on Q-GAP	1.31 (0.51) (n = 45)	1.54 (0.94) (n = 41)	n.s.	1.53 (0.85) (n = 60)	1.15 (0.37) (n = 26)	**	1.33 (0.55) (n = 30)	1.27 (0.46) (n = 15)	n.s.	1.69 (1.06) (n = 30)	1.00 (0.00) (n = 11)	**
5. Ever received government training on IPM (1 = yes) (%)	34.7	18.0	***	27.6	25.0	n.s.	37.5	32.0	n.s.	18.0	18.0	n.s.
6. Ever received government training on the use of organic fertilizer (1 = yes) (%)	55.1	36.0	**	54.1	37.0	**	58.3	52.0	n.s.	50.0	22.0	***

CH: Chanthaburi, NST: Nakhon Si Thammarat; CERT: Certified, UNCERT: Uncertified. *** p < 0.01, ** p < 0.05, * p < 0.10; n.s. = not significant at 0.1. Standard deviation in parentheses. The number of samples is indicated as n in parentheses for items that fall short of the complete sample N.

). Those who had such experience were asked for the number of days for which they had participated in the training. The uncertified farmers in NST, with six of them joining a 15-day program, had significantly more days for pesticide use training in comparison to their certified counterparts (p < 0.10).

In both CH and NST, there is a significantly higher proportion of certified farmers who had Q-GAP training in comparison to their uncertified counterparts (p < 0.01). The Q-GAP-certified farmers in NST who ever joined Q-GAP training took significantly more days for the training than uncertified farmers (p < 0.05). Regarding the farmers’ experience in receiving government training on IPM, no significant differences were detected in either province. Finally, there was a significantly higher proportion of certified farmers in NST who had government training on the use of organic fertilizer (p < 0.01), although there was no significant difference in CH.

3.5. Certified Farmers’ Experiences of Audit

In both provinces, farmers had, on average, 1.5 instances of an audit out of the maximum of 3 (

Table 7. Audit experiences of Q-GAP-certified farmers.

Variable Description	Certified Farmers
	CH (N = 48)	NST (N = 50)	Prob > Chi-sq
Number of times DoA audit was needed to receive Q-GAP certification	1.50	1.50	n.s.
Received advance notice on the date of the first audit (1 = yes) (%)	52.0	64.6	n.s.
Time taken for the first audit (minutes)	51.5 (n = 36)	34.7 (n = 47)	***
Checked in audit on the record-keeping of agricultural practices (1 = yes) (%)	58.3	28.0	***
Handed durian samples directly to DoA officers for pesticide residue test (1 = yes) (%)	4.0%	31.3%	***

CH: Chanthaburi, NST: Nakhon Si Thammarat. *** p < 0.01; n.s. = not significant at 0.1. The number of samples is indicated as n in parentheses for items that fall short of the complete sample N.

). In GAP certification in general, farmers are not supposed to receive notice pertaining to the date of the auditor’s visit in advance so that they cannot prepare for an audit. However, from the auditors’ viewpoint, visiting farmers without prior reservations makes auditing less efficient, for farmers may not stay at home at the time of the audit visit. The results show that over half of the farmers in both provinces received advance notice. Regarding the time spent on an audit in the first round, the certified farmers in CH took approximately 17 min more than those in NST, with there being a significant difference (p < 0.01). Additionally, 30.3% more farmers in CH compared to those in NST were checked by the auditor on their record-keeping of agricultural practices, with there being a significant difference (p < 0.01). Further, in principle, crop samples should be chosen by the auditors themselves, but a significantly higher proportion of farmers in NST handed their durian produce directly to the DoA auditor as samples for pesticide residue analysis in comparison to those in CH (p < 0.01).

3.6. Pesticide Use

In CH, no significant differences are observed in the adoption of synthetic pesticides between the two farmer types (

Table 8. Pesticide use by respondent farmers.

Variable Description	CH (N = 98)	NST (N = 100)	Prob	CERT (N = 98)	UNCERT (N = 100)	Prob	CH-CERT (N = 48)	CH-UNCERT (N = 50)	Prob	NST-CERT (N = 50)	NST-UNCERT (N = 50)	Prob
Insecticides
Use (1 = yes) (%)	98.0	84.0	***	94.9	87	*	95.8	100.0	n.s.	94.0	74.0	***
Annual amount of active ingredients (mL/ha)	6022 (10,291) (n = 95)	1784 (3752)	***	4079 (7538) (n = 98)	3615 (8355) (n = 97)	n.s.	6233 (9491)	5806 (11,147) (n = 47)	n.s.	2011 (4142)	1557 (3343)	n.s.
Fungicides
Use (1 = yes) (%)	95.9	63.0	***	87.8	71.0	***	93.8	98.0	n.s.	82.0	44.0	***
Annual amount of active ingredients (mL/ha)	2374 (3923) (n = 91)	712 (1325) (n = 99)	***	2416 (3809) (n = 97)	561 (1197) (n = 93)	***	3751 (4865)	836 (1383) (n = 43)	***	1108 (1523) (n = 49)	324 (963)	**
Herbicides
Use (1 = yes) (%)	71.4	60.0	*	61.2	70.0	n.s.	68.8	74.0	n.s.	54.0	66.0	n.s
Annual amount of active ingredients (ml/ha)	3578 (6552) (n = 97)	1413 (3191)	***	2796 (5265) (n = 98)	2166 (5202) (n = 99)	n.s.	3843 (6112)	3319 (7011) (n = 49)	n.s.	1791 (4115)	1036	n.s.

CH: Chanthaburi, NST: Nakhon Si Thammarat; CERT: Certified, UNCERT: Uncertified. *** p < 0.01, ** p < 0.05, * p < 0.10; n.s. = not significant at 0.1. Standard deviation in parentheses. The number of samples is indicated as n in parentheses for the farmers who could provide information.

). Meanwhile, in NST, there is a significantly higher proportion of certified farmers who adopted insecticide (p < 0.01). The same also applies to fungicide. In both provinces, fungicides show a significantly larger evidence of use by the certified over the uncertified farmers in terms of milliliters per ha.

It is worth noting that the certified farmers in CH used approximately 3.1 times more insecticides, 3.4 times more fungicides, and 2.1 times more herbicides per ha in comparison to those in NST. Likewise, the uncertified farmers in CH used about 3.7 times more insecticides, 2.6 times more fungicides, and 3.2 times more herbicides per ha in comparison to those in NST. All of these differences are statistically significant, except for the amount of herbicide use by the uncertified farmers (

Table A4. Regional difference in the respondents’ pesticide use by farmer type.

Variable Description	CERT		UNCERT
	Difference (CH Minus NST)	Prob	Difference (CH Minus NST)	Prob
Insecticides
Annual amount of active ingredients (mL/ha)	4222	***	4249	**
Fungicides
Annual amount of active ingredients (mL/ha)	2643	***	512	**
Herbicides
Annual amount of active ingredients (mL/ha)	2053	**	2283	n.s.

CH: Chanthaburi, NST: Nakhon Si Thammarat; CERT: Certified, UNCERT: Uncertified. *** p < 0.01, ** p < 0.05; n.s. = not significant at 0.1.

in Appendix A).

3.7. Alternative Pest Management

No significant differences are found in either area in the adoption of at least one kind of alternative pest management (

Table 9. Respondent farmers’ adoption of alternative pest management methods by percentage.

Variable Description (1 = Yes)	CH (N = 98) (%)	NST (N = 100) (%)	Prob	CERT (N = 98) (%)	UNCERT (N = 100) (%)	Prob	CH-CERT (N = 48) (%)	CH-UNCERT (N = 50) (%)	Prob	NST-CERT (N = 50) (%)	NST-UNCERT (N = 50) (%)	Prob
Farmers who use at least one alternative pest management	71.4	71.0	n.s.	66.3	75.0	n.s.	68.8	74.0	n.s.	64.0	76.0	n.s.
Adoption of specific method
Herbal insecticides	3.1	1.0	n.s.	3.1	1.0	n.s.	4.2	2.0	n.s.	2.0	0.0	n.s.
EM insecticide	2.0	1.0	n.s.	1.0	2.0	n.s.	0.0	4.0	n.s.	2.0	0.0	n.s.
Naphthalene for insect expulsion	0.0	1.0	n.s.	1.0	0.0	n.s.	0.0	0.0	n.s.	2.0	0.0	n.s.
Biological fungicides	8.2	2.0	**	8.1	2.0	**	12.5	4.0	n.s.	2.0	0.0	n.s.
Cutting weeds using a weed cutter	54.1	66.0	*	51.0	69.0	***	45.8	62.0	n.s.	56.0	76.0	**
Removing weeds by hand	10.2	0.0	***	7.1	3.0	n.s.	14.6	6.0	n.s.	0.0	0.0	n.s.

CH: Chanthaburi, NST: Nakhon Si Thammarat; CERT: Certified, UNCERT: Uncertified. *** p < 0.01, ** p < 0.05, * p < 0.10; n.s. = not significant at 0.1.

). In both areas, the alternative method adopted by most farmers is a mechanical weed cutter. In NST, there is a significantly higher proportion of uncertified farmers who adopted this method in comparison to their certified counterparts (p < 0.05), while there are no significant differences in CH. No other method is used extensively in either area, which indicates durian farmers’ dominant reliance on synthetic pesticides for crop protection.

3.8. Record-Keeping

In CH, there is a significantly higher proportion of certified farmers who have kept records of the use of the particular farming input since their latest certification for all the six items of comparison (p < 0.01) (

Table 10. Respondent farmers’ record-keeping of agrochemical use.

Variable Description (1 = %)	CH (N = 98) (%)	NST (N = 100) (%)	Prob	CERT (N = 98) (%)	UNCERT (N = 100) (%)	Prob	CH-CERT (N = 48) (%)	CH-UNCERT (N = 50) (%)	Prob	NST-CERT (N = 50) (%)	NST-UNCERT (N = 50) (%)	Prob
Insecticide use	37.5 (n = 96)	3.6 (n = 84)	***	35.4 (n = 93)	6.9 (n = 87)	***	65.2 (n = 46)	12.0 (n = 50)	***	6.4 (n = 47)	0.0 (n = 37)	n.s.
Fungicide use	38.3 (n = 94)	1.6 (n = 63)	***	37.2 (n = 86)	7.0 (n = 71)	***	68.9 (n = 45)	10.2 (n = 49)	***	2.4 (n = 41)	0.0 (n = 21)	n.s.
Herbicide use	32.9 (n = 70)	5.0 (n = 60)	***	43.3 (n = 60)	0 (n = 70)	***	69.7 (n = 33)	0.0 (n = 37)	***	11.1 (n = 27)	0.0 (n = 33)	**
Use of other pest management methods	37.5 (n = 64)	0.0 (n = 68)	**	82.7 (n = 81)	44.8 (n = 67)	***	63.6 (n = 33)	9.7 (n = 31)	***	0.0 (n = 32)	0.0 (n = 36)	n.s.
Use of chemical fertilizers	33.3 (n = 96)	5.2 (n = 97)	***	35.1 (n = 94)	4.0 (n = 99)	***	61.0 (n = 46)	8.0 (n = 50)	***	10.4 (n = 48)	0.0 (n = 49)	**
Use of other fertilization methods	26.0 (n = 77)	5.8 (n = 86)	***	30.1 (n = 83)	0 (n = 76)	***	54.1 (n = 37)	0.0 (n = 40)	***	10.9 (n = 46)	0.0 (n = 36)	**

CH: Chanthaburi, NST: Nakhon Si Thammarat; CERT: Certified, UNCERT: Uncertified. *** p < 0.01, ** p < 0.05; n.s. = not significant at 0.1. The number of samples is indicated as n in parentheses for farmers who use the particular production input.

). In NST, a significant difference in favor of the certified farmers is observed with the use of herbicides, chemical fertilizers, and other fertilization methods. The results of the certified farmers show a striking contrast between the two provinces. While about 54–70% of the certified farmers in CH keep a record of the use of the farming inputs studied, only around 0–11% of the certified farmers in NST do so for the same items.

3.9. Factors Affecting the Quantity of Pesticide Use

All the 12 multivariate regressions were statistically significant, and the R² values ranged from 0.530 to 0.736 (

Table 11. Influence of various factors on the quantity of insecticides, fungicides, and herbicides applied (all the variables considered are listed).

Variable Description	Insecticides		Fungicides		Herbicides
	Certified	Uncertified	Certified	Uncertified	Certified	Uncertified
Chanthaburi	(N = 48)	(N = 47)	(N = 48)	(N = 43)	(N = 48)	(N = 49)
Socioeconomic Factors
Gender (male = 1, female = 0)	0.943 (0.663) *	−0.466 (0.775)	−0.145 (0.411)	0.147 (0.097) **	−0.470 (0.545) *	−0.466 (0.775) *
Age (no. of years)	0.006 (0.036)	−0.025 (0.058)	−0.002 (0.022)	0.002 (0.007)	0.015 (0.024)	−0.025 (0.058)
Education (no. of years)	0.069 (0.093)	0.004 (0.112)	−0.048 (0.059)	−0.007 (0.014)	0.104 (0.067) **	0.004 (0.112)
Total farm size (ha)	0.005 (0.072)	0.005 (0.072)	0.009 (0.019)	−0.002 (0.009)	0.023 (0.023)	0.005 (0.072)
Durian farm size (ha)	−0.045 (0.044)	−0.016 (0.082)	−0.013 (0.027)	−0.002 (0.010)	−0.028 (0.031)	−0.016 (0.082)
No. of durian trees/ha	0.046 (0.047)	0.020 (0.032)	0.046 (0.029) **	−0.002 (0.004)	0.010 (0.035)	0.020 (0.032)
Durian yield (kg/ha)	0.243 (0.340)	−0.445 (0.462)	−0.137 (0.210)	0.014 (0.058)	−0.272 (0.265) *	−0.445 (0.462)
Durian sales (THB)/ha	−0.799 (0.484) **	0.610 (0.458) *	−0.036 (0.319)	0.016 (0.057)	−0.205 (0.333)	0.610 (0.458)
Farmers’ perceptions (1 = yes, 0 = no)
Can relate GAP to food safety	−0.747 (0.718)	−0.067 (0.332)	0.232 (0.715)	−0.177 (0.146) **	−0.245 (0.865)	0.507 (1.165)
Know what IPM is	−0.062 (0.163)	0.507 (0.165)	0.137 (0.467)	0.050 (0.166)	−0.504 (0.559)	−0.367 (0.72)
Not harmful to producer health	−0.572 (0.863) ***	0.151 (0.165)	−0.073 (0.529)	0.117 (0.146)	−0.037 (0.649)	0.151 (0.165)
Not harmful to consumer health	−0.633 (0.565) ***	−0.048 (0.921) *	−0.529 (0.343) **	−0.029 (0.115) *	−0.234 (0.453) **	−0.048 (0.221) *
Not harmful to the environment	0.602 (1.438)	0.207 (0.241)	−0.462 (0.870)	−0.037 (0.155)	−0.716 (0.887)	0.207 (1.241)
Received government support	0.993 (0.731) *	0.158 (0.014)	−0.551 (0.447) **	−0.084 (0.127)	−0.648 (0.594) **	−0.158 (1.014) **
Training experience
Training on pesticide use (no. days)	−0.071 (0.474)	−0.056 (0.178)	−0.703 (0.314) ***	−0.002 (0.022)	−0.264 (0.582) *	0.056 (0.178)
Training on Q-GAP (no. days)	−0.465 (0.625) **	−0.724 (0.994) *	−0.071 (0.547) *	−0.177 (0.124) ***	−0.537 (0.406) ***	−0.724 (0.994) *
Training on IPM (1 = yes, no = 0)	−0.455 (0.588)	0.138 (1.556)	−0.866 (0.842)	−0.066 (0.195)	−0.049 (0.478)	0.138 (1.556)
Training on organic fertilizer (1 = yes, no = 0)	0.209 (0.895) **	0.799 (0.835)	−0.491 (0.565)	0.058 (0.094)	−0.652 (0.631) ***	0.799 (0.835)
Record-keeping practices(1 = yes, 0 = no)
Record-keeping on insecticides	0.345 (0.198)	0.572 (1.222)	−0.100 (0.685)	0.057 (0.104)	−0.125 (0.769)	−0.318 (0.142)
Record-keeping on fungicides	−0.425 (0.138)	−0.318 (0.142)	−0.412 (0.747) **	−0.089 (0.143)	0.552 (0.737)	−0.278 (1.104)
Record-keeping on herbicides	0.044 (0.818)	n/a	0.227 (0.469)	n/a	−0.231 (0.506)	n/a
Constant	0.975 (0.684) *	−0.820 (0.391)	0.148 (3.800)	−0.102 (0.799)	0.851 (0.092) *	−0920 (0.391)
R2	0.670	0.594	0.736	0.614	0.710	0.530
Nakhon Si Thammarat	(N = 50)	(N = 50)	(N = 49)	(N = 50)	(N = 50)	(N = 50)
Socioeconomic Factors
Gender (male = 1, female = 0)	−0.032 (0.236)	−0.123 (0.171)	−0.041 (0.094)	−0.067 (0.048) *	0.194 (0.103)	0.025 (0.103)
Age (no. of years)	−0.002 (0.011)	0.001 (0.012)	0.002 (0.004)	0.003 (0.003)	0.006 (0.007)	0.005 (0.007)
Education (no. of years)	0.010 (0.029)	−0.033 (0.025) **	−0.013 (0.012) **	−0.006 (0.006)	0.044 (0.015) **	0.033 (0.013) ***
Total farm size (ha)	−0.0111 (0.022)	0.003 (0.021)	0.002 (0.006)	−0.009 (0.005) ***	0.009 (0.012)	−0.002 (0.010)
Durian farm size (ha)	−0.007 (0.016) **	−0.025 (0.030) *	−0.013 (0.009) **	0.008 (0.007)	0.002 (0.018)	−0.011 (0.018)
No. of durian trees/ha	0.002 (0.009)	0.003 (0.007)	−0.000 (0.005)	0.000 (0.002)	−0.002 (0.007)	−0.003 (0.004)
Durian yield (kg/ha)	0.299 (0.146) ***	0.003 (0.131)	−0.041 (0.075)	−0.036 (0.041)	−0.037 (0.086)	−0.048 (0.086)
Durian sales (THB)/ha	0.015 (0.085)	−0.071 (0.117)	0.015 (0.044)	0.007 (0.033)	0.120 (0.115)	0.034 (0.069)
Farmers’ perceptions (yes = 1, no = 0)
Can relate GAP to food safety	−0.136 (0.317) **	−0.358 (0.254) **	0.117 (0.127)	−0.188 (0.072) ***	−0.674 (0.423)	0.046 (0.150)
Know what IPM is	−0.130 (0.363) ***	−0.878 (0.726) *	−0.298 (0.156) **	0.023 (0.066)	−0.305 (0.496) **	−0.328 (0.430)
Not harmful to producer health	−0.177 (0.207) *	−0.172 (0.213)	−0.037 (0.107)	−0.075 (0.126) *	−0.139 (0.259)	−0.275 (0.126) *
Not harmful to consumer health	−0.430 (0.291) **	−0.225 (0.448) ***	−0.072 (0.153)	−0.130 (0.117)	−0.653 (0.352) **	−0.136 (0.266) *
Not harmful to the environment	−0.027 (0.224)	−0.324 (0.336) **	−0.015 (0.118)	−0.095 (0.091)	−0.334 (0.306)	−0.193 (0.199)
Received government support	−0.326 (0.288) **	0.105 (0.266)	−0.137 (0.123) *	0.057 (0.062)	−0.234 (0.324) *	−0.074 (0.158)
Training experience
Training on pesticide use (no. days)	0.034 (0.037)	−0.001 (0.017)	−0.014 (0.015) *	−0.003 (0.005)	−0.068 (0.875) **	−0.008 (0.010)
Training on Q-GAP (no. days)	−0.238 (0.091) ***	−0.267 (0.226) **	−0.039 (0.036) **	−0.027 (0.205) *	−0.197 (0.334) ***	0.018 (0.157)
Training on IPM (yes = 1, no = 0)	−0.529 (0.455) ***	−0.189 (0.326)	−0.346 (0.182) ***	0.181 (0.204)	−0.604 (0.593)	0.219 (0.368)
Training on organic fertilizer (yes = 1, no = 0)	0.003 (0.224)	−0.220 (0.722) *	−0.034 (0.090)	−0.048 (0.066)	−0.604 (0.593)	0.376 (0.428)
Record-keeping practices (yes = 1, no = 0)
Record-keeping on insecticides	0.325 (0.448)	n/a	0.236 (0.232)	n/a	0.447 (0.612)	n/a
Record-keeping on fungicides	1.446 (1.176)	n/a	−0.153 (0.607)	n/a	0.620 (0.605)	n/a
Record-keeping on herbicides	−0.469 (0.486)	n/a	0.189 (0.251)	n/a	−0.363 (0.663)	n/a
Constant	−0.422 (1.241) ***	0.988 (0.198) *	0.304 (0.640) **	0.722 (0.338)	−0.495 (0.694) *	−0.806 (0.710)
R2	0.702	0.613	0.694	0.603	0.672	0.611

*** p < 0.01, ** p < 0.05, * p < 0.10; the results without an asterisk are not significant. Standard error in the parentheses. n/a = not available due to the lack of farmers’ record-keeping practices. The number of samples is indicated as N in parentheses for those who could provide the information on the amount of the particular pesticide used and are thus included in the analysis.

). All the statistically significant results related to the growers’ training show a negative coefficient, suggesting that the more training the growers have, the less pesticides they would use. This highlights the importance of training in lowering pesticide use. In particular, growers’ training days on Q-GAP show significant results for all the pesticide types of the farmers in the two regions, except for the herbicide use by the uncertified farmers in NST. Training days on pesticide use by the certified farmers also show significant results for fungicides and herbicides in both regions. Of them, the coefficient value of those in CH is relatively high for fungicides (−0.703), indicating the higher impact of pesticide use training on the reduction of fungicides sprayed compared to other variables.

The certified farmers in NST have a significant effect of their knowledge of IPM in lowering the use of each type of pesticide. Their IPM training experience also has a significant effect on reducing the quantity of insecticides and fungicides sprayed. Meanwhile, these variables have no significant effect on the farmers in CH.

The variables in the farmers’ perceptions category also show some significant results. Especially noteworthy is that the variable “not harmful to consumer health” shows significant negative effects on the farmers’ use of all types of pesticides in CH and on the farmers’ use of insecticides and herbicides in NST. This suggests farmers who are more confident to assure their consumers’ health tend to spray less pesticides.

4. Discussion

In both provinces, the certified farmers were found to have an economic advantage. Without a price premium specifically attached to the Q-GAP-certified durian produce, the significant differences in durian sales per ha between the two types of farmers in both regions would be largely due to the effect of Q-GAP certification on farmers’ access to high-value markets. This finding is in line with Amekawa et al. (2017) on durian for MyGAP [19], Krause et al. (2016) on mangoes for Q-GAP [26], and Pongvinyoo et al. (2015) on mangosteen for Q-GAP [27]. Accordingly, the major motivations of the certified farmers to have applied for Q-GAP are related to economic incentives. Meanwhile, “expected product safety/consumer health improvements through the adoption of Q-GAP” was found to play a lesser motivational role in both areas. This could be the case considering that both areas involve durian export sales that solicit growers’ economic incentives. It is also noteworthy that the value of total annual durian sales of the certified farmers in CH is particularly pronounced―approximately 2.7 times higher than their counterparts in NST. This comparatively high marketability of durian produce at the farm household level in CH may help explain the male bias of the interview respondents in the region, who claimed themselves to be the most knowledgeable and responsible person for durian farming in the household.

Regarding the effectiveness in raising awareness, in both areas, nearly 90% of the certified farmers demonstrated an understanding of the basic purpose of the Q-GAP policy. Meanwhile, the uncertified farmers in both areas showed a much lesser understanding, given that many of them did not even know about Q-GAP itself. The uncertified farmers’ lack of knowledge would also be grounded in their significantly limited experience of participation in GAP training. As several extant studies on ASEAN public GAP standard reported [19,28,29,30], GAP training experience is a key factor for the growers’ adoption of Q-GAP. We also notice that overall, farmers in NST tend to hold a more positive view of the possible effects of pesticide use in comparison to those in CH. This may be related to the finding that farmers in NST tend to use significantly less insecticides and fungicides than those in CH.

Regarding the control in the quantity of pesticides sprayed, no significant differences between the two farmer types were found in insecticides or herbicides in either study area. This suggests that compliance with Q-GAP does not necessarily lead to a reduction in pesticide use. Further, the significantly greater amount of fungicides used by the certified farmers in both areas implies that Q-GAP certification may even become a driving force to increase farmers’ fungicide use in the context where IPM is barely adopted. A possibility of this result is that recognizing the certified farmers for their improved access to high-value markets through Q-GAP certification (e.g., more sales, better prices) would make them buy and use more fungicides to protect their durian trees from canker disease (caused by Phytophthora). Another possibility suggested by the local Department of Agriculture and Agricultural Extension (DoAE) staff was that the certified farmers’ trained knowledge of appropriate fungicide usage to treat the affected part of the tree could facilitate the quantitative gap between the certified and uncertified farmers.

It was also found that the certified durian farmers in CH used significantly more pesticides per ha in comparison to those in NST. This result suggests that regional contextual differences (e.g., economic, aesthetic, and agroecological) could dictate farmers’ pesticide use far more saliently than the adoption of GAP practices. According to the local DoAE staff, the larger evidence of pesticide usage by the durian farmers in CH might be related to the possible existence of a more developed hotbed for pests because of the longer history and larger scale of durian farming in the region, coupled with the farmers’ more developed technical knowledge of pesticide usage, as compared to those in NST.

Twelve different multivariate linear regressions were performed with respect to insecticides, fungicides, and herbicides, under both certified and uncertified farming conditions, and in the two considered Thai provinces. For both CH and NST, training days on Q-GAP had a significant effect in reducing the use of all three types of pesticides by the certified farmers. Likewise, training days on pesticide use by the certified farmers in both regions had significant negative effects on fungicide and herbicide use. These results imply the critical importance of training farmers in these subjects in reducing pesticide use through Q-GAP adoption.

Regarding the process control outcomes related to pesticide use, less than three-quarters of the certified farmers in CH keep records for various pesticide and non-pesticide items they actually use. Worse still, only a very limited proportion of the certified pesticide user farmers in NST keep records. The poor record-keeping practices in NST seem to be related to the poor auditing process in the region.

5. Conclusions

The implementation of the Thai public GAP certification standard (Q-GAP) was analyzed in depth through a survey that explored multiple levels of information (technical, economic, social, etc.). The statistical processing allowed for a good understanding of the perceptions and practices of the compared Q-GAP-certified and uncertified durian farmers operating in two contrasting Thai provinces.

A few striking conclusions of this work can be highlighted concerning the understanding of the ASEAN public GAP literature and consequent policy implications:

• The multi-level approach applied in this work seems to be unique and distinctive in comparison with many previous ASEAN public GAP studies retrieved in the literature.
• The local implementation of the Q-GAP standard to a fruit crop was investigated in this study, and the most updated Q-GAP code was assumed as reference (TAS 9001–2013). Although the survey was performed in 2016, the conditions are not drastically different, and the conclusions to be drawn remain almost unchanged: Q-GAP-certified farmers have a definite advantage over uncertified farmers concerning the knowledge and understanding of GAP, while the effectiveness of the public GAP standard in reducing pesticide use looks doubtful. This evaluation may imply the need for the Thai government to consider increasing stringency in compliance related to farmers’ pesticide use, although it may decrease the current level of farmers’ participation in Q-GAP.
• Growers’ pesticide use and performance of other agricultural practices are affected by the farmers’ regional context. In this respect, Q-GAP does seem very effective to reduce the amount of pesticide use if strong exports and, therefore, strong economic revenues are concerned.
• Pesticide use training is critically important in reducing pesticide use by certified farmers, which implies that Q-GAP applicants should be encouraged to participate in this training as part of the GAP application or renewal procedure.

Just a few limitations can be detected; although the sample size was probably limited, we believe the results are correct and substantially in line with those of previous similar studies. Moreover, no direct fruit tests were performed to check the content in pesticide residues, nor were farmers’ record-keeping notebooks referenced. Indeed, only the farmers’ declarations on pesticide use were taken into account.