Cambodian rice farmers’ knowledge, attitudes, and practices (KAPs) regarding insect pest management and pesticide use

ABSTRACT Insect pests are a serious problem for rice farmers in Cambodia, threatening livelihoods and food security. To protect their yields, farmers are dependent on broad-spectrum pesticides, which has led to a pesticide ‘lock-in’ scenario. Our study aimed to better understand farmers’ current pest management practices and the underlying knowledge and attitudes driving these behaviours. We surveyed rice farming households (n = 168) from five different villages in Battambang Province, Cambodia. Survey respondents considered insect pests to be the most significant cause of yield loss. Respondents depended solely on chemical insecticides to manage insect pests. Combinations of agronomic and pesticide application practices contributed to pesticide dependency, including high seeding rate, insecticide timing and application rate. Additionally, 77% of respondents misidentified the beneficial lady beetle, Micraspis discolor, as a pest. Respondents indicated a desire for education and training on pest and beneficial arthropod identification, safe pesticide use, and improved cropping practices. We suggest critical areas for education and training are damage thresholds of local pests, the role of natural enemies, and improved awareness of pesticide toxicity and exposure. Promoting local knowledge sharing to engage and empower farmers to make informed decisions about their own pest management can forge a pathway away from pesticide dependency.


Introduction
Cambodia is heavily dependent on agriculture economically and for food security.Approximately 80% of the country's population is involved in agriculture, with 75% of cultivated land dedicated to rice production (FAPDA, 2014;Seng, 2014).Rice is primarily produced by smallholder farmers who typically run low input rainfed production systems as the primary source of food and income (FAO, 2017).Though the incidence of rice cultivation is high, rice yields in Cambodia are among the lowest in Southeast Asia (Seng, 2014).
Farmers often misuse pesticides by applying them too early in the rice growth stage, at incorrect quantities, and were unaware of the target pests (Jensen, Konradsen, Jørs, Petersen, & Dalsgaard, 2011;Preap & Sareth, 2015).As a result, the rice paddy ecosystem becomes disrupted, leading to a decrease in natural enemies, increased pest populations, secondary pest outbreaks such as the Brown planthopper (BPH) and the evolution of pesticide resistance (Matteson, 2000).
In Cambodia, pest management and pesticide problems have been credited to social (educational, stakeholder attitudes and arrangements) and ecological issues.Social issues include low levels of pest and beneficial arthropod management education, incorrect pest management information from pesticide retailers, the push for agricultural intensification and expansion of rice production, particularly by the Cambodian government (Flor et al., 2020).Management issues include specific agronomic practices such as incorrect seeding rate, over fertilization (Flor, Maat, Hadi, Kumar, & Castilla, 2019) and minimal understanding of target pests.
One over-arching issue for rice farmers is a lack of awareness of, or confidence in, alternative pest control methods (Flor, Chhay, Sorn, Maat, & Hadi, 2018;Matsukawa, Ito, Kawakita, & Tanaka, 2015;Preap & Sareth, 2015).In combination with social and management practices, these mechanisms contribute to a pesticide lock-in (Flor et al., 2020).In short, a lock-in is a repeated preference for one technology or strategy (in this case, the 'technology/ strategy' is pesticides) and uncertainty over alternative 'technologies/strategies' such as integrated pest management (IPM), which ultimately leads to a situation where adopting newer technologies becomes difficult (Arthur, 1989;Flor et al., 2020;Perkins, 2003).
Alternative management approaches often take the form of integrated pest management (IPM), which has been proposed in Cambodia since the 1990s to help tackle pest management issues (Flor et al., 2018).In Cambodia, however, IPM is not commonly adopted for rice, despite significant investment from the Cambodian Government (Flor et al., 2018).In the past, IPM regimes have been attempted for rice using data from other countries such as Thailand and Indonesia (Fazeli, 2017;Pretty & Bharucha, 2015;Thorburn, 2015).However, these have failed mainly due to their lack of contextually relevant practices or technologies, lack of attention to farmers' priorities regarding their agricultural productivity and lack of appreciation of the socio-economic contexts within which smallholder farmers operate, which includes the pesticide lock-in scenario (Parsa et al., 2014;Thornton et al., 2017).
Arguably, a deeper understanding of farmers' current pest management practices and the underlying factors driving these behaviours are needed if alternative pest control methods are to be adopted in the Cambodian context (Parsa et al., 2014).Meijer, Catacutan, Ajayi, Sileshi, & Nieuwenhuis (2015) argued that the potential adopter's knowledge, perceptions, and attitudes towards the innovation play a vital role in the potential adoption and implementation of innovations.However, this has been less studied in the context of agricultural innovations such as IPM (Meijer et al., 2015).
More recently, several studies have investigated pesticide dependency and its causes in rice farming contexts in Cambodia (Castilla et al., 2020;Flor et al., 2019;Flor et al., 2020) and have identified that rice farmers are caught in a pesticide lock-in.These studies concluded that agronomic practices and stakeholder knowledge networks are key factors encouraging the pesticide lock-in.However, further investigation into farmers' knowledge and perceptions is required to get the level of detail needed to improve pest management.
The knowledge, attitudes, and practices (KAPs) study performed by Schreinemachers et al. (2017) concluded that vegetable farmers in Cambodia are heavily reliant on pesticides.While farmers realized the health risks of pesticide use, farmers considered them indispensable for vegetable production.Moreover, farmers who had lower knowledge of beneficial insects and sought information from pesticide sellers were associated with higher pesticide usage (Schreinemachers et al., 2017).
Given the prevalence of pesticide misuse in Cambodia, a better understanding of the knowledge of farmers and the attitudes, and behaviours (practices) driving 'pesticide lock-in' is needed.Hence, our study aims to understand the knowledge, attitudes, and practices (KAPs) of smallholder rice farmers' regarding insect pest management and to identify potential reasons behind chemical dependence.We then use the knowledge generated by our study to suggest entry points for successful IPM adoption.

Conceptual framework
Our study followed a similar method to the one used by Schreinemachers et al. (2017), which applied the concepts of knowledge, attitudes, and practices (KAP) survey to investigate pesticide dependence in yard-long bean and leaf mustard farmers in Southeast Asia.The KAPs methodology investigates the progressive relationship among knowledge, attitudes, and behaviours, where knowledge is the foundation of behaviour change, and belief and attitudes are the driving force of behaviour change (Fan et al., 2018).A KAPs methodology allows an in-depth assessment and analysis of the farmers' knowledge of the problem, their feelings about it and how they currently behave or manage the issue (Schreinemachers et al., 2017).As such, KAPs assessments have been used to help inform educational intervention strategies to encourage behavioural change (Fan et al., 2018;Schreinemachers et al., 2017).Here we adopted a KAPs methodology to investigate driving forces behind farmers' knowledge of and behaviours towards insect pest management and promotion of sustainable methods.
We adopted a mixed-methods approach using a survey methodology to gather qualitative and quantitative information relating to insect pest management knowledge, attitudes, and practices.We also collected background information about the respondents' farms (farm size, rice yields, rice cultivars, water access).
Knowledge refers to the respondents' understanding of common pests and beneficial arthropods present within rice agroecosystems and their general knowledge about integrated pest management and personal protective equipment (PPE).These questions aimed to identify knowledge gaps where information and education efforts could be employed (Gumucio et al., 2011).
To assess knowledge of beneficial and pest insects, respondents were shown 14 coloured pictures of beneficial and pest arthropods common in rice (Supplementary data Table 1).The four beneficial arthropods shown to respondents were: a spider (Araneae), a lady beetle (Micraspis discolor), a damselfly (Coenagrionidae: Odonata), and Asian honeybee (Apis cerana) (Apidae: Hymenoptera).Ten pictures of pest insects were shown to the respondents which included, Brown planthopper (BPH) (Nilaparvata lugens (Stål)), Rice leaf folder (Cnaphalocrocis medinalis), Army worm (Mythimna separata), Thrips (Stenchaetothrips biformis (Bagnall)), Asian rice bug (Leptocorisa sp.), Leaf miner (Hydrellia sp.), Rice caseworm (Parapoynx stagnalis), Yellow stem borer (Scirpophaga incertulas), Striped stem borer (Chilo suppressalis) and Asian rice gall midge (Orseolia oryzae).Respondents were asked to indicate whether they thought the pictured arthropod was good or bad for their rice crop.A choice was awarded 1 point if the respondent correctly classified the arthropod as a pest or beneficial.Wrong classifications or 'do not know' answers were rewarded a score of zero.Respondents were also asked if they had seen the arthropod in their rice crop in the last five years using a 'yes', 'no', 'do not know' scale.
Open-ended questions were used to assess the respondents' knowledge of protective clothing ('why is it important to wear PPE when spraying') and alternative insect pest management practices ('Please explain other forms of insect control you use').
Attitude refers to respondents' beliefs about pesticides with a specific focus on insecticides and their role in rice production, insect pest management, and the effects on human and environmental health.We measured respondents' attitudes using 11 statements (Table 3).To develop the attitude statements, we used an amalgamation of statements from LePrevost, Blanchard and Cope (2011) and Schreinemachers et al. (2017) with adjustments made to ensure relevancy with context to Cambodian rice farmers and insecticide use.Farmers' attitude towards insect pest management and pesticide use was measured with a series of statements such as 'Insect pests are the main cause of yield loss' and 'I cannot farm rice productively without using pesticides to control pests', where the respondents were asked to rate their extent of agreement on a five-point scale from 1 to 5 where 1 was strongly disagree; 2, disagree; 3, neutral; 4, agree; 5, strongly agree.Those who agreed or strongly agreed were classified as having a strong belief.
Practices or behaviours are the observable actions of an individual (Gumucio et al., 2011).Here respondents were asked about rice production practices, insecticides used, quantity and frequency of application, mixing of insecticides, cost of insecticides, their sources of pest management information, and protection used when applying pesticides.
The survey was written and designed in English and translated into Khmer by native speaker Dr Van Touch.The survey was coded into CommCare, version '2.46.0' (Dimagi inc), in 2018, a mobile acquired data (MAD) collection platform, and uploaded onto Android tablets used in the field.All surveys were conducted face-to-face in Khmer by four Cambodian master students from the National University of Battambang (NUBB) trained as interviewers.We performed pilot surveys on four farmers in Svay Cheat village to ensure appropriate questions, translations, and configuration.Any minor changes required were made after the pilot surveys.Each survey took approximately 30-60 min to complete, and all respondents provided informed and signed consent before undertaking the survey.Approval from the University of Sydney Human Research Ethics Committee (HREC) (Project number: 2016/882) was obtained before this survey was conducted.
The surveyed farmers were selected using an opportunistic sampling method in each village with snowballing sampling to reach further participants, such as obtaining mobile numbers of other farmers from village leaders.The head of households/ farm management decision-maker made up most participants for the survey (65%).However, if they were unavailable for reasons such as being away in the field, a representative of the farming household participated instead.Other is the combination of the rice varieties: Romdoul, Neang Khon, Neang Minh, Car 15 and Jasmin.

Analytical approach
Quantitative data were analysed using descriptive statistics, percentages, means and standard deviations in R studio software version 4.0.2(2020-06-22).The specific variables analysed include village demographics, causes of yield loss, insect incidence, sources of pest management advice, farmers' knowledge and attitudes towards pests and pest management, and pest management practices.For money spent on pesticides, the standard errors were large, and hence, medians are presented.
For qualitative data, thematic analysis was performed through manual coding of open-ended verbatim question 'how do you decide when to spray insecticide?' into common themes relevant to farmers' decisions.Direct verbatim quotes from the open-ended questions are used as descriptive narratives for themes found in the thematic analysis.

Village demographics
Rice production in Northwest Cambodia is small scale, with 60% of farming enterprises focused on rice only.However, some farmers produce rice in conjunction with vegetables (20%), fruit (5%), and livestock (8%) (Table 1).The majority of farm management decision-makers are men, with only 13% being women (Table 1).
The land used for rice production across all five Northwest villages averaged 5.13 (4.58) hectares (ha) (Table 1).Farmers typically plant two rice crops per year in the early and late wet seasons (69%).Farmers who only plant one rice crop per year typically plant in the early wet season (Table 1).Only 17% of respondents produce rice in the late wet and dry season, and even fewer (11%) produce rice only in the early wet season.
92% of respondents used the broadcast sowing method to sow their rice seeds.The broadcast sowing method is a direct seeding method and involves sowing seed by scattering it by hand onto the soil (International Rice Research Institute, 2007).Conversely, mechanical direct drill methods were seldom used; only 3.6% of farmers used a drum seeder, a form of the mechanical seeder, for wet sowing, and 3.6% used a direct drill seeder (Table 1).
Sen Kra Oub, a fragrant short-duration rice variety, was planted by 71% of farmers.Other rice varieties include longer duration cultivars such as Srongae (19%).Farmers plant an average of 180.8 (standard error 40.98) kg of rice seeds per hectare (Table 1).The amount of rice harvested each year did not vary greatly between villages, with the average being 4.23, 4.30 and 5.06 t/ha for 2016, 2017, 2018, respectively (Table 1).
There was little variation in the average amount of rice sold annually (t/ha) between villages, ranging from 4.02 t/ha in Prey Toteung to 5.84 (t/ha) rice sold annually in Ous Tok (Table 1).The average across all villages was 4.68 t/ha rice sold annually.The annual income earned per household from rice varied, with the median incomes for both Ous Tok and Svay Cheat more than double that of the other villages.
The majority of respondents earned an income from rice, and as such, rice production in the five villages is not one of traditional subsistence farming (Table 1).Only two farmers did not earn an income from the rice they grew and harvested for personal consumption only.
Respondents identified several ecological and social constraints when producing rice.Across all five villages, 159 out of 168 (95%) respondents considered insect pests one of the top three most important factors causing crop loss in rice, followed by rice disease and farmers having no water (Figure 1).
Seventy-six percent of respondents thought that the insect population had increased in the last five years.Similarly, 75% of respondents thought insect damage had increased in the past five years (Figure 2).
To manage and solve pest issues and incidence, 92% of respondents received pest management advice from pesticide sellers, neighbours/ friends (68%) or other farmers (56%) (Figure 3).

Knowledge and attitudes
Respondents were shown photos of various commonly occurring pest and beneficial arthropods to establish their ability to distinguish between pest and beneficial arthropods (Table 2).Farmers were somewhat able to distinguish between beneficial and pest arthropods (70% correct on average).Farmers' ability to identify pest arthropods (77% correct on average) was much greater than correct identification of beneficial arthropods (51% correct on average) (Table 2).Brown planthopper, for example, was identified as a pest correctly by 91% of respondents.
Damselflies and Asian honeybee were correctly identified as beneficial by an average of 77% and 79% of respondents, respectively.In comparison, only 4% of respondents correctly identified lady beetles as a beneficial insect, with 77% incorrectly identifying the lady beetle as a pest insect for their rice crop.The remaining respondents said they did not know or identified lady beetle as neither good nor bad.The functional role spiders play within the rice ecosystem appeared unclear to respondents, with 46% of respondents correctly identifying spiders as beneficial arthropods, while 29%, 8%, 17% identified spiders as bad, do not know or neither good nor bad for their rice crop, respectively.A summary of the respondents' attitudes regarding pests and beneficial arthropods are in Table 3.Most respondents believed that insects are the leading cause of yield loss and agreed that insecticides should be applied whenever insects are seen in the rice field.However, almost all respondents agreed  that some insects are good for pest control and that insecticides can be harmful to these beneficial insects.
Similarly, all respondents believed that pesticides could be harmful to human and environmental health.However, many respondents thought that rice cannot be farmed productively without pesticides and that no other pest control is needed.Moreover, there is a strong belief that pesticides increase rice yield.While respondents somewhat hold a strong belief that pesticide use should be stopped or reduced, the belief that alternative pest control methods can be just as effective as pesticides was low.

Pest management practices
Chemical insecticides are the primary form of insect management used by 99% of respondents (Table 4).Ous Tok village was the only village that did not have 100% of respondents indicate they used chemicals as their primary source of pest control.Very few respondents use alternative pest control methods (Table 4).Resistant rice varieties such as Srongae, used by 6.4% of respondents, was the most used alternative pest control method.
Most farmers used insecticides multiple times per season, with the average number of insecticide applications per season being 2.10 (1.33) with slight variation between villages (Table 4).Typically, farmers are applying an average of 2.02 (1.33) different insecticides per season.Mixing different insecticides is common with 63% of respondents mixing different insecticides during the same treatment.
Table 5 shows a list of the names and types of insecticides respondents used on their rice crop.The main insecticides used were Prevathon® (Chlorantraniliprole), Winner® (Emamectin benzoate + Chlorpyrifos), and Golden Dragon 585 EC® (Chlorpyrifos-ethyl + Cypermethrin).However, many respondents could not remember or did not know the name or type of insecticide they had been using.The median spent on insecticides was 36 USD per season, based on the average 4.58 ha rice planted.There was some slight variation of insecticide use practices between villages (Table 4).Respondents from Svay Cheat village used fewer insecticides and spent less on insecticides (Table 4).Conversely, Boeing Pring respondents used more insecticides, thus spending more.
When applying insecticides, 99% of respondents agreed it was important to wear protective clothing to protect their health.Some respondents added extra details about their perception of potential chemical hazards.For example: 'Protecting from bad effects on health that may later lead to other diseases' and 'Protecting chemicals getting in through the skin because it is dangerous to our health'.These quotes indicate that respondents are aware of chemicals entering the body through the skin, and thus believed that the skin requires protection.
Respondents commonly wore items covering the arms, legs, and head when applying insecticides (Table 4).Protective items to protect the hands and feet were less frequently used.Protective clothing items to protect chemical absorption through the eyes were rarely worn.However, 86% of respondents indicated wearing a paper mask while spraying, although only 1% wore a gas mask to prevent inhalation of chemicals.
Respondents were asked how they decided when to apply insecticides which resulted in responses relating to what factors trigger insecticide spraying events.We categorized the responses into common re-occurring themes.We identified five shared themes that respondents identified as triggers to decide when to apply insecticides (Figure 4).
'When insects are present' was the most common trigger with respondents, as evidenced through quotes such as 'When there is small insect population in the field'.Perception of insect damage is another common decisionmaking factor, for example, 'Spray immediately when Table 5.A list of the names and types of insecticides respondents used on their rice crop.Active ingredients and concentrations came from packaging of the insecticides (if they were available).Thedepicts that the information could not be found (Insecticide Resistance Action Committee, 2021; University of California, 2014;Socheata, 2005) According to some respondents, insecticides are a prevention method, 'Spray to prevent the occurrence of insects'.
Other respondents decided to spray based on the calendar spraying method, a spray method based on an exact time after sowing or stage of rice growth, for example, 'when the plants are 30 days after sowing' and 'During panicle formation stage'.

Input access and potential for future education
Respondents indicated having access to various inputs relating to agricultural production and pest management (Figure 5).We found 81% of respondents have access to input sellers, with considerably fewer indicating they have access to other agricultural education sources such as agricultural training (43%) or extension officers (36%).
Only 5% of respondents indicated they had access to general pest management information.Technology-based pest information is scarcely accessible to farmers, with 2% of respondents indicated having access to a smart mobile phone, 1% access to pest management phone applications and 1% having access to the internet (Figure 5).
Regarding training and education, 97% of respondents agreed they would attend a field school or training programme for various reasons (Table 6).Respondents wanted further education and training on general rice production, non-chemical insect control methods, more effective pesticide applications, and improved cropping practices to increase productivity.

Discussion
Our study suggests that pest management knowledge, attitudes, and practical aspects of pest management in smallholder rice farms in Cambodia contribute to 'pesticide lock in' and exacerbate  insecticide dependency (Flor et al., 2019;Flor et al., 2020).We found that all respondents considered insect pests to be the main constraints to rice production.The management of insects by 99% of respondents consists of chemical insecticides, typically without any other form of pest management and inadequate PPE, which increases the risk of exposure to poisoning.Moreover, respondents have minimal knowledge of beneficial arthropods, targeting them as pests while obtaining most of their pest management advice from pesticide sellers.The summary of our findings is consistent with studies from Cambodia and other SE Asian countries where insect pest management and pesticide dependence is an issue (Castilla et al., 2020;Flor et al., 2019;Jahn, Sophea, Bunnarith, & Chanthy, 1997;Ngin et al., 2017;Norton, Heong, Johnson, & Savary, 2010;Schreinemachers et al., 2017).Exaggerated expectations of yield loss associated with the presence of insects within respondents' rice fields are driving rice farmers to use insecticides (Horgan et al., 2017).This, coupled with minimal extension support and lack of adequate basic ecological knowledge, means risk-averse farmers tend to overreact to minor infestations and apply insecticides whenever they see an insect, a symptom of pesticide dependency (Heong & Escalada, 1997;Horgan et al., 2017).Moreover, there is little understanding of actual yield losses due to rice pests in Cambodia, as farmers often cannot correctly predict realistic damage potential, i.e. economic thresholds (Horgan et al., 2017).Poor understanding of the damage potential of insect pests is evident as insects in the field are the main reason for applying insecticides (Figure 4), and 76% of respondents agreed they should apply insecticides whenever they see any insects in their rice field (Table 3).
Two practices, seeding rate and the use of broadcast sowing, may contribute to increased pest problems.Higher seeding rates and direct seeding techniques such as broadcast seeding are correlated with increased incidence and spread of insect pests and diseases (Horgan, 2017).In our study, the average seeding rate was 180 kg/ha, much higher than the recommended seeding rate of 100 kg/ha by the International Rice Research Institute (IRRI) (IRRI, 2018) and higher than the seeding rate of 150 kg/ha in another study in Battambang by Flor et al. (2019).Flor et al. (2019) found that farmers who used a higher seeding rate tended to report higher insecticide usage (more applications) in the dry season.In the wet season, lower seed rates and lower numbers of fertiliser applications were correlated with no insecticide use (Flor et al., 2019).Although our study did not examine the correlation between seeding rate, seeding method and insecticide use, it is possible that high seeding rates could exacerbate pest problems, which could in turn cause farmers to use a greater volume of insecticides.
Many of the insect pests observed by respondents in the fields are generally not problematic, especially if they are managed efficiently.A review by Matteson (2000) concluded that farmers' common practice of early-season insecticide sprays against stem borers and defoliators is generally unnecessary.For example, an extensive experimental study performed by Savary, Willocquet, Elazegui, Castilla, and Teng (2000) on pest damage in lowland rice-producing areas in Asia found that of insect injuries, only 'whiteheads' caused by stem borers were of any significance, causing 2.3% in yield losses.The yield loss caused by weed competition was considerably more significant (20%) than insects (Savary et al., 2000).Defoliators, such as the Rice leaf folder (Cnaphalocrocis medinalis), attack rice in the early crop stages, causing obvious leaf injury.However, due to the ability of rice plants to compensate for early foliage loss, the injury often does not result in yield loss (Matteson, 2000;Ngin et al., 2017;Norton et al., 2010), and the control of leaf feeders generally does not increase rice yields (Ngin et al., 2017;Heong et al. 1994;Savary et al., 2000).Ngin et al. (2017) determined that the indiscriminate insecticide application did not increase the rice yield but rather decreased the net income of farmers in Cambodia.This is particularly true if insecticide application encourages secondary pests such as the Brown planthopper (Nilaparvata lugens Stål) and reduces the presence of predatory arthropods that would otherwise be present in farmers' rice fields.
Pest identification or misidentification likely plays a role in pesticide application, i.e. farmers misidentifying spiders and lady beetles as pests and spraying to eradicate them.Seventy-six percent of respondents identified lady beetles as a pest, yet 85% observed them in their field.Farmers' misidentifying beneficial arthropods is of major concern.Micraspis discolor is considered an important beneficial arthropod for controlling rice pests, including Brown planthopper (Rahaman et al., 2016;Begum et al 2002;Begum, Jahan, Bari, Hossain, and Afsana 2002).The misidentification of Micraspis discolor is a novel finding for rice farmers in Cambodia in this study, although other studies have found similar results in vegetable (Schreinemachers et al., 2017) and mungbean farmers (Hinchcliffe, Quinnell, Martin, Touch, & Tan, 2019).
Our data are consistent with those of Schreinemachers et al. ( 2017), who found vegetable farmers in Cambodia were much better at identifying harmful arthropods (69% correct on average) than beneficial arthropods (23% correct on average).In their study, 38% of farmers identified spiders as beneficial, and only 1% understood lady beetles were beneficial.This trend is supported by an earlier study by Jahn, Sophea, Bunnarith, and Chanthy, (1997) in Cambodia in 1995, where 43% of 1265 rice farmers were unaware that natural enemies (beneficial arthropods) exist, and 43% of dry season rice farmers reported lady beetles as pests.These data combined with the data from the present study suggest that from 1995 to 2019, there has been no apparent improvement in farmers' awareness of beneficial arthropods within the field sites studied.Schreinemachers et al. (2017) attributed the low percentage of identification to a lack of knowledge of beneficial arthropods and a tendency to regard all arthropods as pests.Schreinemachers et al. (2017) found an association between a greater knowledge of beneficial arthropods with less pesticide use.Therefore, improving farmers' knowledge of beneficial arthropods to encourage a reduction in pesticides is an example of how knowledge can be the foundation of behaviour change.In our survey, respondents highlighted areas for further education, which included insect management (education about beneficial insects and economic/ damage thresholds) and practical pesticide usage, effective timing, further education about risk and exposure to pesticides, including one respondent saying they 'want to learn about non-chemical insect controlling practices' (Table 6).
Farmers' current knowledge and attitudes towards the role of pesticides within the rice agroecosystem is influenced by their experiences and the sources providing the information.Respondents agreed that pesticides are harmful to their health and the environment and that insecticides can harm good insects.However, many believed that rice yields would increase with insecticides, and without insecticides, they cannot farm productively (Table 3).Their pest management knowledge sources likely influenced farmers' attitudes towards pesticides, and in our study, most farmers trust pesticide sellers to provide pest management advice.
Similarly, several other studies have found that Cambodian farmers are inclined to follow the instructions of pesticide sellers and selected the most 'effective' products based on sellers' opinions (Flor et al., 2020;Matsukawa et al., 2015;Matsukawa et al., 2017;Seng & Chhunhy, 2012).Respondents seeking advice from pesticide sellers were observed by Schreinemachers et al. (2017) to use much more pesticides than farmers seeking advice from friends and neighbours.The ease in access to pesticidebased recommendations is an example of a typical pesticide lock-in issue (Flor et al., 2020).Improvements in the quality of pest management information farmers have access to, and increasing farmers' access to various sources of information, such as extension officers, NGOs, and village meetings between farmers themselves, could promote a more comprehensive understanding of alternative pest management methods and which could encourage a reduction in insecticides (Flor et al., 2020).
Reducing insecticides is a crucial step towards sustainable pest management, particularly because many of the insecticides that the respondents used are known to have environmental and human health impacts.Organophosphate insecticides used by respondents such as chlorpyrifos and cypermethrin are toxic to members of the Coccinellids family (such as lady beetles) (Chemseddine, 2002;Sattar, Azam, Sarwar, Amjad, & Malik, 2018) and are highly toxic to humans (Damalas & Koutroubas, 2016).To reduce health impacts, promoting more selective insecticides or biological insecticides such as Beauveria, currently used for mungbeans in Cambodia, could be a practical solution (Martin, Montgomery, Yous, & Rien, 2020).However, the promotion of any form of insecticide should be in conjunction with safety measures, such as encouraging the use of PPE while spraying.
Respondents were generally aware of the importance of wearing protective clothing when spraying pesticides.Respondents were aware of the risks of absorption through the skin and protected themselves by wearing long sleeves and long pants.This is particularly important in rice paddies as the farmers often work with their feet and legs in the water, which can have high concentrations of pesticides (Migheli, 2021).However, correct protective clothing or personal protective equipment (PPE) was rarely fully implemented to eliminate risk.Respondents in our study reported a consistent lack of adequate protection of the face, eyes, hands, and feet.
Moreover, the efficiency of some items worn more commonly such as the paper / surgical mask, have been reported to have a very low filtration efficiency for insecticides and likely offers inadequate protection, particularly in comparison to a respirator (Sapbamrer, Hongsibsong, Naksata, & Naksata, 2021).Unfortunately, exposure during typical pesticide handling often occurs in body areas that remain uncovered by protective clothing, such as the face and the hands and through inhalation (Damalas & Koutroubas, 2016;Damalas & Eleftherohorinos 2011).Schreinemachers et al. (2017) found that most applicators applying pesticides to vegetables in Cambodia reported having experienced a range of pesticide poisoning symptoms after spraying (headaches, nausea).Farmers' exposure to pesticides can be minimized by reducing pesticides and through the correct use of protective clothing at all stages of pesticide handling (Damalas & Koutroubas, 2016).However, adequate PPE items such as a respirator may be inaccessible to smallholder rice farmers, particularly because their socio-economic conditions do not allow them to obtain sufficient PPE (Sapbamrer & Thammachai, 2020;Sapbamrer et al., 2021).Reasons behind the lack of PPE are often due to the type of PPE available, and the instructions suggested on the packaging are often general, and not specific to hot, humid tropical climates, which are unrealistic for Cambodian farmers.The heat in tropical countries can discourage the use of heavy PPE equipment (Williams, 2002).
Going forward, considerations around PPE for smallholder farmers should be based on pesticide toxicity and practical use, ensuring balance between the risks from pesticide exposure and acceptance of PPE use (Sapbamrer et al., 2021).For example, a study in Thailand found several common face coverings such as bandanas and cotton masks available at local markets provided greater filtration of insecticides than paper masks and might be a suitable alternative, particularly for low to middle-income countries in tropical climates (Sapbamrer et al., 2021).However, it is important to investigate forms of PPE that suit the location, climate, and application of the pesticides to ensure the best protection for farmers.Enhancing farmers' education of suitable PPE options that are more practical could reduce the risk of exposure and could be a public health strategy that could improve farmers' health and the wider community.A strategy such as this could be particularly beneficial if incorporated into a principle-based on reduced use, reduced risk, and reduced dependence (Damalas & Koutroubas, 2016;Thetkathuek, Suybros, Daniell, Meepradit, & Jaidee, 2014;Williams, 2002).
Another important factor to be considered, along with PPE when applying pesticides is the application equipment used (Food and Agriculture Organisation (FAO) and World Health Organisation (WHO) 2020).While pesticide application is increasingly popular in smallholder communities, the application methods and equipment are often problematic (Abhilash & Singh, 2009).Our study did not investigate the application practices and equipment used by Cambodian rice farmers, however, other studies in Southeast Asia found that smallholder farmers are often using poor pesticide application methods, which can cause detrimental health problems and contribute to environmental pollution (Chhun, Kumar, Martin, Srean, & Hadi, 2020;Thuy, Van Geluwe, Nguyen, & Van der Bruggen, 2012).
Typically for small-scale farmers, the choice of pesticide spraying equipment is dictated by what is affordable and available, which may involve sharing equipment between neighbours (Mathews, 1998).However, a lack of training on the equipment's use and maintenance, the lack of training on dosage and calibration (e.g.wrong equipment setting, too much liquid output and too much pressure, wrong nozzle choice), lack of spare parts and the use of faulty equipment can cause incorrect application of pesticides, which has been found to result in leakages (Mathews, 1998;Abhilash & Singh, 2009;Food and Agriculture Organisation (FAO) and World Health Organisation (WHO) 2020; Chhun et al., 2020).Coupled with insufficient PPE, leakages and poor application practices can result in pesticide poisoning of the spray operator (Abhilash & Singh, 2009;Mathews, 1998).There is room to expand research into improving application equipment and training operators on safe pesticide application practices.
Regarding education and training about pesticide use and IPM techniques, we suggest promoting face to face training and local knowledge transfer for pest management education at a village level.Our study found that respondents appear to prefer faceto-face training, with 97% of farmers wanting to attend and participate in farmer field schools and village meetings.The farmer field school (FFS) model has been used as an extension tool to promote and educate farmers about IPM in Cambodia to date (Chhay et al., 2016).However, we suggest a more comprehensive implementation of these FFS, focusing on promoting farmer trainers within villages, an educational method promoted as part of the National IPM Programme (Flor et al., 2020).Respondents want to share the knowledge with other farmers and their community.One respondent said, '[I] want to get more knowledge and then share with the other farmers who don't know ' and '[I] want to gain more knowledge on agriculture to help develop my village' (Table 6).Utilizing community knowledge structures to disseminate sustainable pest management advice could be an appropriate bottom-up approach to empower farmers out of a pesticide lock-in.

Conclusion and implications
This study detailed smallholder rice farmers' knowledge, attitudes, and practices regarding insect pest management and investigated how these attitudes and behaviours impact pesticide dependency and contribute to a pesticide 'lock-in'.Combinations of agronomic and pesticide application practices contribute to pesticide dependency, including high seeding rate, broadcast sowing, insecticide timing and application rate, and use of broad-spectrum insecticides.These behaviours were driven by respondents' knowledge and attitudes towards pest management and pesticide dependency, including reliance on pesticide sellers for information, misidentification of lady beetles as a pest, and the belief that pesticides will increase yield.
To encourage behavioural change, we identified several areas where education and knowledge could be strengthened, providing a pathway out of pesticide lock-in and improving IPM.
Firstly, education and training should focus on identifying the role of beneficial arthropods, particularly the common predator found in farmers' rice fields, Micraspis discolor.
Secondly, there should be education, training, and further Cambodian specific research on the economic or damage thresholds of commonly occurring rice pests such as Rice caseworm (Parapoynx stagnalis) and Rice leaf folder (Cnaphalocrocis medinalis), which will help to reduce unnecessary or early pesticide spray events.
Thirdly, an investigation into improving cropping practices, such as optimal seeding rate to reduce pest incidence without pesticide application.
Fourthly, using alternative pesticides such as Beauveria alongside research into appropriate economic thresholds should be encouraged to reduce reliance on broad-spectrum insecticides.
The fifth entry point is improving awareness of pesticide toxicity and exposure and how that influences health risks.Here, we suggest further investigation into PPE alternatives that are available and suitable such as bandanas, to reduce the risk of exposure.
A sixth entry point we suggest is to investigate how pesticides are used and the pesticide application equipment available to or utilized by farmers.There is an opportunity to improve application practices (such as dosing, calibration, nozzle choice) to help minimize misuse and pesticide poisoning.
It is suggested that education and training be provided through face to face, farmer orientated, handson learning in field days or, by extension, officers visiting farmers' fields.
Highlighting critical areas required for IPM in Cambodia is essential and finding methods to engage and empower farmers to make informed decisions about their pest management will forge a pathway away from pesticide dependency.

Figure 1 .
Figure 1.The main causes of rice yield loss according to farmers.

Figure 3 .
Figure 3. Information sources used by farmers to obtain pest management information.

Figure 5 .
Figure 5. Inputs accessible to across all five villages in Battambang province.

Table 1 .
Farm demographics of rice farming households throughout five villages.Total is the %/average of total respondents.Standard deviation in ()

Table 2 .
Ability of farmers to correctly identify the pest and beneficial arthropods that commonly occur in rice fields.

Table 3 .
Attitudes /beliefs farmers' have regarding pest management and pesticide use.The values indicate the proportion of farmers that agreed or strongly agreed with each statement, a higher score indicates a stronger belief.

Table 4 .
Respondents' pest management practices in rice production in five villages in Northwest Cambodia, average per farm, 2018-2019.Total is the %/average of total respondents.The results of removing/catching/trapping insects have been combined due to them being similar control methods.Standard deviation in ().
. eat the crop' and 'First I go to check the rice field and when I see the symptom of attack by insect, and I spray insecticide' Figure 4. Factors that trigger insecticide spraying events categorised into five themes re-occurring within the farmers' responses insects began to

Table 6 .
What respondents want to learn at field schools and indications for future education.