Which factors determine adaptation to drought amongst farmers in Northern Thailand? Investigating farmers’ appraisals of risk and adaptation and their exposure to drought information communications as determinants of their adaptive responses

As Southeast Asia’s second largest economy, Thailand’s development significantly affects its neighbours, and the country plays an important role in policy-making within the wider region (Kiguchi et al. 2021). Agricultural production in Thailand, like other countries in Southeast Asia, is important for global food supply, with the region as a whole contributing 40% of global rice exports (Yuan et al 2022). However, rice yields have stagnated in Thailand in recent decades, as they have in several neighbouring countries, with drought having played a major role in crop losses in Thailand, Cambodia and Myanmar since 2015 (Venkatappa et al. 2021). Across Southeast Asia, there is an increasing risk of agricultural drought in the near term (Amnuaylojaroen and Chanvichit 2019), with projections suggesting countries across the region will experience longer, more intense and more severe droughts prior to 2050 due to the impacts of climate change under a range of probable emission scenarios (Supharatid and Nafung 2021). Identifying strategies to support farmers to stabilise and increase yields whilst reducing their vulnerability to drought is therefore a key concern for policy makers across the region that may look to Thailand as an example.

Drought has already had major impacts on the availability of water for irrigation in Thailand in recent years (Kiguchi et al. 2021), with this problem set to grow in the near future (Chang et al. 2021; Singhrattna and Babel 2011). Due to the recent incidence and severity of agricultural drought impacts, the Ping River catchment (a major tributary of Thailand’s main interior river, the Chao Phraya) was chosen for data collection. Significant regional droughts took place in 2015–2016 and 2019–2020 (Khan and Gilani 2021; Pak-uthai 2019), with reports that the timing of government drought warnings and advice hampered effective farm responses (Khan and Gilani 2021). Despite the major agricultural impacts and increasing risk of drought in Thailand, few studies have focused on climate risk communications (Arunrat et al. 2017; Lebel et al. 2018). As such, Northern Thailand provides a suitable context for investigating how institutional drought communications currently support farmers to build resilience to drought and for identifying strategies to optimise such support in future.

The research team purposively selected a sample of villages exposed to different levels of drought over the last decade. Villages both within and outside of formal irrigation zones were selected to represent a range of typical agricultural production profiles (such as rice, longan and livestock production). According to the customary context, the village headperson (a village elder who acts as the community leader) in each village was contacted initially, and the research team requested their permission to conduct the research. The village headperson then invited farm household heads to participate. As opposed to random sampling of households, this method may have resulted in a degree of sampling bias, meaning that the sample should not be interpreted as fully representative. In particular, readers should bear in mind the possibility that due to the selection of participants by the village headperson, older, wealthier farmers may be over-represented, whilst poorer, younger farmers and female-headed farms may be under-represented. However, this sampling approach was considered the most appropriate given the local socio-cultural context, and the validity of the analysis reported herein is not contingent on a perfectly representative sample. Trained enumerators undertook data collection after obtaining informed consent. In total, 176 questionnaires were completed in January 2020 with a close to even distribution of respondents coming from the provinces of Chiang Mai (n = 41), Lamphun (n = 45), Kamphaeng Phet (n = 45) and Tak (n = 45). The structured questionnaire collected socioeconomic, agricultural, risk appraisal, adaptation appraisal, adaptation behaviour and drought communications data from farmers. The questionnaire was originally written in English, translated into Thai and administered before translation back to English for analysis. Questionnaires were anonymised, and the questionnaire and dataset can be found at https://​doi.​org/​10.​17862/​cranfield.​rd.​16553136.

Statistical tests demonstrated the role of socioeconomic factors in determining access to drought warnings. Older farmers reported receiving drought communications through significantly fewer channels (mean = 1.0 vs 1.3 for farmers aged under 60, t = −2.883, p = .004) and tended to report receiving drought communications from a smaller number of other farmers (although p = .059). Respondents in older age groups were less likely to have received drought warnings that were accompanied by advice on what to do (Fisher’s exact test = 14.998, p = .003), as were respondents with lower education levels (Fisher’s exact test = 33.219, p = .000).

Wealth was also a factor, with the highest mean household asset and land wealth scores found within households that received both a drought warning and advice on what to do (10.0), compared to lower scores amongst those that only received warnings (8.9), and those that received neither (5.4), F(2172) = 5.416, p = .005.

Adaptation appraisal scores followed the same pattern, with the lowest mean scores in households that received no warning or advice (−0.85), mean scores of 0.36 in households that only received the drought warning and mean scores of 0.73 in households that received both the warning and advice, F(2169) = 15.419, p = .000. Respondents that received drought warnings accompanied by adaptation advice were significantly more likely to have changed something as a result, than those that received the drought warning alone (67.5% compared to only 25.0%, χ2 = 24.696, p = .000).

Figure 1 provides a conceptual overview of the main relationships reported in the results above, with the routes connecting the four adaptation outcome variables with risk appraisal and adaptation appraisal illustrated. Also shown are other sociodemographic and policy factors found to have significant effects on either adaptation outcomes or the cognitive pathways associated with adaptation appraisal and/or risk appraisal. Being in a formal irrigation zone is indicated as a policy intervention since farmers based in these areas benefit from irrigation zone policies which provide additional government-funded support. The results synthesised in Fig. 1 show that adaptation appraisal interacted directly with all four of the adaptation outcome variables whilst only a weak association was found between risk appraisal and one of the adaptation outcome variables. This lends weight to the idea (reported in the context of flooding by Babcicky and Seebauer 2019) that, although there may be two separate PMT pathways, only the one operating through adaptation or coping appraisal leads to the implementation of protective measures. As such, the results support Babcicky and Seebauer (2019), suggesting that the traditional interpretation of PMT, where high risk appraisal and high adaptation appraisal combined lead to the implementation of protective measures, may not be empirically accurate in this kind of context.

Looking at the relationship between PMT variables and the adaptation outcome variables analysed, this study reveals a much more pronounced connection between adaptation and adaptation appraisal than between adaptation and risk appraisal, which reflects findings from many previous PMT studies exploring a range of different risks and contexts (Bubeck et al. 2012; Milne et al. 2000; Poussin et al. 2014; Sheeran et al. 2014; Truelove et al. 2015). It has been argued that risk appraisal could be higher prior to an adaptation being undertaken, but then decline in light of adaptation success (Bamberg et al. 2017). Our finding of significantly higher mean risk appraisal scores amongst respondents that indicated wanting to implement further adaptations supports this possibility. However, since no significant associations were found between adaptation counts and other more stable factors that correlate with risk appraisal (for example, the number of droughts and drought impacts respondents reported having experienced), the authors argue that the findings suggest adaptation appraisal is the more robust behavioural predictor of the two PMT constructs (Milne et al. 2000).

The finding of a weak negative correlation between adaptation appraisal and risk appraisal illustrates that, rather than being purely additive, it is possible that high levels of risk perception may lower perceived capacity to adapt, or vice versa. The conceptual diagram in Fig. 1 depicts positive feedbacks between adaptation and adaptation appraisal, showing that direct experience of implementing adaptation and perceived adaptation success may increase cognitive readiness to adapt in future. These findings suggest that direct experience, observation and learning about adaptation and drought risk continually reshape perceptions, illustrating processes of adaptation as continuous rather than finite and suggesting that adaptation experience and a pro-adaptive mindset are mutually reinforcing. This identification of positive adaptation feedbacks mirrors those identified by Noll et al. (2022) in the context of flood adaptation. On the basis that adaptation actions may trigger further adaptation intentions, drought communications and adaptation campaigns should not be regarded as standalone responses, but rather as elements within long-term support programmes geared towards developing cultures of resilience and capacity-building (Ewbank et al. 2019). This finding is of particular relevance for countries where agricultural impacts from drought are projected to increase in coming decades.

Whilst direct personal experience of climate impacts strongly shapes adaptation motivations (Niles et al. 2015), climate and adaptation information received through personal and institutional networks also affects adaptation decision-making (Babcicky and Seebauer 2017). Further, contrary to the suggestion that top-down campaigns may have limited capacity to affect the uptake of protective measures against extreme climate events (Haer et al. 2016), our results indicate that institutional drought communications (via community, private or government institutions) interact significantly with adaptation attitudes and actions amongst agricultural producers. The diamonds in Fig. 1 illustrate where drought communications or direct policy interventions (the provision of drought compensation or official irrigation support) significantly predicted adaptation outcome variables. It is evident that both the size of a farmer’s local drought information network and the nature of the drought information they receive via institutional channels interact significantly with adaptation intention and behaviour. The study has shown that receiving more different types of drought information as well as the perceived efficacy of the information received are both positively associated with higher confidence in implementing adaptations and with the number of adaptations farmers have employed in the past. Additionally, our findings support the importance that drought information is accompanied by practical advice on what to do to limit negative impacts (Ewbank et al. 2019; Haer et al. 2016; Koerth et al. 2013).

Although the institutional drought communication measures significantly contributed to adaptation outcome variables (namely, adaptation appraisal scores, past adaptation counts and whether respondents reported having adapted in response to official advice), they appeared not to contribute to respondents’ drought risk appraisal. Instead, risk appraisal was influenced by local communication factors, such as the number of farmers drought information was received from, and whether information was received at a village meeting. This supports findings that local and informal risk communications are more likely to alter risk perceptions (Binh et al. 2020). Additionally, risk appraisal was influenced by direct personal experience of drought, supporting the suggestion that factors which are more psychologically close (either in time, space or socially) determine risk perceptions more strongly than official communications (Binh et al. 2020; Niles et al. 2015; Steynor et al. 2021).

The finding of a significant influence of local farmer communications on risk appraisal and past adaptation points to the role of social connectedness in facilitating access to information and as an important determinant of adaptation responses. Moreover, the apparent role of smartphone messaging group membership in enhancing responsiveness to official advice should not be overlooked. Overall, these findings highlight the benefits of strengthening opportunities for all farmers to participate in local and institutional drought communications networks via locally appropriate and accessible means, particularly in agricultural development contexts similar to the case study presented here.

Demographic variables such as age and education are important determinants of disaster warning response, with evidence that different segments of at risk populations may comprehend information differently (Mayhorn and McLaughlin 2014). Our findings reflect socio-economic patterns well-established by diffusion of innovations scholars (E. M. Rogers 2003), namely that education, age and wealth determine capacity to implement new behaviours. The study found that older farmers were more likely to indicate adaptation constraints, such as a lack of time or energy. Older farmers were also likely to receive drought information from fewer local farmers and via fewer communication channels. As such, this study highlights that older farmers may be less able to adapt to drought and supports findings that older individuals may have different communication needs and support should be tailored accordingly (Walkling and Haworth 2020). As drought information communications move online in future, the need may grow to implement additional communication strategies which ensure that older and poorer population segments do not end up missing out.

The study findings also suggest that livestock producers experience drought risks and impacts differently, resulting in measurably higher risk appraisal scores and lower perceived success rates for past adaptations. This indicates that households which focus on livestock production may likewise benefit from specific targeted drought support, reflecting findings from different regional contexts that drought responses within the livestock sector are different to other agricultural sectors (Salmoral et al. 2020).

The absence of a strong link between risk appraisal and adaptation behaviour (aligning with many other PMT studies) suggests that seeking to heighten risk appraisal by incorporating fear appeals within official drought communications is unlikely to enhance adaptation behaviour amongst farmers and signals the possibility that fear appeals could be counterproductive. Whilst this study supports the communication of targeted drought information to farmers, emphasis should be placed upon accompanying drought information with positive messages about household coping capacity and practical tailored advice about adaptation strategies which enhance farmers’ perceptions of both self- and adaptation efficacy (Binh et al. 2020; Ong and Araral 2021).

Although the findings do not support risk appraisal directly determining adaptation actions, higher risk appraisal levels may promote adaptation desire and/or intention (Milne et al. 2000). The findings indicated that risk appraisal levels may be higher following a direct experience of drought, but then decline thereafter. Drought communications should be targeted to take advantage of potential synergies between aspects of risk and adaptation appraisal (Sheeran et al. 2014), by ensuring that drought adaptation information and physical/financial assistance are made available soon after farming households have directly experienced drought, since this may lead to greater uptake than if only provided at a later stage.

The findings indicate that the constraint of not having enough money to implement the necessary adaptations may be delaying farmers from implementing desired further adaptations. Given the lack of a correlation between financial and the other adaptation constraints measured (time and energy), providing financial support to farmers who indicate an inclination to adapt may boost the uptake of drought adaptation measures. In some instances, pursuing this proactive approach to building drought resilience may be preferable to providing drought compensation after losses have already been incurred, particularly given the negative relationship between compensation for major crops such as fruit and rice and responsiveness to official drought warnings and adaptation advice. This finding reflects Tabe-Ojong et al. (2020), who show that relief assistance may negatively influence household resilience strategies for coping with extreme weather events. However, given that our study is cross-sectional, it is not possible to establish whether compensation arrived before or after farmers failed to take action in response to official drought advice; hence, longitudinal research on this issue is warranted.