Abstract
Evidence points to past bee-mediated crop pollination deficits in Chanthaburi province, Eastern Thailand. Conversely, no such evidence has yet been reported for Chiang Mai province (Northern Thailand), suggesting that wild pollination is delivered there above the requirements of local orchards. Discrete choice experiments (DCE) were conducted to elicit the preferences of pollinator-dependent orchard farmers with regard to three pollinator conservation measures and their possible effects on of native bee populations in each region. We fitted random parameter logit (RPL) models on the resulting data to capture preference heterogeneity and to obtain willingness to pay (WTP) point estimates. To test our results' robustness, we also inspected for scale heterogeneity by fitting generalized mixed logit (GMXL) models on the pooled and individual datasets. This yielded WTP space estimates (i.e., directly from WTP distributions) and made possible the comparison of farmers' preferences for a native bee conservation policy in both regions. The results hint at significant WTP differences for some of the conservation policy attributes between both provinces. Furthermore, unobserved contributions to choice seem to have been more random in Chiang Mai. Our analyses also suggest that farmers who engage in bee-related activities are WTP more for a conservation policy that includes bee husbandry.