National agroforestry program in Mexico faces trade-offs between reducing poverty, protecting biodiversity and targeting forest loss

Abstract

Abstract National reforestation initiatives with ambitious targets and multiple objectives are becoming the norm across the Global South. The extent to which these large-scale initiatives are actually achieving their multiple and potentially conflicting objectives, however, is largely unknown. Sembrando Vida, a national initiative in Mexico implemented in 2019, pays smallholder farmers to plant agroforests in order to reduce poverty and forest loss, and protect biodiversity. We assessed to what degree program recruitment met its stated objectives via its selection of participating municipalities and households. Because program data are not publicly available, we consolidated and harmonized >14 million policy payments (totaling ∼$4 billion USD) to smallholder farmers, thus creating the first spatiotemporal dataset of program outcomes. We found that ∼450k rural households in ∼1000 municipalities across the country participated in the program consistently from 2019 to 2022. The program was reasonably well targeted to achieve its poverty reduction objectives. Significantly more households (ANOVA, p < 0.001) were enrolled in high-poverty (10.4%) than low-poverty (4.9%) municipalities, despite more money being transferred in absolute terms to low-poverty municipalities. The program did not reach some regions that best fit its three goals. Using a zero-inflated negative binomial model, we showed that the distribution of participating households was more likely to address poverty (coefficient = 0.51, p < 0.001 at household level) and forest cover loss (0.1, p = 0.01) than to restore areas important for biodiversity (−0.08, p = 0.02). Finally, we conducted a spatial analysis showing that there is technically sufficient rural land (4.29 Mha) and households (491k) to maximize the potential of all policy objectives simultaneously, but this would require that the program operate in only 83 municipalities across 10 states. Our results highlight the challenges in reaching high poverty regions while meeting multiple other objectives when scaling up forest landscape restoration.