Adaptation strategies to the impacts of climate change: smallholder farmers’ decision-making on climate-smart practices adoption

Abstract

Climate change severely threatens stallholder agriculture in Mozambique, where frequent droughts, floods, and cyclones undermine food security. Climate-smart agriculture (CSA) offers resilience-enhancing practices, yet despite decades of promotion, adoption among smallholders remain low, limiting climate adaptation. This thesis investigates farmers’ perceptions of CSA practices and determinants of adoption, and synthesizes evidence on how digital agricultural technologies influence CSA decision-making. The study comprises a cross- sectional survey of 400 smallholder farmers in Zavala district, Mozambique, and a systematic review of 82 studies across sub-Saharan Africa. Results show that awareness of 13 CSA practices ranged from 47-100%. Three awareness-knowledge patterns emerged–awareness exceeding knowledge (rainwater harvesting, 81% awareness vs. 39% knowledge), knowledge exceeding awareness (improved varieties, 67% knowledge vs. 47% awareness), and aligned awareness and knowledge patterns for eight practices. Understanding of CSA principles lagged considerably (r = 0.151) with knowledge. Perceived compatibility (β = 0.342) and trialability (β = 0.418) were the strongest adoption drivers. The systematic review identified three complementary pathways through which digital technologies influence perceptions: real-time information provision, predictive analytics and integrated indigenous-scientific systems. Education was the strongest predictor of awareness-knowledge pattern: secondary-educated farmers knew nearly triple the practices of primary-only farmers (11.60 vs 6.82, p < 0.001). Women outperformed men on awareness (11.34 vs 7.47), knowledge (10.21 vs 6.82) and principles (9.45 vs 6.24) all at p < 0.001. Multinomial logistic regression showed that age (OR = 0.391), male gender (OR = 0.068), farming experience (OR = 1.322), and cooperative membership (p < 0.026) significantly differentiated adoption levels. Linear regression explained 83% of variance in awareness, 87% in knowledge but only 7.6% in understanding of principles. Systematic review identified interconnected infrastructural, economic, institutional, and gender as barriers to adoption. These findings imply that extension should shift from awareness campaigns to principle-based participatory training, align terminologies with farmers language, invest in secondary education, leverage women as peer educators, and address structural barriers like infrastructural, credit, and land rights. Digital technologies require simultaneous investment in enabling environments to scale, and adoption depends more on deep understanding and compatibility than simple awareness