Climate-smart agriculture: Strategies for sustainable crop production under climate change: A review

Climate change threatens crop production through interacting changes in temperature, water availability, extreme events and biological pressures. Climate-smart agriculture (CSA) provides a framework for pursuing productivity and income gains, adaptation and resilience, and greenhouse gas mitigation where feasible. This critical narrative review examines how crop improvement, diversification, conservation agriculture, water and nutrient management, agroforestry, integrated pest management, adjusted crop calendars and climate information services can contribute to these objectives. It interprets foundational literature alongside selected recent syntheses, distinguishing plausible mechanisms from measured outcomes and identifying limits to transferability. The evidence supports locally adapted management, but does not justify treating individual practices as universally climate smart. Diversification can improve several ecosystem services without consistently increasing yield; no-tillage performance depends on accompanying residue and rotation management; and irrigation efficiency at field scale does not necessarily reduce basin-scale water consumption. In rice, alternate wetting and drying illustrates the need to evaluate yield, methane and nitrous oxide together. Across interventions, short study durations, heterogeneous comparators, incomplete greenhouse gas accounting and weak measurement of household resilience constrain inference. Adoption additionally depends on labour, finance, tenure, markets and access to usable advice. The review proposes an outcome-based assessment linking biophysical performance to economic feasibility and distributional effects. Priorities include long-term, multi-location evaluations of practice combinations, transparent resource accounting and participatory implementation. CSA is most defensible as an adaptive decision framework whose benefits must be demonstrated against explicit baselines, rather than inferred from technology adoption alone.