Climate-Resilient Agriculture: Systems, Science and Solutions for a Warming World

Drought resilience in crops is often treated as a unitary tolerance phenotype, yet field performance emerges from interactions among genotype, developmental stage, soil water distribution, atmospheric demand and management. This critical narrative review evaluates how genetic, physiological and agronomic innovations can be combined to stabilise crop yield under water limitation while avoiding misleading gains in survival or leaf-level water-use efficiency that do not translate into harvestable production. Literature published from 1 January 2000 to 14 June 2026 was identified through complementary searches of major open scholarly indexes and agriculture- and life-science databases, with selective inclusion of foundational studies. Evidence was appraised for mechanistic strength, field realism, yield relevance, replication, environmental characterisation and transferability across target production environments. The synthesis indicates that the strongest individual advances are those that connect a defined drought scenario with a causal trait and a yield endpoint, as illustrated by deeper rooting, stay-green canopy regulation and selected engineered alleles. Nevertheless, trait value is conditional: deeper roots require accessible subsoil water, conservative transpiration may protect late-season water at the cost of early biomass, and osmotic or antioxidant responses can improve stress status without guaranteeing yield. Agronomic interventions show the same context dependence. Regulated deficit irrigation, residue retention, mulching and rhizosphere manipulation can improve water productivity, but their benefits vary with crop stage, soil, rainfall pattern, input access and long-term system effects. The central inference is that durable drought resilience is best pursued as a genotype x environment x management optimisation problem rather than by stacking nominal tolerance traits. Progress therefore depends on managed-stress phenotyping, explicit target-environment definition, genomic prediction linked to crop models, recovery phenotyping and multi-environment validation of integrated trait-management portfolios. This framework shifts evaluation from whether an intervention reduces stress symptoms to whether it reliably preserves yield, resource productivity and system viability across credible drought scenarios.