Global warming increases the frequency with which drought and heat stress occur simultaneously, especially in semi-arid regions. Such combined stress imposes a non-additive and more severe impact on plant growth, yield, and quality than either stress alone. Here, we integrate recent physiological, biochemical, and multi-omics studies to compare individual and combined stress responses and to dissect the underlying signal transduction networks. We show that drought-dominated phases rapidly elevate ABA concentrations and activate SnRK2–AREB cascades, whereas heat pulses trigger jasmonic acid and ethylene signals that antagonize ABA-driven stomatal closure. Under combined stress, these hormonal modules converge on a “competitive TF marketplace”, where ABA, JA, and GA cis-elements co-regulate invertase–sugar checkpoints, heat shock factor/ROS oscillators, and chromatin-remodeling events that determine reproductive fate. Recent advances using multi-omics approaches and systems biology have further elucidated these complex networks. These insights will inform future breeding strategies aiming to develop stress-tolerant crops. We highlight emerging tools—weighted gene co-expression networks, kinetic multi-omics, and cis-regulatory CRISPR editing—that can exploit these signaling hubs for breeding crops with improved combined stress tolerance.

Combined drought–heat stress is not adequately described by adding the responses to drought and heat in isolation. Water limitation restricts evaporative cooling, whereas high temperature increases atmospheric demand and threatens photosynthesis, proteostasis, and reproduction. To assess what has been demonstrated rather than merely proposed, we combined a structured narrative review with a study-level evidence map spanning genomics, multi-omics, and emerging artificial intelligence (AI). The PubMed search, updated on 9 August 2026, returned 254 records; targeted and backward searches contributed eight seminal or reviewer-nominated publications. Independent screening retained 90 verified publications: 69 direct combined-stress studies, 11 genomics studies, six original AI or prediction studies, and four reviews or quantitative syntheses used for context. No conserved molecular signature emerged. Response direction varied with soil water status, vapor-pressure deficit, stress order, tissue, and developmental stage. Carbon allocation, redox and proteostasis balance, reproductive failure, and recovery recurred across omics layers, although neither their direction nor their adaptive value was consistent. Genomic loci and predictive performance were also environment dependent. Direct uses of AI remain uncommon and currently support prioritization more convincingly than causal inference. We organize this uneven evidence as a ladder extending from molecular association to functional validation, multi-environment prediction, and cultivar deployment. The main shortage is not another nominal omics layer but harmonized factorial experiments that measure recovery and reproduction, report field-relevant environmental metadata, and include external validation.