Priming-induced drought memory in rapeseed and cotton: A meta-analytic assessment

Seed priming is increasingly used to enhance crop resilience to drought, yet whether it induces functional stress memory, and how this varies across crops, remains unclear. We conducted a global meta-analysis of 28 studies comprising 761 effect sizes to evaluate priming-induced drought memory in two major crops, rapeseed (Brassica napus) and cotton (Gossypium hirsutum). Effect sizes were calculated as standardized mean differences (Hedges' g) and synthesized using a random-effects model. Responses were evaluated across physiological, biochemical, and yield-related traits, while drought intensity and priming type were tested as key moderators to explain variation among studies. Priming significantly improved drought tolerance in both crops, but through divergent strategies: rapeseed responses were generally consistent with enhanced cellular protection via robust antioxidant activation (superoxide dismutase, catalase, ascorbate peroxidase) and reduced oxidative damage (malondialdehyde and hydrogen peroxide), particularly under moderate drought stress. In contrast, cotton may reflect a reproductive assurance strategy, sustaining leaf water status (RWC), accumulating proline, and significantly enhancing seed and boll yield, even when antioxidant responses were non-significant. Nutrient/mineral and microbial priming were most effective in cotton, whereas nutrient/mineral and hormonal/biochemical priming excelled in rapeseed. High heterogeneity in cotton (I² = 90.99%) indicates that priming responses are strongly influenced by experimental context, including drought severity and priming treatment. Overall, our findings are compatible with a functional, phenotypic interpretation of stress-memory-related responses rather than direct evidence of epigenetic memory, and highlight the potential value of crop-tailored priming strategies for improving drought resilience and supporting climate-smart agricultural production under increasingly water-limited conditions.