CBGG

Climate change amplifies drought stress, threatening global crop yields and highlighting the urgent need for effective soil amendment strategies. SAPs are increasingly proposed as soil amendments to improve drought resilience, although their agronomic performance across species and trait categories remains insufficiently characterized. We systematically evaluated three SAP types—one biopolymer-based and two fossil-based—across 14 morphological, physiological, and biochemical traits in two maize and two rapeseed genotypes under controlled drought conditions. PCA and a RF-based feature selection identified SFW, PH, and total PFW as key responsive traits. However, SAP applications did not significantly enhance biomass accumulation or antioxidant activity, with pronounced genotype-specific responses, particularly in rapeseed. Fossil-based SAPs consistently outperformed the biopolymer-based variant. Our results demonstrate that SAP efficacy is species- and genotype-dependent, challenging the assumption of its universal benefit under drought. This multi-trait, dual-species analysis underscores the need for genotype-specific SAP strategies and highlights the value of machine learning approaches for predicting treatment outcomes. Overall, our findings contribute to a more nuanced understanding of SAP–soil–plant interactions and support the targeted development of climate-resilient agricultural technologies.

SAP; machine learning; stress response; abiotic stress