Photosynthetic performance index (PIabs) and malondialdehyde (MDA) content determine rice biomass under combined salt stress and prohexadione-calcium treatment.

backgroundPlant growth regulators (PGRs) have proven effective in alleviating salt stress damage in crops, yet the predictive hierarchy of physiological traits governing biomass under combined salt-PGR regimes remains unclear. In this study, we applied random forest machine learning with variable importance measures to quantify the relative importance of key physiological traits - including photosynthetic rate (Pn), photosynthetic performance index (PIabs), electron transport rate (ETR), maximum photosystem II efficiency (Fv/Fm), superoxide dismutase (SOD) activity, and malondialdehyde (MDA) content- in rice subjected to NaCl (0.3%) and Prohexadione-Calcium (PCa) co-treatment. Using contrasting rice varieties (salt-sensitive 9311 and salt-tolerant Changmaogu), we measured physiological responses at 14 and 21 days after treatment (DAT).

resultsThe analysis revealed MDA content as a universal stress biomarker, showing consistent negative correlations with biomass in both varieties. Notably, our analysis revealed distinct genotype-specific Pn-biomass relationship. While the salt-sensitive 9311 showed positive Pn-biomass correlations (r = 0.34 at 14 DAT; r = 0.63 and p = 0.37 at 21 DAT), the tolerant Changmaogu displayed negative relationships (r = -0.15 at 14 DAT; r = -0.75 and p = 0.25 at 21 DAT). These opposing patterns underscore the challenges of relying solely on conventional correlation analysis for biomass prediction in different genetic backgrounds. The random forest model effectively addressed this complexity, identifying photosynthetic performance index (PIabs) and MDA content as the most robust predictors of biomass.

conclusionBy integrating machine learning with traditional physiological analysis, this work provides valuable insights for rice breeding and management programs under saline conditions. The identification of PIabs and MDA as key biomarkers offers a practical framework for varietal selection. Particularly, PIabs serves as an effective positive selection marker, enabling rapid and non-destructive screening of salt-tolerant genotypes in large breeding populations.