Multidimensional detection and predictive modeling of Tricholoma matsutake freshness during postharvest storage

Tricholoma matsutake is a highly perishable edible fungus whose freshness largely determines its commercial value, yet current evaluation methods remain subjective or incomplete. Sensory scoring cannot capture physicochemical changes, and single analytical techniques fail to describe spoilage dynamics. To address this gap, we established an integrated framework combining sensory and texture profiling with electronic-nose responses, gas chromatography-mass spectrometry (GC-MS), and amino acid analysis, followed by partial least-squares regression (PLSR) modeling for continuous freshness prediction. Controlled trials at 2 °C and 20 °C revealed clear temperature-dependent deterioration. Hardness decreased by 85.7 % at 2 °C and 96.0 % at 20 °C; chewiness declined by 32.0 % and 71.1 %, while adhesiveness rose by >10³-fold. Moisture loss reached 31.4 % at 2 °C versus 67.5 % at 20 °C by day 15. Volatile profiles showed accumulation of aldehydes and alcohols, while the C8 alcohol 2-octen-1-ol declined. Arginine decreased early at low temperature and, together with Sensor-5 response and 2-octen-1-ol, formed a compact biomarker set. The resulting PLSR achieved R² = 0.94 (2 °C) and 0.93 (20 °C), with no significant differences from sensory scores (paired t-tests, p > 0.05) and homogeneous variances (F-tests, p > 0.59). This multidimensional approach offers an objective, data-driven basis for freshness monitoring and provides deployable biomarkers for intelligent cold-chain management of Tricholoma matsutake.