作者
Xianghui Lu,Haina Zhang,Zhihong Xia,Shuai Tian,Ming Jin,Zuwen Liu
摘要
Rare-earth tailing substrates severely restrict the growth and productivity of woody aromatic plants because of poor fertility, weak water- and nutrient-retention capacity, and acidity–salinity stress. In 2023, a pot experiment was conducted using one-year-old Cinnamomum camphora seedlings grown in rare-earth tailing substrate. A total of 12 treatments were established, including two controls, two fertilizer-only treatments, and eight biochar–fertilizer combined treatments, with 10 replicates per treatment and 120 pots in total. At three developmental stages (June, August, and October), canopy hyperspectral reflectance, soil plant analysis development (SPAD) value, gas-exchange parameters, growth traits, leaf nitrogen concentration, biomass, and essential oil yield were measured. This study aimed to evaluate the physiological regulation of biochar–fertilizer co-application on camphor seedlings in rare-earth tailing substrate, develop three-dimensional optical spectral indices (3D-OSIs) combined with machine-learning algorithms for harvest-stage essential oil yield prediction, and use partial least squares regression (PLSR) to identify key physiological drivers. Biochar–fertilizer co-application generally improved chlorophyll status, gas exchange, growth, and harvest-stage productivity compared with the controls. The F2 treatment achieved the highest essential oil yield (1.61 g plant −1 ), whereas F1D1 produced the highest biomass (21.82 g plant −1 ). Spectral performance was stage-dependent, with the highest absolute correlation coefficient occurring in October (0.57), followed by August (0.50) and June (0.44). Extreme gradient boosting showed the best prediction performance. The October model achieved the highest validation accuracy (R 2 = 0.83, RMSE = 0.17, MRE = 17.8%), while the August model still provided strong predictive capability two months before harvest (R 2 = 0.78, RMSE = 0.20, MRE = 17.0%). PLSR further indicated that August represented a key functional window driven mainly by transpiration- and photosynthesis-related processes, with comparable stage contributions across June, August, and October. Overall, this framework links biochar–fertilizer regulation, physiological attribution, and hyperspectral prediction, providing practical support for precision camphor cultivation in reclaimed rare-earth tailing lands.