Over 90% of global heavy rare earth elements (HREEs) reserves are found in the Nanling region of China in weathering crust deposits derived from granites, a deposit type unique in the world. However, the Nanling region is heavily vegetated, where dense plant cover acts as a natural barrier, posing significant challenges to traditional mineral exploration. We applied and validated a phytogeochemical remote sensing method using the indicator plant Dicranopteris dichotoma . Through controlled pot experiments under varying rare earth element (REE) stress, we discovered that specific spectral transformations and feature parameters could reveal significant correlations between leaf reflectance and soil REE contents (Ce, Gd, Y). Our analysis identified optimal, element-specific spectral indicators, with their Pearson correlation coefficients (r) relative to soil REE contents as follows: the absorption width at 570–720 nm for Ce (r = 0.75), and the first-derivative at 603 nm for Gd (r = 0.53). Notably, the first-derivative at 1856 nm and the absorption width in the 1300–1670 nm band were both identified as optimal indicators for Y, each achieving the highest correlation (r = 0.82). We further established that support vector regression (SVR) models significantly outperformed conventional methods, yielding an average increase of 0.05 in the coefficient of determination ( R 2 ) against stepwise multiple linear regression (SMLR) and a substantial increase of 0.21 against polynomial fitting, confirming superior accuracy and robustness. Our findings validate that plant spectra can function as a reliable bio-indicator for subsurface REE enrichment. This work proposes a novel approach for REE exploration in vegetated regions, representing a critical advancement towards sustainable and efficient resource discovery.
Li et al. (Fri,) studied this question.