A novel method based on high-resolution imaging spectroscopy to predict fine-root traits and the economics space of...

Tree fine-root morphological, anatomical, and chemical traits are important to reflect belowground resource acquisition strategies to support tree growth, but their measurements involve several labor-intensive and time-consuming procedures. We examined the efficacy of visible (VIS), near-infrared (NIR), and shortwave infrared region (SWIR) spectral reflectance for predicting a comprehensive set of fine-root traits. Spectral features of fresh fine roots in 20 tree species were analyzed using three distinct spectral datasets: the VIS-NIR (458-960 nm), the SWIR (967-2391 nm), and the combined spectral region (full-range; 458-2391 nm). The partial least square regression (PLSR) model using root spectral reflectance was developed to estimate nine functional root traits. Comparison of the models revealed that the prediction accuracy was higher when using the SWIR and full-range datasets than when using the VIS-NIR dataset. The trait with the highest prediction accuracy was root diameter. The predicted multiple root traits successfully reproduced the ordination of the multidimensional trait space, exhibiting a pattern highly similar to that of the observed traits. This method can advance our understanding of multidimensional fine-root trait spaces and resource strategies by enabling efficient measurement and spatial modeling of root traits across diameter classes.