View article

To examine the effects of tannin content in acorns of Quercus serrata on the fate of individual seeds, we developed a system for nondestructive chemical analysis of single acorns by near-infrared spectroscopy. We modeled the relationship between the normalized second-derivative spectra and tannin content by partial least squares regression. We obtained the best calibration model (r ² = 0.836; standard error of calibration = 1.90; standard error of validation = 1.43; bias = 0.10). The regression coefficient plots showed that the key band for the model was 938 nm, close to the absorption band of tannic acid in aqueous solution.