Comparative spectroscopic analysis of bee propolis: NIR, FT-IR, FT-Raman and 1H NMR. Focus on provenance and antioxidative correlation

Propolis is a polyphenol-rich resinous substance well known for its antioxidative properties. This study describes and compares the chemical composition and radical scavenging activity (RSA) of propolis from Australia (N = 8), Denmark (N = 38), and Norway (N = 7) using infrared (IR), Raman, near-infrared (NIR), and proton nuclear magnetic resonance ( 1 H NMR) spectroscopy through multivariate data analysis. Principal Component Analysis (PCA) showed that Danish and Norwegian propolis samples had a higher content of aromatic compounds, whereas Australian samples were richer in carbohydrates and terpenoids. The RSA of Danish and Norwegian propolis samples was found to be similar ( p ≥ 0.05), while the Australian samples exhibited significantly lower RSA. The strongest spectral correlation (R 2 = 0.64) to RSA was found at 1631 cm -1 (Raman) and 1666 nm (NIR), attributed to conjugated C=C stretching vibrations and first overtones of C-H stretching vibrations from aromatic/conjugated structures, respectively. NIR spectroscopy provided the most accurate predictions of RSA (R 2 = 0.75, RMSECV = 6.66%), highlighting its potential as a rapid, non-destructive quality assessment tool for propolis. Overall, this study demonstrates that the combined use of multiple non-targeted spectroscopic techniques provides complementary insights into complex matrices, enabling a robust characterisation of propolis and its antioxidant activity. • Bee propolis from Australia, Denmark, and Norway was analyzed using IR, Raman, NIR, and 1 H NMR spectroscopy • Danish and Norwegian propolis exhibited higher aromatic content, while Australian samples were richer in carbohydrates and terpenoids • RSA of the Danish and Norwegian samples was comparable and significantly higher than that of the Australian propolis. • Key RSA-related bands were identified at 1631 cm -1 (Raman) and 1666 nm (NIR). • NIR spectroscopy provided the most accurate RSA prediction (R 2 = 0.75, RMSECV = 6.66%)