Pupal age estimation of Dermestes maculatus (Coleoptera: Dermestidae) using hyperspectral imaging and machine...

Dermestes maculatus DeGeer, 1774 is a necrophagous beetle associated with the late stages of decomposition and skeletonization, and its pupal age is of forensic importance for long-term postmortem interval (PMI) estimation. However, pupal age estimation in necrophagous beetles remains poorly explored, particularly using non-invasive spectroscopic methods. In this study, hyperspectral imaging (HSI) combined with machine learning was applied for the first time to estimate pupal age in a necrophagous beetle. Under constant temperature conditions (25.0 ± 0.5 °C), D. maculatus pupae from D0 to D7 were examined daily by morphological observation and HSI acquisition from frontal and lateral views. Six regression models (SVR, RF, PLS-R, XGBR, PCR, and CLR) were developed using Raw, SG, and SNV spectra from frontal, lateral, and combined datasets. Results showed that morphological changes during the pupal stage were limited. Hyperspectral reflectance gradually decreased with increasing pupal age, and PCA showed continuous spectral variation throughout pupal development, with substantial contributions observed across the full spectral range. Across all models, R²P ranged from 0.70 to 0.94 and RMSEP from 0.56 to 1.26 d. SVR showed the best overall predictive performance across datasets. These results demonstrate that HSI provides a feasible non-invasive approach for pupal age estimation of D. maculatus and may improve the forensic application of necrophagous beetles in long-term PMI estimation.