Species discrimination from dried urine stains using ATR-FTIR spectroscopy coupled with multivariate analysis

Species identification from body fluids is a persistent challenge in forensic practice, especially when samples are recovered from crime scenes in a dried or contaminated condition. Urine is an underexplored body fluid with rich biochemical signatures, which may support species discrimination. This study evaluated species discrimination from dried urinary stains using ATR FTIR spectroscopy coupled with multivariate analysis. Four domestic animal species – cow (Bos indicus), buffalo (Bubalus bubalis), goat (Capra hircus), and sheep (Ovis aries) – were selected for urine sample collection, with 13 independent animals sampled from each species (n = 52). Unsupervised PCA and supervised LDA were performed to discriminate species. The first three principal components accounted for 98.61% of the total spectral variance. PCA-LDA model achieved an overall classification accuracy of 86.53% using leave-one-out cross-validation and 84.62% using five-fold cross-validation. The confusion matrix revealed successful classification of cow and buffalo samples, while some misclassifications occurred between goat and sheep samples. The AUC values for all species were ≥ 0.96. These findings support ATR-FTIR spectroscopy coupled with multivariate analysis as a rapid, non-destructive proof-of-concept screening approach for species discrimination from dried urine stains; however, further validation using larger datasets and within-sample replicate measurements is required before routine forensic application.