Please login first
CHEMOMETRIC MULTIVARIATE ANALYSIS FOR SPECIES CLASSIFICATION IN THE GENUS BACCHARIS (ASTERACEAE)
* 1 , 1 , 1 , 2 , 2 , 1 , * 1
1  Department of Pharmacy, Federal University of Paraná, Curitiba, 80210-170, Paraná, Brazil
2  State University of Ponta Grossa, Ponta Grossa, 84010-790, Paraná, Brazil
Academic Editor: Gill Diamond

Abstract:

Within the genus Baccharis (Asteraceae), about 50 species collectively known as “Carquejas” are notable for their high content of essential oils and their widespread use in Brazilian folk medicine, particularly for their digestive, anti-inflammatory, and hepatoprotective properties. Despite this extensive use, only Baccharis trimera (Less.) DC. currently has an officially established quality control protocol in Brazil. This protocol, however, relies exclusively on morphological analysis, a restricted approach that is insufficient to ensure the rigorous quality standards required for commercially available products, such as capsules, extracts, and sachets. This study aimed to develop a chemometric exploratory and classification model using the MATLAB toolbox (version R2025a) to analyze hydroalcoholic extracts from five Carqueja species (B. trimera; B. articulata (Lam.) Pers.; B. sagittalis (Less.) DC; B. junciformis DC.; B. pentaptera (Less.) DC.). The dataset comprised 250 mid-infrared Fourier Transform spectroscopy spectra (4000–400 cm⁻¹), with 50 spectra per species. Principal Component Analysis (PCA) and Partial Least Squares - Discriminant Analysis (PLS-DA) were employed to construct the exploratory and classification models, respectively. PCA explained 98,0% of the total variance, allowing clear separation of the five Baccharis species using six principal components. The model with PLS-DA algorithm was built using the same dataset and evaluated through 10-fold Venetian Blinds cross-validation. It achieved a total explained variance of 88,9%, with an overall classification accuracy across classes of 90,0%, and specificity and sensitivity values of 90,0% and 71,0%, respectively, using four latent variables. The most relevant spectral regions contributing to the discrimination were located between 1750 cm⁻¹ and 750 cm⁻¹. Both models, the exploratory and the classification approaches, proved effective in evaluating and highlighting the differences among the Baccharis species, confirming the potential of chemometric tools as an innovative strategy for the quality control and regulatory monitoring of Carqueja-based products.

Keywords: Baccharis trimera; Multivariate Chemometric analysis; Plant classification
Top