# Exploring Environmental Element Monitoring Data Using Chemometric Techniques: A Practical Case Study from the Tremiti Islands (Italy)

## TL;DR

This paper shows how chemometric methods can help analyze environmental data from the Tremiti Islands to detect pollution trends and improve data interpretation.

## Contribution

The study introduces a generalizable chemometric workflow combining multivariate techniques and data fusion for environmental monitoring.

## Key findings

- PCA and PLS-DA identified spatial and temporal trends in trace elements and BVOCs.
- Data fusion of BVOC and element data improved discrimination of environmental patterns.
- PARAFAC analysis revealed latent structures across chemical, spatial, and temporal dimensions.

## Abstract

Environmental element monitoring is essential for assessing environmental quality, identifying pollution sources, evaluating ecological risks, and understanding long-term contamination trends. Modern monitoring campaigns routinely generate large volumes of complex data that require advanced analytical strategies. This study applied chemometric techniques to analyze elements and BVOCs (biogenic volatile organic compounds) measured from Posidonia oceanica and related environmental matrices (seawater, sediment, and rhizomes) during three sampling campaigns in the Tremiti Islands (Italy). Twenty-two trace elements were quantified, and BVOC profiles were obtained from the leaf samples. The dataset was analyzed using a combination of univariate visualizations, unsupervised and supervised multivariate techniques, and multi-way methods. PCA (Principal Component Analysis) and PLS-DA (Partial Least Squares-Discriminant Analysis) revealed distinct spatial (leaf section) and temporal (sampling period) trends, supported by consistent elemental markers. A low-level data fusion approach integrating BVOC and element data improved group discrimination and interpretability. PARAFAC (PARAllel FACtor analysis) applied to a three-way array successfully separated background trends from meaningful compositional changes, uncovering latent structures across chemical, spatial, and temporal dimensions. This work illustrates the usefulness of chemometrics in environmental monitoring and the effectiveness of combining multivariate tools and data fusion to improve the interpretability of complex environmental datasets. The methodology used in this study is fully generalizable and applicable to other environmental multi-way datasets.