Unveiling Cu extractability and statistical insights: Hydrocotyle umbellata L. as a catalyst for environmental remediation.

The key objective of the research was to investigate the potential of Hydrocotyle umbellata L. as a hyperaccumulator in copper (Cu)-contaminated environments and to enhance the understanding of its phytoextraction efficiency through the application of unsupervised machine learning techniques alongside statistical comparisons. The effects of Cu toxicity on pigment content, total flavonoids, total phenolic content, electrolyte leakage, translocation, and bio-concentration factors were analyzed in H. umbellata L. using analysis of variance (ANOVA), paired t-tests, and correlation analysis. Machine learning (ML) was applied to various experimental outputs of H. umbellate L. after Cu phytoextraction. The ML techniques included cluster analysis and classification and regression tree (CART). There were 48 samples available for the clustering analysis, with three variables (TF observations, plant parts, and treatment levels). Results indicated that the highest metal uptake was by the roots, with a TF value of 1.114, making the plant appropriate for Cu phytoextraction. TF emerged as the most crucial, followed closely by chlorophyll and carotenoid content, total flavonoid content, total phenolic content, leakage ratio, fresh weight, and dry weight. Notably, our analysis suggested that bioaccumulation factor (BCF) may not be a reliable indicator for assessing Cu uptake within the specific context of this investigation. This study represents one of the first attempts to show the effects of Cu toxicity on physiology, biochemical compounds, and leakage ratio, along with BCF and TF in H. umbellata L. Moreover, new insights from ML model interpretation, alongside statistical models, could guide effective phytoremediation by identifying the phytoextraction ability of H. umbellata L.