Assessing Distribution Patterns and Prediction of Metals in Wildlife Sanctuary by Employing Lichen, Expanding Machine Learning Techniques

The national parks and wildlife sanctuaries are the protected areas for the conservation of wildlife and act as policy tools to sustain biodiversity protection and sustainable application of resources. In the present study the dissemination, qualitative and quantitative valuation of metals at 80 sampling sites of Govind Wildlife Sanctuary (GWLS) were evaluated for the first time by employing lichen Heterodermia diademata. The mean concentration of As, Al, Cd, Cr, Cu, Fe, Hg, Mn, Ni, Pb and Zn varied from 0.78–12.05, 5.15–79.42, 0.09–0.84, 4.26–295.31, 7.09–63.47, 12.56–418.90, 0.11–0.99, 0.45–16.85, 1.06–25.66, 0.54–9.07, 6.49–209.18 μg/g dw in the study area. The seven diverse advanced machine learning algorithms were applied to determine precise predictive models for different metals with respect to altitude in the area. The Ridge Regression and Random Forest Regression models displayed the superior linearity between predictions and observations than other models. The Pearson’s correlation coefficient and root mean square error and effects of most influential metal in different altitudes using most effective machine learning algorithm have been studied here. The most influential metals in altered altitudes can be predicted using various machine learning models. In addition, the study provide baseline evidence on the range of metal contamination at different altitudes and emphasised the need to prevention of pollution in GWLS further.