Insights into Zooplankton Abundance Dynamics in Tropical Temporary Ponds using Machine Learning and Explainable AI

Determining the health and resilience of ecosystems depends on an understanding of the dynamics of zooplankton abundance in tropical temporary ponds. Using ensemble modelling and explainable artificial intelligence (XAI) techniques, this paper explores the complex dynamics of zooplankton abundance in tropical temporary ponds. This study uses a Machine Learning framework to evaluate the predictive performance of models such as k-nearest Neighbors ($k N N$) regression, Random Forest Regression, Lasso Regression, Passive Aggressive Regression, Ridge Regression, and Ensemble Voting Regression. Among the machine learning models tested, the ensemble model, which combines ridge and ridge CV, performs best, with metrics such as MAE of 0.02, MSE of 0.00, and $\mathbf{R 2}$ of 1.00. The Shapley Additive ExPlanations (SHAP) analysis identifies ‘Rotifers’, ‘Culex’, ‘Chironomodidae’, and ‘Calanoids’ as significant predictors. Data preprocessing techniques improve the dataset derived from a study of fish predation effects in tropical temporary ponds. Our research advances our understanding of zooplankton ecology and provides insights into conservation strategies and management practices in these ecosystems. Using ensemble modelling and explainable AI, we contribute to developing accurate predictive models for the long-term management of tropical temporary pond ecosystems.