Mangroves are valuable ecological resource providing habitat for many species. Species assemblages vary in life history strategies and spatial and temporal scales, thus recording real diversity using traditional methods is almost impossible. Here, we aim to predict species interaction networks combining eDNA and machine learning to provide insights in ecosystem function and emergent properties, a fundamental information to more appropriate conservation strategies and restore ecosystem services.