Dynamical assessment of aboveground and underground biodiversity with supportive AI

Abstract

Ecosystems are complex open systems, and ecological studies have been developing through extensive approaches taking the newly measurable variables into account. Here we explore a dynamical framework based on the interactive expansion of models in environmental assessment supported by artificial intelligence, and demonstrate examples for both aboveground and underground biodiversity assessment in the Synecoculture projects as a typical scenario. We first applied statistical modeling to obtain the scoring of index species with respect to the productivity and functionality of the field ecosystems, which incorporate the dynamical changes of species composition in real-time management and succession. Secondly, we applied machine learning to the estimation and validation of soil microbiological diversity and activity based on the aboveground plant species diversity and other physicochemical soil parameters, through interactive parameter selection and dynamic reconfiguration of deep learning neural network structure. Finally, we simulated the augmentation of ecosystem services in Meiji-Jingu precincts based on the suggestion of introducible species from ecological interaction networks. The framework of interpretation towards a comprehensive evaluation of these processes is discussed.

Keywords

• Environmental assessment
• Aboveground and underground biodiversity
• Artificial intelligence
• Statistical model
• Machine learning
• Deep learning neural network