Multimodal data integration to model, predict, and understand changes in plant biodiversity: a systematic review

Résumé fourni par la source

The integration of multimodal data to analyze, model, and predict changes in plant biodiversity is critical for addressing global conservation challenges. This systematic review examines the current landscape of plant biodiversity data, focusing on the identification, classification, and evaluation of key open-access data sources and integration methodologies. We highlight the strengths and limitations of major biodiversity platforms, emphasizing their contributions to species occurrence, trait data, taxonomic checklists, and environmental variables. The review also explores computational tools for data integration. We describe and analyze the role of Darwin Core standards in data standardization, harmonization, and interoperability, highlighting the importance of tools such as Species Distribution Models and machine learning. Additionally, we assess the tools available for multimodal data integration and analysis of the effects of environmental drivers (e.g., temperature, precipitation, topography) on biodiversity. We find significant advancements in biodiversity informatics over the last decades. Still, challenges persist in achieving interoperability across datasets, in addressing spatial and temporal biases, and in integrating remote sensing with in situ observations. By identifying both the challenges and emerging solutions, this review contributes to advancing biodiversity monitoring strategies, aligning with global conservation goals outlined by the Convention on Biological Diversity and the United Nations Sustainable Development Goal 15. Ultimately, the findings underscore the importance of harmonized data integration frameworks to enhance predictive modeling capabilities and inform effective conservation policies. • Reviews 12 biodiversity platforms and their data integration strategies. • Assesses data quality, coverage, and interoperability for ecological modeling. • Explores SDMs and deep learning in biodiversity prediction. • Identifies gaps in spatial-temporal coverage and the role of remote sensing data. • Proposes multimodal data integration like a novel approach for ecological analysis.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.