EUVIMEDEUVIMED entdecken
Crossref

Spatiotemporal Prediction of River Fish Biodiversity by Multimodal Deep Learning of eDNA and Satellite Imagery

Song Zhang iD, Xiaowei Zhang, Wenjun Zhong iD, Yifan Wang, Shuo Zong, Loïc Pellissier

Environmental Science & Technology · 18.08.2026 · 60 (32) · 22641-22652

DOI: 10.1021/acs.est.6c05038 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Environmental Science & Technology, online erschienen am 06.08.2026. Verfasst von Song Zhang und 5 weiteren Autor:innen. Beteiligte Einrichtung: State Key Laboratory of Pollution Control & Resource Reuse, School of the Environment, Nanjing University, Nanjing210023, China. Hauptthemen laut MEDLINE: Deep Learning, Biodiversity, Rivers, Fishes. Die Arbeit nennt 76 Literaturangaben. Gefördert von: China Scholarship Council, National Key Research and Development Program of China.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

High-resolution spatiotemporal biodiversity maps are essential for conservation and environmental management of freshwater ecosystems. Yet existing eDNA-based approaches remain largely static and spatially extrapolative, discarding fine-grained landscape structure and overlooking interannual dynamics. Here, we developed a multimodal deep learning framework that integrates raw satellite imagery with environmental tabular covariates to predict fish species richness across river networks, using eDNA-derived richness from 1302 sites across six Chinese river basins as training labels. The framework couples a U-Net encoder–decoder with a novel water-guided spatial attention mechanism that leverages MNDWI-derived water masks to prioritize aquatic and riparian pixels, and an attention autoencoder that compresses high-dimensional environmental variables into a latent representation. Ablation experiments confirmed that multimodal fusion of imagery and tabular data substantially outperformed single-modality baselines (R2 from 0.54 to 0.76), and that the water-guided attention mechanism yielded further gains (R2 = 0.81), demonstrating that raw imagery preserves ecologically critical spatial information lost by conventional buffer-based summarization. The model was validated on fully independent basins (middle Yangtze and Yuan River) and multiyear field surveys (2018–2023). It achieved strong spatial transferability (r = 0.83 and 0.69) without region-specific retraining, and captured interannual recovery trends following the Yangtze fishing ban without any direct policy input. By shifting from static spatial extrapolation to prediction that generalizes across space and time, this framework offers a scalable pathway for freshwater biodiversity monitoring and conservation evaluation under the Kunming–Montreal Global Biodiversity Framework.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningBiodiversityRiversFishesSatellite ImageryAnimalsEnvironmental MonitoringDNA, Environmental

Stichwörter der Autor:innen

biodiversity monitoringconservation evaluationenvironmental DNAmultimodal deep learningremote sensingspatiotemporal prediction

Substanzen

DNA, Environmental

Autor:innen und Institutionen

  • State Key Laboratory of Pollution Control & Resource Reuse, School of the Environment, Nanjing University, Nanjing210023, China.
  • Ecosystems and Landscape Evolution, Department of Environmental Systems Science, ETH Zürich, 8092Zürich, Switzerland.
  • Ecosystems and Landscape Evolution, Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), Birmensdorf8903, Switzerland.
  • State Key Laboratory of Vegetation Structure, Function and Construction (VegLab), Yunnan Key Laboratory for Plateau Mountain Ecology and Restoration of Degraded Environments, School of Ecology and Environmental Science,…
  • School of Earth Sciences, Zhejiang University, 310027Hangzhou, China.

ORCID-Profile: Song Zhang, Wenjun Zhong

Förderung

  • China Scholarship Council
  • National Key Research and Development Program of China

Zitieren

APA

Zhang, S., Zhang, X., Zhong, W., Wang, Y., Zong, S., & Pellissier, L. (2026). Spatiotemporal Prediction of River Fish Biodiversity by Multimodal Deep Learning of eDNA and Satellite Imagery. Environmental Science & Technology, 60(32), 22641-22652. https://doi.org/10.1021/acs.est.6c05038

Vancouver

Zhang S, Zhang X, Zhong W, Wang Y, Zong S, Pellissier L. Spatiotemporal Prediction of River Fish Biodiversity by Multimodal Deep Learning of eDNA and Satellite Imagery. Environmental Science & Technology. 2026;60(32):22641-22652. doi:10.1021/acs.est.6c05038

BibTeX
@article{Zhang2026,
  author = {Zhang, Song and Zhang, Xiaowei and Zhong, Wenjun and Wang, Yifan and Zong, Shuo and Pellissier, Loïc},
  title = {Spatiotemporal Prediction of River Fish Biodiversity by Multimodal Deep Learning of eDNA and Satellite Imagery},
  journal = {Environmental Science & Technology},
  year = {2026},
  volume = {60},
  number = {32},
  pages = {22641--22652},
  doi = {10.1021/acs.est.6c05038}
}
Kostenlos starten →