Uneven research attention to biodiversity loss revealed by large language models
- 1LMU Munich, Munich Center for Machine Learning
Abstract
Biological diversity is crucial for ecosystem functioning and human well-being but is declining under multiple pressures from human activity. Research on biodiversity loss spans multiple ecological contexts, yet existing syntheses typically cover only subsets of the evidence base. We developed a systematic mapping framework based on large language models to screen and synthesize the scientific literature on biodiversity loss at scale. Applying our framework to 2.4 million bibliographic records, we identified 253,081 scientific articles on biodiversity loss, which we classified across five dimensions: geography, ecosystems, taxa, direct drivers (e.g., land- and sea-use change, climate change, and pollution), and specific threats (e.g., agriculture, hunting, and urban development). Chordates accounted for 34.8% of research attention but only 4.6% of described species. We find comparatively limited research attention to non-chordate animals, studies in Africa, and direct exploitation in marine systems. Overall, our systematic map provides a multidimensional view of research attention to biodiversity loss across various ecological contexts and can help identify evidence gaps to inform future research priorities.
Related
- BIRDS: Characterizing and Understanding Biodiversity Impact of Large Language Model Serving
- BioCosmos: computer vision and large language models for large-scale image-based biodiversity data and trait exploration
- Maturational constraints on tracking of temporal attention in infant language acquisition.
- Spatiotemporal dynamics of lexical and morphosyntactic planning during language production revealed by magnetoencephalography
- The ADHD advantage: Comorbid attention-deficit/hyperactivity disorder is associated with earlier diagnosis and better academic outcomes in children with specific language disorder
BibTeX
@misc{mohiuddin2026uneven,
title = {{Uneven research attention to biodiversity loss revealed by large language models}},
author = {H. Mohiuddin and K. Forster and S. Feuerriegel},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.23.761449},
url = {https://www.biorxiv.org/content/10.64898/2026.10.23.761449v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Uneven research attention to biodiversity loss revealed by large language models
AU - H. Mohiuddin
AU - K. Forster
AU - S. Feuerriegel
PY - 2026
DA - 2026/10/23
PB - bioRxiv
DO - 10.64898/2026.10.23.761449
UR - https://www.biorxiv.org/content/10.64898/2026.10.23.761449v1
AB - Biological diversity is crucial for ecosystem functioning and human well-being but is declining under multiple pressures from human activity. Research on biodiversity loss spans multiple ecological contexts, yet existing syntheses typically cover only subsets of the evidence base. We developed a systematic mapping framework based on large language models to screen and synthesize the scientific literature on biodiversity loss at scale. Applying our framework to 2.4 million bibliographic records, we identified 253,081 scientific articles on biodiversity loss, which we classified across five dimensions: geography, ecosystems, taxa, direct drivers (e.g., land- and sea-use change, climate change, and pollution), and specific threats (e.g., agriculture, hunting, and urban development). Chordates accounted for 34.8% of research attention but only 4.6% of described species. We find comparatively limited research attention to non-chordate animals, studies in Africa, and direct exploitation in marine systems. Overall, our systematic map provides a multidimensional view of research attention to biodiversity loss across various ecological contexts and can help identify evidence gaps to inform future research priorities.
ER -