May 2026· arXiv.org· Vol abs/2605.27480· 0 citations· 55 references
BiologyComputer Science
TL;DR
BIRDS defines request-level functional units, quantifies operational and embodied biodiversity impact, and introduces Quality-Normalized Biodiversity Impact (QNBI) to jointly analyze ecological impact and response quality.
Abstract
Large language model (LLM) serving creates environmental impacts beyond carbon and water, including ecosystem damage through biodiversity-related pathways. We present BIRDS, a framework for Biodiversity Impact of Request-Driven LLM Serving. BIRDS defines request-level functional units, quantifies operational and embodied biodiversity impact, and introduces Quality-Normalized Biodiversity Impact (QNBI) to jointly analyze ecological impact and response quality. Across diverse workloads, models, GPUs, and regions, BIRDS reveals that biodiversity impact accumulates at scale and exposes quality-aware serving tradeoffs. The code is available at https://github.com/TianyaoShi/BIRDS.
This study presents a TRL-4 prototype that integrates a configurable Digital Twin core with a generative AI conversational interface for conservation-oriented modeling in Doñana National Park, Spain, a UNESCO World Heritage site facing significant environmental challenges.
Pablo Vicente-Martínez, Adrián Chust-Ros, Ismerai David Gutiérrez-Rodríguez et al.· Environments· 0 citations
Invasive alien species are widely recognised as a major driver of biodiversity loss and ecosystem change. Their ecological impacts are expressed as measurable changes in ecosystem properties across levels of organisation, from genes to ecosystems, and vary in both direction and magnitude depending on social-ecological...
Eva Malta-Pinto, Ingolf Kühn, H. Roy et al.· ARPHA Conference Abstracts· 0 citations
Abstract Large-scale online biodiversity databases, particularly the Global Biodiversity Information Facility (GBIF), have become indispensable resources for macroecology, biogeography, and conservation biology. Their accessibility and extensive coverage have promoted widespread application in species distribution mode...
F. Petrozzi, L. Luiselli· Journal of Tropical Ecology· 1 citation
A Sampling-Aware Global Evaluation (SAGE) benchmark is introduced, combining GBIF records for training with sPlotOpen vegetation plots for presence-absence evaluation across 5771 plant species, and an evaluation framework that groups species based on two properties, sampling effort and relative prevalence is proposed.
Emilia Arens, Nina Van Tiel, Robin Zbinden et al.· 0 citations
Evaluating ecological time series is critical for benchmarking model performance in many important applications, including predicting greenhouse gas fluxes, tracking soil moisture and soil–atmosphere interactions, and monitoring hydrological cycles. Traditional numerical metrics (e.g., R-squared, root mean square error...
Qi Cheng, Licheng Liu, Yi-Xuan Chen et al.· Proceedings of the 32nd ACM...· 3 citations
The negative impact of alien species is recognised as a major threat to biodiversity. An impact indicator, a measurable proxy of the status of alien species impacts over space and time, is essential for informing biological invasion policies and management strategies. We introduce an open-source workflow for computin...
M. M. Yahaya, S. Kumschick, S. MacFadyen et al.· NeoBiota· 0 citations
