SAGE: A sampling-aware global evaluation benchmark for species distribution modeling
Abstract
Knowing where species occur is fundamental for biodiversity research and conservation. Species distribution models (SDMs) link species observations to environmental conditions to estimate their spatial distribution. However, accuracy varies with the underlying data and models, making it essential to know for which species models can be trusted. Deep-learning-based SDMs ("DeepSDMs") now jointly model thousands of species, drawing on hundreds of millions of community-science records. At this scale, averaging performance hides substantial species-level variability, particularly for rare species, often of greatest conservation concern. Records are also strongly biased, making occurrence counts misleading. Accounting for these factors is essential for a reliable and informative evaluation of multi-species SDMs. Here, we introduce a Sampling-Aware Global Evaluation (SAGE) benchmark, combining GBIF records for training with sPlotOpen vegetation plots for presence-absence evaluation across 5771 plant species. We propose an evaluation framework that groups species based on two properties, sampling effort and relative prevalence, which describe how densely a species' range is sampled and how frequently the species is recorded. Evaluating single-species SDMs and multi-species DeepSDMs, we find that Random Forests and DeepSDMs perform best overall, but neither dominates: DeepSDMs outperform single-species SDMs for infrequently recorded species while offering no consistent advantage for well-sampled ones. Crucially, this advantage emerges only when established bias-correction practices, such as spatial thinning and reweighting, are carried over to the deep-learning setting. SAGE helps identify the species and data conditions for which a given approach is beneficial, thereby supporting the development of more transparent and ecologically credible SDMs. Data and code: https://earens.github.io/sage/
Related
- Genetic Perturbation Prediction Models Fail Out-of-Distribution: A Rigorous Benchmark and Systematic Evaluation
- STEM-LM: Spatio-Temporal Ecological Modeling via Masked Language Model for Multi-Species Distribution
- Quantification of total fish eDNA concentration to optimize the detection of rare species and sampling effort
- Protein novelty, sequence similarity, and negative sampling controls reveal benchmark specific discrimination in protein protein interactionprediction
- OmniEEG-Bench: A Standardized Evaluation Benchmark for EEG Foundation Models
BibTeX
@misc{arens2026sage,
title = {{SAGE: A sampling-aware global evaluation benchmark for species distribution modeling}},
author = {Emilia Arens and Nina van Tiel and Robin Zbinden and Damien Robert and Lukas Drees and Chiara Vanalli and Benjamin Kellenberger and Niklaus E. Zimmermann and Loïc Pellissier and Devis Tuia and Jan Dirk Wegner},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.31082},
eprint = {2609.31082},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.31082},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - SAGE: A sampling-aware global evaluation benchmark for species distribution modeling
AU - Emilia Arens
AU - Nina van Tiel
AU - Robin Zbinden
AU - Damien Robert
AU - Lukas Drees
AU - Chiara Vanalli
AU - Benjamin Kellenberger
AU - Niklaus E. Zimmermann
AU - Loïc Pellissier
AU - Devis Tuia
AU - Jan Dirk Wegner
PY - 2026
DA - 2026/09/25
PB - arXiv
DO - 10.48550/arxiv.2609.31082
UR - https://arxiv.org/abs/2609.31082
AB - Knowing where species occur is fundamental for biodiversity research and conservation. Species distribution models (SDMs) link species observations to environmental conditions to estimate their spatial distribution. However, accuracy varies with the underlying data and models, making it essential to know for which species models can be trusted. Deep-learning-based SDMs ("DeepSDMs") now jointly model thousands of species, drawing on hundreds of millions of community-science records. At this scale, averaging performance hides substantial species-level variability, particularly for rare species, often of greatest conservation concern. Records are also strongly biased, making occurrence counts misleading. Accounting for these factors is essential for a reliable and informative evaluation of multi-species SDMs. Here, we introduce a Sampling-Aware Global Evaluation (SAGE) benchmark, combining GBIF records for training with sPlotOpen vegetation plots for presence-absence evaluation across 5771 plant species. We propose an evaluation framework that groups species based on two properties, sampling effort and relative prevalence, which describe how densely a species' range is sampled and how frequently the species is recorded. Evaluating single-species SDMs and multi-species DeepSDMs, we find that Random Forests and DeepSDMs perform best overall, but neither dominates: DeepSDMs outperform single-species SDMs for infrequently recorded species while offering no consistent advantage for well-sampled ones. Crucially, this advantage emerges only when established bias-correction practices, such as spatial thinning and reweighting, are carried over to the deep-learning setting. SAGE helps identify the species and data conditions for which a given approach is beneficial, thereby supporting the development of more transparent and ecologically credible SDMs. Data and code: https://earens.github.io/sage/
ER -