Declining Collections and ‘Collecting Smarter’: Data-Driven Approaches to Sampling Design and Research Prioritisation in Ecology
Abstract
Biological collections, holding over two billion specimens, provide a unique spatiotemporal record of Earth's biodiversity. Over the last 20 years, digitisation efforts and platforms like the Global Biodiversity Information Facility (GBIF) have enhanced the accessibility and value of these collections. However, our analysis reveals significant declines in specimen data availability over the past 40 years, varying across taxonomic groups and geographical regions. This trend threatens to erode the utility of biological collections as global research infrastructure.
Resource constraints necessitate a ‘collect smarter’ approach for sampling design and allocation of collecting effort in biodiversity research. To improve our ability to understand spatiotemporal trends in ecological phenomena and respond to environmental crises, we propose a data-driven framework for research operational strategy for collections science. Implementing a systematic approach for resource allocation is crucial in the face of mounting environmental crises, especially as specimen data expand their relevance beyond taxonomy to fields like environmental monitoring, agriculture, biosecurity, and public health.
Using Australian specimen-based data from GBIF, we demonstrate the application of Bayesian decision theoretic principles and optimal design tools to create a prototype interactive sampling design platform for collections scientists. This platform combines spatial value of information (VoI) metrics with derived indices for spatial distribution of ecological threats, introducing a 'need for information' (NfI) measure. By integrating these approaches, we aim to enhance ecological sampling strategies and optimise the impact-to-effort ratio in addressing urgent environmental challenges.