Leveraging Massive Opportunistically Collected Datasets to Study Species Communities in Space and Time

Area of Science:

• Ecology
• Biodiversity Science
• Statistical Modeling

Background:

• Online portals enable extensive biodiversity data collection by naturalists.
• Opportunistic data present challenges for ecological inference due to detection biases.
• Existing methods struggle to fully utilize heterogeneous, widely available biodiversity datasets.

Purpose of the Study:

• To develop a novel statistical approach for analyzing opportunistic biodiversity data.
• To overcome limitations of variable detection probabilities and observer behavior in ecological datasets.
• To provide high-resolution insights into biological communities from large-scale, naturalist-collected data.

Main Methods:

• Embedding a spatiotemporal joint species distribution model within a site-occupancy framework.
• Modeling phenological patterns to address variable detection probabilities across visits and species.
• Utilizing latent variables to characterize observer-specific detection and reporting behavior.

Main Results:

• The model successfully analyzes over 100,000 opportunistic records of lentic odonates in Flanders.
• High-resolution insights into community phenology, interannual trends, and environmental associations were obtained.
• Spatiotemporal co-distributional patterns in community composition were revealed.

Conclusions:

• The novel statistical approach effectively leverages opportunistic biodiversity data for ecological research.
• The model enhances our understanding of biological communities, including their dynamics and environmental drivers.
• This method opens new avenues for utilizing large-scale, citizen-science biodiversity datasets.