Efficient occupancy model-fitting for extensive citizen-science data.

Area of Science:

• Ecology and Biodiversity
• Computational Biology
• Environmental Science

Background:

• Large-scale citizen science data offer extensive spatial coverage for biodiversity modeling.
• Existing occupancy modeling approaches often use random effects, necessitating time-consuming Bayesian analysis.
• This is particularly challenging for annual modeling of numerous species, such as British butterflies.

Purpose of the Study:

• To present an efficient alternative to complex occupancy modeling for citizen science data.
• To utilize logistic regression with environmental covariates for classical inference-based model fitting.
• To enable faster and more accessible biodiversity trend analysis and distribution mapping.

Main Methods:

• Developed a logistic regression approach to model site variation using environmental covariates.
• Applied classical inference for efficient occupancy model fitting on presence-only data.
• Validated the approach using both real (UK Butterflies) and simulated datasets.

Main Results:

• The classical inference approach yielded similar conclusions on trends compared to random effects models.
• Classical model-fitting facilitates comparison of alternative models, covariate identification, and model fit assessment.
• Enabled the construction of regional occupancy indices and dynamic occupancy maps for species distribution.

Conclusions:

• Classical inference provides an efficient and accessible method for analyzing citizen science biodiversity data.
• This approach aids in understanding ecological processes and monitoring species' responses to environmental change.
• The methods offer valuable tools for biodiversity conservation, particularly in the context of climate change, and can engage citizen scientists.