Imperfect detection biases β diversity estimates based on whether species are shared between sites

Area of Science:

• Ecology
• Biodiversity Science
• Conservation Biology

Background:

• Imperfect detection is a known source of sampling error affecting species occurrence and diversity measures (α, γ).
• Its impact on beta diversity (β), especially when considering species traits and evolutionary history, is less understood.
• Beta diversity quantifies species composition variation among sites.

Purpose of the Study:

• To quantify the impact of imperfect detection on taxonomic, functional, and phylogenetic β diversity.
• To investigate how imperfect detection influences relationships between β diversity and island attributes (area, isolation).
• To develop a general framework for understanding and correcting detection bias in β diversity.

Main Methods:

• Applied occupancy models to avian monitoring data from 36 subtropical islands over a decade.
• Quantified taxonomic, functional, and phylogenetic β diversity, correcting for imperfect detection.
• Conducted simulations with varying ecological and sampling drivers to test broader applicability.

Main Results:

• Imperfect detection consistently overestimated taxonomic, functional, and phylogenetic β diversity in both empirical and simulated data.
• Under-detection of shared species was the primary driver of this overestimation, obscuring diversity-attribute relationships.
• In empirical data, bias increased with island area differences; simulations showed repeated surveys reduced bias.

Conclusions:

• Imperfect detection systematically biases all facets of β diversity by altering observed species composition.
• Accurate quantification of β diversity, particularly functional and phylogenetic aspects, is essential for ecological research.
• The findings provide a general framework applicable across taxa and ecosystems for improving biodiversity measurements.