Biases in raw opportunistic observation data may produce unreliable richness estimates

Study evaluates effects of inclusion criteria on measures of α-diversity and β-diversity when using citizen science data

GBIF-mediated data resources used : 1,184,984 species occurrences
Tadorna tadorna
Tadorna tadorna (Linnaeus, 1758) observed in Sweden by Ulf Teghammar (CC BY-NC 4.0)

Measures of biodiversity are crucial to understanding community assembly and for conservation planning. Separating transients from core species in standardized surveys is possible by applying specific inclusion criteria, but with abundant citizen science observation data available, when do we consider a species part of a local community?

Extracting occurrences published by Artportalen (Swedish Species Observation System) of 77 bird species during a 90-day breeding season at 107 frequently visited wetland sites in Sweden, this study examined the effects of varying species inclusion criteria on measures of α-diversity (species richness) and β-diversity (community dissimilarity) when relying on raw high-density opportunistic observational data vs corrected estimates using a site-use occupancy model.

Applying thresholds of 1–30 days for inclusion, they found both α and β diversity to be highly sensitive to the criteria used. Estimates of richness generated from raw data, however, were consistently lower than when based on occupancy models. The effect was even more pronounced under the criteria of consecutive days.

These results suggest that occupancy model estimates are more stable and accurate and that, despite its abundance, raw opportunistic observation data may not produce reliable local richness estimates.

Ruete A, Arlt D, Berg Å, Knape J, Żmihorski M and Pärt T (2020) Cannot see the diversity for all the species: Evaluating inclusion criteria for local species lists when using abundant citizen science data. Ecology and Evolution. Wiley 10(18): 10057–10065. Available at: https://doi.org/10.1002/ece3.6665