Observer-oriented approach improves species distribution models from citizen science data

Area of Science:

• Ecology
• Conservation Biology
• Biodiversity Informatics

Background:

• Citizen science platforms generate vast species location data valuable for conservation.
• Limited information on surveyed sites and sampling effort hinders the use of citizen science data for species distribution modeling.
• Traditional methods often lack robust surrogates for sampling effort, impacting model accuracy.

Purpose of the Study:

• To evaluate the effectiveness of an observer-oriented approach for generating pseudo-absences and sampling effort proxies in species distribution models.
• To compare the predictive accuracy of models using observer-oriented pseudo-absences versus random pseudo-absences.
• To enhance the utility of citizen science data for accurate species distribution modeling and conservation planning.

Main Methods:

• An observer-oriented approach was developed, using occurrences of non-target species as pseudo-absences and observer activity as sampling effort proxies.
• Species distribution models (SDMs) were constructed for 15 Italian mammal species using ensemble predictions from nine SDM algorithms.
• Model performance was assessed via cross-validation, comparing predictions derived from observer-oriented pseudo-absences against those using random pseudo-absences.

Main Results:

• The observer-oriented approach significantly improved the predictive accuracy of species distribution models compared to using random pseudo-absences.
• Models incorporating observer-derived pseudo-absences and sampling effort proxies demonstrated higher predictive power.
• This approach effectively addressed limitations associated with missing data on sampling effort in citizen science datasets.

Conclusions:

• Citizen science data, when processed with an observer-oriented approach, can yield robust species distribution models.
• This methodology enhances the capacity to accurately predict species geographic ranges, supporting effective conservation strategies.
• The study highlights the importance of accounting for sampling effort and observer bias when utilizing citizen science data for ecological modeling.