Leveraging local species data, a global database, and an occupancy model to explore bee-plant interactions.

Area of Science:

• Ecology
• Conservation Biology
• Bioinformatics

Background:

• Global bee declines threaten pollination services and ecosystem stability.
• Understudied bee-plant interactions create knowledge gaps in ecosystem vulnerability.
• Large-scale data aggregation from online databases (e.g., GloBI) offers new research opportunities.

Purpose of the Study:

• To investigate factors influencing bee-plant interactions and detection probabilities.
• To compare interaction networks derived from raw data versus occupancy models.
• To test hypotheses on bee size, sociality, floral traits, and data collection methods.

Main Methods:

• Utilized the Global Biotic Interactions (GloBI) database and curated species checklists.
• Employed an occupancy model to account for imperfect detection in bee-plant interactions.
• Compared interaction networks generated from raw data and occupancy model outputs.

Main Results:

• Smaller bees showed higher interaction probabilities, contradicting the hypothesis on bee size.
• Blue and closed-shaped flowers had higher interaction probabilities.
• Larger bee size, blue flowers, bowl shapes, and community science data increased detection probability.
• Occupancy model networks exhibited higher evenness, nestedness, and connectance than raw data networks.

Conclusions:

• Detection biases significantly influence our understanding of bee-plant interactions.
• Occupancy modeling provides a more comprehensive view of interaction networks.
• Community science data is valuable for detecting ecological interactions, but biases must be considered.