Using historical data to estimate bumble bee occurrence: Variable trends across species provide little support for community-level declines
2021
Abstract Bumble bees are globally important pollinators, especially in temperate regions, and evidence suggests that many species are declining. One recent high profile study by Soroye et al. (2020) applied occupancy models to dated historical collection data to quantify declines across North America and Europe. The authors modelled 66 species across a set of sites spanning both North America and Europe, rather than confining species to sites where they might be expected to occur. In addition, they inferred non-detections for time intervals where there is no evidence that the site was visited (by forcing every site to have exactly 3 visits in each era). We use simulated data to (i) investigate the validity of methods used in that study and (ii) test whether a multi-species framework that incorporates species' ranges and site visitation histories produces better estimates. We show that the method used by Soroye et al. (2020) yields biased estimates of declines, whereas our framework does not. We use such a model to provide revised and appreciably lower estimates for bumble bee community declines, with species-specific trends more closely matching classifications from IUCN. The species level trends we provide can help inform future species-at-risk assessments. Well-parameterized occupancy models may be a powerful tool for assessing species-wide trends using curated historical collection data.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
56
References
6
Citations
NaN
KQI