In GBIF records of 116 bumblebee species, 1901–2025, sampling effort explains most raw range-limit shifts; a 28–74 km southern retreat remains in North America only

Kerr et al. (2015) reported that bumblebee southern range limits in North America and Europe retreated by up to about 300 km and that northern limits did not expand. Their metric is the mean latitude of the 5 most extreme records. We tested this metric with GBIF record counts for 116 Bombus species (ad-hoc map API, 0.35 x 0.18 degree cells, 3.9 million species records) in the Kerr periods and in 2011-2017 and 2018-2025, relative to 1901-1974. For each species we calculated an effort-only null: within 1-degree cells, the species keeps its share of all Bombus records, while the amount and location of Bombus sampling change as observed. The effort-adjusted shift is the observed shift minus the null mean. The null predicts most raw shifts. In North America the raw southern shift to 1999-2010 is 149 km (median across species), the null 107 km and the adjusted shift 28 km (95% species-bootstrap interval 2 to 98 km). For 2018-2025 the adjusted southern shift is 74 km (15 to 136 km) with all records and 23 km (6 to 72 km) with specimen records only. In Europe the adjusted southern shifts are 5 to 6 km and their intervals include zero. Adjusted northern shifts are 0 km in North America and 13 to 17 km in Europe, while June-August isotherms moved 150 to 290 km. In simulated species with the real effort, the raw metric gave false median shifts from -137 to +78 km; the adjusted estimator was within 4 km of zero and recovered 86 to 106 km of a 117 km southern shift, but only 52 to 56 km (Europe) and 0 to 3 km (North America) of a northern shift. Thus North American northern-limit change cannot be tested with these records. We also found that GBIF density-map species layers hold a median of 53% of the records in the search index, with spatially uneven deficits; we did not use them for the results. Limitations: opportunistic records, identification and habitat changes inside cells, GBIF backbone taxonomy, latitude only, and too-narrow per-species p-values.

bumblebees · Bombus · range shift · range limits · climate change · sampling effort · GBIF · null model · species distributions · museum specimens · community science · North America · Europe · isotherm tracking · reproducibility