1. Species exhibit a range of specialisation in diet and other niche axes, with specialists typically thought to be more efficient in resource use but more vulnerable to extinction than generalists. Among herbivorous insects, dietary specialists seem more likely to lack acceptable host plants during the insect’s feeding stage, owing to fluctuations in host-plant abundance or phenology. Like other herbivores, bee species vary in host breadth from pollen specialisation (oligolecty) to generalisation (polylecty).

2. Several studies have shown greater interannual variation in flowering phenology for earlier-flowering plants than later-flowering plants, suggesting that early-season bees may experience substantial year-to-year variation in the floral taxa available to them.

3. We therefore reasoned that, among bees, early phenology could be a more viable strategy for generalists, which can use resources from multiple floral taxa, than for specialists. Consequently, we expected median dates of collection of adult specimens to be earlier for generalist species than for specialists. To test this, we obtained phenology data and pollen diet information on 67 North American species of the bee genus Osmia.

4. Controlling for latitude and phylogeny, we find that dietary generalisation is associated with significantly earlier phenology, with generalists active, on average, 11–14 days earlier than specialists.

5. This result is consistent with the generalist strategy being more viable than the specialist strategy for species active in early spring, suggesting that dietary specialisation may constrain the evolution of bee phenology—or vice versa.

lectydata.csv: Contains data on lecty and host-plant family for oligoleges for each of the 67 Osmia species used in our analysis. Lecty was determined from an exhaustive search of the literature combined with solicitation of expert opinion, Krombein's Catalog of Hymenoptera in America North of Mexico, and Fowler and Fowler & Droege's online databases of North American specialist bees. All lecty searching was done blind to phenology data. Classifications we deemed to be based on biased evidence (i.e., surveys of visitors to single plant taxa) were excluded. To stay consistent with the classifications used in most of the literature, monolectic species were classified as oligoleges, and eclectic oligoleges and mesoleges were classified as polylectic.

occurrencedata.txt: Contains the dataset of Osmia bee occurrences that we used in our analyses, post cleaning (34 098 occurrences, one row per occurrence). Reference for original (pre-cleaning) dataset: GBIF. GBIF Occurrence Download [Internet]. 2021 [cited 2021 Feb 19]. Available from: https://doi.org/10.15468/dl.z2rupk. We downloaded the raw data from GBIF, including dates of capture, coordinates, and species names, for all Osmia bees between the latitudes of 45°N and 35°N and between the longitudes of 125°W and 60°W (70 126 total occurrences).

R_code.pdf: all R code used to conduct our analyses. This pdf was made by knitting to html with Rmarkdown in R version 4.2.1.

Only R and the lectydata.csv and occurrencedata.txt files are needed to repeat all of our analyses using the code in R_code.pdf. Descriptions of all columns in the csv and txt files are all in the README.md document.