Flowering phenology of herbarium specimens reveals century-scale shifts in allergenic pollen seasonality

Data to support our paper, "Flowering phenology of herbarium specimens reveals century-scale shifts in allergenic pollen seasonality ". The CSV file contains a list of GBIF IDs representing herbarium specimens within 100km of a pollen monitoring site, and our flowering phenology model's prediction for whether or not the specimen is in flower. This dataset is derived in large part from data published with Williamson et al., 2025, (https://doi.org/10.1111/nph.70139) the supporting data for which is also available on Zenodo here: https://doi.org/10.5281/zenodo.14230753 Columns: gbifID, eventDate, year, month, day, decimalLatitude, decimalLongitude, class, order, family, genus, taxonRank, species, prediction, site. All but the last 3 columns are standard Darwin Core Archive columns as used by GBIF. species combines the DwCA columns genus and specificEpithet to give binomial species names. prediction gives our flowering prediction model's estimate of whether a flower is present in the herbarium sheet image associated with the occurrence, continuous in (0,1). site is the name of the pollen monitoring site for which that prediction was used. Note that because some pollen monitoring regions overlap, the same occurrence may appear more than once, associated with different sites. Note also that the site kjevik corresponds to Kristiansand in the paper, and other site names have been modified to be lower case and exclude non-Latin characters (e.g., bodo rather than Bodø).