Data Package Metadata   View Summary

Overviews of the study sites, study design, and objectives can be found in Forrest & Chisholm (2017, Ecology, doi: 10.1002/ecy.1655) and Wong & Forrest (2021, Journal of Animal Ecology, doi: 10.1111/1365-2656.13460). Study sites are visited every 3-10 days throughout the growing season.

Temperature data are recorded hourly at each site by a HOBO pendant data-logger (light & temperature; Onset Computer, Bourne, Massachusetts, USA) attached to the bottom of one centrally located trap-nest (nesting block). The data-logger is not properly shielded from solar radiation, but it is shaded by the block. The light sensor faces out (=down). The logger is attached by a wire to the pipe-strapping that attaches the block to the tree, and it is additionally secured by another piece of pipe-strapping or wire mesh that supports it against the bottom of the trap-nest (so it isn’t dangling). Typically, the temperature data are downloaded once in the spring (May-June) and again at the end of summer (August). Batteries are replaced at the same time if necessary. Data from all sites and years are compiled into a single data file. One data-logger (at the VB site) failed over the winter of 2013-2014, so those readings are missing. Occasionally, single hourly readings were missed due to errors with time stamps or downloading being done at the time of temperature recording.

As soon as any bee activity or flowering is observed at a site, floral surveys begin at that site (though see below for exceptions). Surveys were done twice a week, as much as possible (once a week at the VB site), from 2013 to 2019. Floral data were not collected in 2020. As of 2021, surveys are done weekly at all sites. Surveys are occasionally delayed by 1–3 days because of poor weather. The specific flower taxa counted have changed slightly over the years, based on our growing knowledge of which taxa are used by bees and of which taxa can be distinguished using their pollen. We have also grouped some taxa to simplify sampling, as shown in Appendix A. Floral densities are measured by counting the number of open flowers (or inflorescences, for Asteraceae) seen while walking in an expanding spiral from a fixed, central point; when 100 flowers are reached, we measure the distance from the centre of the site to this flower. Distances under 30 m are normally measured using a surveyor’s tape, to the nearest 0.1 m. Distances greater than 30 m are measured to the nearest metre using a Trimble GeoXH GPS unit. If 100 flowers are not reached within a 100 m radius of the centre of the site, ">100 m" is recorded for the distance and the number of flowers observed is recorded in the "notes" column. Distances are converted to densities using the fact that the measured distance represents the radius within which 100 flowers were observed.

In the first year of the project (2013), floral resources were quantified only once the first bees began nesting at a site, so floral densities prior to that point are unknown. In all years, the end of flowering may have been missed for some taxa because of the team’s departure from the field station in August. (The last date recorded for a given site in each year represents the last date on which floral surveys were conducted at that site in that year.)

Pollen samples are taken from nests constructed by the bee occupants of the trapnests. Nests are observed at each site visit so that nesting bees can be identified and individually marked, as described below. Bees are identified to genus or species based on field observations of females and their nest architecture, and/or by collecting adult offspring as vouchers (housed at the Forrest lab at the University of Ottawa and the Canadian National Collection of Insects, Arachnids and Nematodes). Nesting female Osmia bees (and sometimes females of other bee taxa) are individually marked using Testors paint applied to the dorsum of the mesosoma so that nests can be assigned to individual mother bees (many of which construct multiple nests sequentially over their lifetime).

Pollen samples are taken from all nests constructed by Osmia spp., and occasional samples are taken from nests constructed by other bee genera (Hoplitis, Megachile, Hylaeus). Pollen samples are taken by cutting a small flap (~1 x 2 mm) in the paper straw housing the nest and removing a tiny amount of pollen using fine-point forceps. The pollen is then melted with a cube of fuchsin gel under a cover slip on a microscope slide. Between 2013 and 2019, pollen was sampled from the innermost and outermost brood cell of each nest comprising four or more brood cells; pollen was sampled only from the innermost brood cell of nests comprising three or fewer brood cells. From 2021 on, pollen was sampled from only the first and last brood cells constructed by each individual bee (i.e., if an individual bee constructed multiple nests, only the innermost brood cell of the first nest and the outermost brood cell of the last nest was sampled). Nests constructed by unknown bees were sampled as in previous years.

The slide-mounted pollen samples are scored by undergraduate research assistants using a compound microscope at 200–400x magnification and a local pollen reference collection. Pollen is normally identified to the genus, tribe, or family level; species-level identification is reliable only for monotypic genera. Observers are instructed to scan the entire pollen sample and identify each distinct pollen type; they record whether each pollen type is a "major" (>25%) or "minor" (<25%) component of the pollen sample, based on their qualitative assessment. Pollen types represented by only a few pollen grains are recorded as "contaminants". Pollen identification is normally checked by a second observer who may correct the first observer’s identifications, reclassify the pollen as major/minor, or record additional pollen types. Pollen slides are stored at the Forrest lab at the University of Ottawa. Pollen was sampled from nests constructed in 2022, but these samples have not yet been identified and are not currently included in the dataset.

This method step describes provenance-based metadata