Making plant–pollinator data collection cheaper for restoration and monitoring

Wildflower plantings are a key tool for wild bee conservation, and plant–bee interaction data are frequently used in seed mix design to ensure that plantings provide sufficient resources for a diversity of bees. Plant–bee interaction data are also used in monitoring programmes to measure the success of bee-supporting habitat. However, collecting plant–bee interaction data can be expensive and these data are not used by many seed mix design practitioners and monitoring programmes. Therefore, a crucial question is how does the conservation value of seed mixes vary as a function of the intensity of the data collection effort underpinning their design? We leverage a plant–bee interaction data set to ask how bee richness is expected to change when informed by different scenarios of reduced data collection effort and cost. From the original, large data set, we created subsets of data by randomly reducing (1) the number of specimens sampled, (2) the number of sites sampled, (3) the number of sampling days per site and (4) using data from a single, representative taxon (bumble bees). We ask whether seed mixes designed from these reduced data sets support comparable bee richness to the full data set, and for any reduction in cost. Reductions in the number of specimens sampled and the number of sampling days per site yield seed mixes that support comparable bee richness to the full data set. Reductions in the number of sites yielded seed mixes that support lower bee richness than the full data set. Using bumble bee interaction data only yields seed mixes that support lower bee richness than the full data set, especially when few plants are included in the mixes. Synthesis and applications. We recommend prioritizing broad spatial, temporal and taxonomic coverage of plant–bee interaction data to guide cost-effective seed mix design. Our results also provide guidance for practitioners designing programmes to monitor bee richness in restored habitat because plant–bee interaction data may be made cheaper by collecting fewer specimens per sampling event or collecting on fewer dates per site (as long as coverage of the full bee flight season is maintained).