Citizen science data reveals the need for keeping garden plant recommendations up-to-date to help pollinators

The study addresses whether widely disseminated “pollinator-friendly” plant recommendations are specific, accurate, and current enough to effectively support pollinators, particularly bumblebees, in urban and garden settings. With documented declines in terrestrial insect pollinators and habitat loss as a major driver, gardening for pollinators has been promoted via books, websites, seed mixes, labels, and lists. However, such labels are often generic, implying uniform benefit across diverse pollinator taxa and species. The authors hypothesize that practitioner lists lack independence, specificity, and timeliness, potentially misaligning with actual plant use by bumblebees. They aim to compare practitioner recommendations with citizen-reported bumblebee foraging records to evaluate concordance and assess species-level differences in plant use.

Prior work highlights pollinator declines and the importance of urban areas and domestic gardens for flower-visiting insects. Published evaluations of garden plant attractiveness show variation among species and cultivars, and caution against overreliance on static lists (“listmania”), noting issues such as lack of independence among sources, outdated information, and limited evidence bases. Phenological shifts under climate change can decouple flowering and pollinator activity, undermining static recommendations. Pollinator-friendly labels influence consumer behavior, but may be imprecise for particular pollinator groups. Studies also document species-specific diet breadth among bumblebees, with some specialist tendencies and morphological constraints shaping plant use.

Data sources included: (1) Practitioner recommendations compiled from 23 widely available UK sources (9 pollinator-friendly lists; 14 bumblebee-specific lists). Plants were recorded at species or genus level per source specificity; cultivars were generally excluded. Each plant’s recommendation frequency was the number of sources listing it. This yielded two ranked lists: pollinator-friendly (465 taxa) and bumblebee-friendly (376 taxa). (2) BeeWatch citizen science records from 25 Aug 2011 to 30 Jun 2017: 6,429 bumblebee–plant interaction records from gardens and public greenspaces across the UK, with uploaded photos, date, location, and plant (free text or from compiled list). Records were opportunistic and not standardized for effort, weather, or floral abundance. Analyses: Plants in BeeWatch were ranked by frequency of occurrence overall and per bumblebee species. Proportions were computed by dividing plant-specific counts by total records per bumblebee species and overall. Pearson correlations tested: (i) recommendation frequencies between pollinator- vs bumblebee-friendly practitioner lists; (ii) pollinator-friendly recommendation frequency vs BeeWatch plant frequency; (iii) bumblebee-friendly recommendation frequency vs BeeWatch plant frequency. Rarefaction analysis (vegan in R 3.4.0) estimated expected numbers of plant species used per bumblebee species by randomly sub-sampling 10 plant choices without replacement, 100 repeats; species with fewer than 10 records were excluded from rarefaction predictions. Correlation between months of activity per bumblebee species and predicted plant richness from rarefaction was assessed. Mapping used ArcGIS Desktop 10.7.

- Practitioner list similarity: Plants frequently recommended as pollinator-friendly and as bumblebee-friendly strongly overlapped (r=0.75, p<0.001), indicating low specificity for bumblebees. - Concordance with BeeWatch usage was modest: recommendation frequency vs BeeWatch frequency r=0.52 (pollinator-friendly) and r=0.57 (bumblebee-friendly), both p<0.001. - Data scope: BeeWatch contained 6,429 records of bumblebee plant use across 334 plant taxa; practitioners listed 465 (pollinator-friendly) and 376 (bumblebee-friendly). Overlap: 250 BeeWatch plants occurred on pollinator-friendly lists; 235 occurred on bumblebee-friendly lists. Only 12 of the top 25 practitioner-recommended bumblebee plants appeared in BeeWatch’s top 25, and less than half of BeeWatch’s top 25 appeared in practitioners’ top 25. - Usage is non-binary and diffuse: Even the most-used plant (Lavandula spp.) accounted for at most 11% of records for any species, and the 7 most-reported plants constituted only 3–8% of total records for all true bumblebees. - Species-specific preferences: Bumblebee species differed markedly in top plants; e.g., B. monticola and B. jonellus showed higher use of Calluna vulgaris; B. pratorum often used Geranium spp.; B. hortorum used Digitalis spp.; B. lucorum agg. and B. terrestris were most similar in diet. Cuckoo bumblebees relied heavily on Cirsium palustre. - Sampling/visibility effects: Many practitioner-listed trees/shrubs (>2.5 m) were absent or underreported in BeeWatch (e.g., Malus spp., Pyrus communis, Salix spp.), likely due to observation/photography constraints. Some common garden plants (e.g., Lobelia, Petunia, Kniphofia, Agapanthus) appeared in BeeWatch but not on practitioner lists. Plants flowering outside typical bumblebee seasons (e.g., Cyclamen, Pulsatilla) were reported, aligning with winter-active B. terrestris. - Diversity vs activity duration: Number of months active correlated with predicted number of plant species used (r=0.73, p=0.005). Rarefaction showed expected plant richness per species increases with sample size, indicating many plants are “good for bumblebees” but most are little used.

Findings indicate that widely promoted pollinator-friendly and bumblebee-friendly plant lists are highly overlapping and offer limited specificity for bumblebee needs. Their modest correlation with actual plant use suggests lists may be outdated, non-independent (derivative of older sources), and insensitive to species-level preferences, changing plant availability, cultivar differences, and phenological shifts. BeeWatch data reveal significant interspecific variation in plant use and a non-binary distribution of attractiveness, with only a small fraction of plants accounting for a modest share of visits and distinct top plants for many species. While labels and lists are effective entry points for public engagement and behavior change, oversimplified dichotomies can mislead expectations and dampen engagement when plants attract fewer pollinators than anticipated. The authors advocate for dynamic, evidence-based recommendations powered by live citizen science, tailored by pollinator species or groups, and potentially delivered via interactive HCI tools to support learning and adaptive gardening practices.

The study demonstrates that static practitioner plant lists lack specificity and only modestly reflect actual bumblebee foraging patterns captured by citizen science. Bumblebee species display distinct, diffuse plant use, undermining binary “pollinator-friendly” labels. The authors propose up-to-date, dynamic planting recommendations driven by live citizen science data, with options to filter by pollinator species or group, to better guide gardeners and support pollinators. Future work should integrate structured components (e.g., effort, floral abundance), address plant abundance/popularity biases, incorporate cultivar-level differences and phenology, and develop interactive platforms that personalize, update, and evaluate recommendations over time.

BeeWatch data are opportunistic, not standardized for effort, weather, or plant abundance, so recorded frequencies reflect both attractiveness and availability/popularity of plants. Tall trees/shrubs may be under-detected due to line-of-sight and photo constraints; some wildflowers may be underrepresented due to rarity or garden avoidance. Practitioner lists may lack independence and exclude newer cultivars/species and academic sources. Rarefaction predictions were not possible for species with fewer than 10 records. Plant identification granularity varied (genus vs species), and cultivars were generally not differentiated.