Differences in visitation of honeybees and bumblebees to ornamental plant varieties can be explained by floral traits

by Verweij et al.

Pollinating insects, like bees, are essential for pollinating 75% of crops and many wild plants, making them crucial for both agriculture and natural ecosystems. However, as a result of climate change, agriculture and urbanization their populations are declining. Urban green spaces can help support bees by providing rich habitats with diverse flowering plants. Many ornamental plants sold in garden centers are marketed as bee-friendly, but often this is not confirmed by actual research.

A combination of flower traits, such as nectar, pollen, colour, flower shape and the number of flowers, determine how attractive a flower is to bees. For example, bees prefer flowers that produce plenty of high-quality nectar and pollen and with bright colours, such as blue and yellow. However, they will adapt to visits other colours if those flowers offer better rewards. Additionally, UV-patterns can help bees to locate nectar, and plants with many flowers are more appealing because they provide more food in one place. The depth of the tube that holds the nectar, also referred to as the corolla tube, can affect how easily bees access nectar, which in turn impacts how efficient they collect the nectar. Breeding ornamental plants may reduce the attractiveness of flowers to bees as they it affects these flower traits, e.g. these flowers may not produce any nectar or pollen or may have become too big to be accessible to many wild bees. This is because ornamental plants have been selected to appeal more to humans, rather than bees.

This study examined how flower traits influence bee visitation to different cultivars of eight ornamental plant genera. Of the 119 cultivars we studied, most were not very attractive to honeybees or bumblebees, but a few were highly attractive. Bees responded differently to traits like nectar sugar content, corolla tube depth and flower colour, and to different combinations of these flower traits depending on the plant genera. For example, both honeybees and bumblebees preferred flowers with a high nectar sugar content, but only honeybees preferred shorter corolla tubes. Our findings show that ornamental plants can support bees if selected carefully based on flower traits. Plant breeders can use this information to breed plants that are both beautiful to humans and beneficial to pollinators. However, native plants will remain critical for supporting more specialized and diverse wild bee species. By combining ornamental and native plants, we can create urban green spaces that help sustain healthy pollinator communities.

Read the scientific article in JPE

Priming bumble bees with caffeine, odour of the target crop, and a food reward, has minimal effects on fruit crop pollination and yield in a semi-commercial setup

by Arnold et al.

Caffeine is produced in the nectar of some plants including coffee and citrus. When honeybees feed on caffeinated nectar, it improves their memory for scents associated with food.  Caffeinated bees might therefore pollinate more effectively as they can more effectively relocate a nutritious flower while at the same time transferring pollen.  We have previously shown in the laboratory, using artificial flowers, that bumblebees exposed to a floral scent in the nest seek out that scent outside the nest when foraging for food, if they consumed caffeine while learning the scent. 

Here we report a semi-field experiment (using individual enclosed polytunnels) in which we tested whether bumblebees could be “primed” to visit strawberry crop flowers more by learning to recall a strawberry flower scent alongside caffeinated sugar-water in the nest before being released. If the commercial bumblebees formed a strong memory for strawberry flower scent, paired with a good food experience in the nest, we expected that they would already have a preference for the crop flowers compared to any other flowers when foraging. 

However, when this was tested in 12m long mini-polytunnels, there was no consistent difference between bumblebees that had received caffeine or not, and whether the artificial strawberry floral odour was present in the nest or not. There was also no consistent difference in strawberry fruit size or quality between tunnels containing control bees, caffeine-primed bees, and bees that received the odour-sugar combination in the nest but no caffeine, meaning that caffeine and floral odour did not make these bees better pollinators.

Read the scientific article in JPE

Utilising affordable smartphones and open-source time-lapse photography for pollinator image collection and annotation

Photography set-up in the field and resulting image

By Stefan et al.

Pollinators, such as bees (Hymenoptera) and flies (Diptera), are essential for helping plants reproduce, including many crops we rely on for food. Understanding which insects visit which flowers is crucial to tracking how environmental changes affect these interactions. Traditional monitoring methods, such as catching insects and identifying them in a lab, are costly, time-consuming, and require killing the insects. Our study explores a non-destructive solution using affordable smartphones and time-lapse photography.

We mounted smartphones above flowers to capture time-lapse photos of visiting insects and tested whether the images provided enough detail for experts to identify the insects accurately. About 90% of all Hymenoptera instances (bees and ants) were identified to the family level, nearly 85% of pollinating families to genus, and only around 25% to species. Smaller insects, such as flies, were harder to identify due to the lack of fine details in their tiny features, like wing veins. For flies, only 26% were identified to genus and just 15% to species.

Because the insects were not physically captured for verification, the identification process involved some degree of subjectivity. Future research should focus on creating identification keys based on features visible in images. When fine identification isn’t possible, grouping insects into broader categories based on visible traits could make analysis more practical. These groupings will also aid in developing realistic artificial intelligence (AI) classes for automated insect classification.

Despite these challenges, our study highlights the potential of affordable smartphones as scalable tools for pollinator monitoring at broader taxonomic levels. Building on the annotated dataset we collected, our next goal is to develop artificial intelligence tools that can localise and identify insects in images more efficiently, advancing pollination ecology research and supporting conservation efforts.

Read the scientific publication in JPE

German version:

Bestäuber, wie etwa Bienen (Hymenoptera) und Fliegen (Diptera), sind unerlässlich für die Fortpflanzung von Pflanzen, einschließlich vieler Nutzpflanzen, die für unsere Ernährung wichtig sind. Dabei ist es entscheidend zu verstehen, welche Insekten welche Blüten besuchen, um nachvollziehen zu können, wie Umweltveränderungen diese Interaktionen beeinflussen. Traditionelle Beobachtungsmethoden, wie das Fangen und Identifizieren von Insekten im Labor, sind jedoch kostenintensiv, zeitaufwendig und erfordern das Töten der Insekten. Unsere Studie untersucht eine nicht-destruktive Alternative mittels erschwinglicher Smartphones und Zeitrafferfotografie.

Wir montierten Smartphones über Blumen, um Zeitrafferfotos von besuchenden Insekten aufzunehmen, und testeten, ob die Bilder ausreichend Details liefern, damit Experten die Insekten präzise identifizieren können. Etwa 90 % aller Hymenoptera-Exemplare (Bienen und Ameisen) konnten bis zur Familienebene identifiziert werden, fast 85 % der bestäubenden Familien bis zur Gattung und nur etwa 25 % bis zur Art. Kleinere Insekten, z.B. Fliegen, waren aufgrund fehlender feiner Details, etwa in den Flügeladern, schwieriger zu bestimmen. Bei Fliegen konnten nur 26 % bis zur Gattung und lediglich 15 % bis zur Art identifiziert werden.

Da die Insekten für eine Verifizierung nicht gefangen wurden, war die Identifizierung zum Teil subjektiv. Künftige Forschung sollte Bestimmungsschlüssel entwickeln, die auf in Bildern sichtbaren Merkmalen basieren. Die Gruppierung von Insekten in gröbere Kategorien anhand sichtbarer Merkmale könnte die Analyse erleichtern, sollte eine genaue Identifikation nicht möglich sein. Mithilfe solcher Gruppierungen könnten dazu beitragen, für künstliche Intelligenz (KI) realistische Klassen zur automatisierten Insektenklassifikation zu entwickeln.

Trotz dieser Herausforderungen zeigt unsere Studie, dass erschwingliche Smartphones durchaus zum Bestäubermonitoring auf höherer taxonomischer Ebene dienen können. Unser nächstes Ziel ist es, mit dem von uns gesammelten annotierten Datensatz eine künstliche Intelligenz zu entwickeln, die Insekten in Bildern effizient lokalisieren und identifizieren kann. Damit möchten wir die Forschung in der Bestäubungsökologie vorantreiben und Naturschutzmaßnahmen unterstützen.