Floral resource availability and honeybee flower visitations in Oslo, Norway

Urban green infrastructure can provide important habitats for pollinators and support urban ecosystem services. Therefore, these areas must be managed to maximize biodiversity and density of pollinating insects. We used DNA metabarcoding to study honeybee pollen resource use over time and space in the city of Oslo, Norway, and to assess the role of green infrastructure as a resource for pollinators and the services they generate. Urban honeybees used diverse pollen resources throughout their active season. There was considerable seasonal turnover in pollen resource use that reflected flowering phenology. Non-native plants (including invasive species) were an important resource early in the season but were replaced by native plants later in the season. Hive location was not strongly correlated with resource use, likely indicating effective long-distance foraging in the fragmented urban landscape. However, flower visitation rates and floral resource density in public urban green spaces were coupled with pollen use. Honeybees collected pollen from a small number of preferred species but also visited other species, likely for nectar. To preserve pollinator services, urban planning should consider flower resource management, with particular focus on planting native species that can act as early season resources for bees. Public campaigns and other incentive mechanisms to promote the cultivation of native plants that are resources for pollinators and the protection of urban semi-natural habitats have the potential to enhance the value of green infrastructure to support urban pollinators and pollination services.

Methods
Floral inventories were conducted in 100 green spaces within the Oslo urban area between July 18 and August 4, 2017, generating a landscape-level estimate of floral resource abundance for this time period. Sampling locations were extensions of previous work, with sites selected to capture a range of habitat suitability values as expressed in early versions of a model for urban pollinator habitat (Stange et al. 2017). The urban green spaces investigated were publicly accessible and included hedgerow, lawn, meadow, and waste areas containing primarily ruderal vegetation, as well as parks and cemeteries containing planted and tended vegetation. Green spaces composed of short cut lawns, forests lacking flowering ground cover, and private garden green spaces were excluded from the inventory. Each green space was visited once to determine floral resource density using 1 to 3 transects according to vegetation patch size and shape for a total of 20 transect meters. All open floral units were recorded within a 40 cm width along each transect. The definition of a unit (individual flower, entire inflorescence) differed between species. For inflorescences with a high number of individual flowers (ex/ racemes, capitula), we used a multiplication factor to calculate the total number of flowers of that species in the transect. Multiplication factors were based on the average number of flowers per recorded unit, retrieved from the literature (Lid and Lid 2005; Mossberg et al. 1992; and the Nordic virtual flora at the Swedish Natural History Museum http://linnaeus.nrm.se/flora/). The mean floral resource density for the landscape was calculated per plant species across all the inventoried green spaces, as well as within a 1-, 3-, and 5-km radius of each individual hive. Landscape-level honeybee visitation rates to different floral resources were calculated from 60 transect meters within each green space. One to four visits (mean 2.3) were made to each green space between July 18 and August 10, 2017, during which we walked each transect for approximately 5 min, allowing for additional time for specimen handling and species identification. We recorded the number of honeybees observed on flowers, including the plant species, within 2.5 m on both sides of the transect. Visitation rates were calculated as a mean across all transects in the landscape, as well as for all transects within a 1-, 3-, and 5-km radius of each individual hive.