What makes a service plant attractive to pollinators? An exploration of the influence of plant traits among 38 flowering plant species

Global declines of pollinator communities have triggered interest in flowering plants to support pollinators in agricultural areas. Flowering plants may differ in their attractiveness for pollinators, and it is not clear which plant traits (co-)determine the attractiveness of flowering plants for different pollinator groups. Here we assessed the communities of hoverflies and bees on 38 flowering plant species by visual counts and sweep net sampling in a replicated field experiment in Korla, Xinjiang, China. There were 13 sampling rounds in 2020 and 15 in 2021. Pollinators from sweep net samples were identified to species level and their body size recorded. Flower cover, corolla depth and width, plant height, and flowering duration were assessed. A total of 2751 hoverflies, 1350 honeybees and 621 wild bees were observed using the two sampling methods. Diplotaxis tenuifolia, Foeniculum vulgare and Anethum graveolens were the most frequently visited plants. The abundance of adult hoverflies and honeybees was negatively associated with corolla depth of flowering plants in 2020, while wild bee abundance was consistently negatively associated with corolla depth across both years. Plant flowering duration not only influenced the period of pollinator abundance, but longer flowering plants also hosted a higher abundance of hoverflies and wild bees in 2021. The abundance of pollinators was not associated with flower cover. The body length of hoverflies and wild bees was not associated with flowering plant traits. Our study demonstrates the potential of trait-based approaches to predict pollinator responses to inform the selection of flowering plants. However, it also reveals interannual variability in trait-pollinator relationships, highlighting the need for continued empirical testing to ensure effective plant selection for supporting local pollinator communities.