Background:Pollination by insects is a key ecosystem service, but there is concern about the decline in pollinators, caused by habitat degradation, diseases and parasites, and climate change. There are key gaps in the pollination science evidence-base, particularly relating to which insects pollinate which plants, and how pollination networks are structured. Most research has focused on bee populations. However, hoverflies (Syrphidae) also feed exclusively on nectar and pollen, making them potential pollinators of a wide range of wild plants and crops. Results: Using DNA metabarcoding to identify pollen, we investigated pollen transport by a range of hoverfly species in conservation grasslands in Wales, UK. It was possible to assign over 98% of sequences to species, genus, or family level. Hoverflies transport pollen from more plant species than previously appreciated. Networks are generalised at the site and species level, but at the individual level varied from specialised to relatively generalised. This suggests that generalised networks may result from a series of short-term specialised feeding bouts by individual insects. Most pollen recorded came from common plant taxa. However, differences in the proportions of various plant taxa in pollen loads between hoverfly species demonstrate some functional complementarity. Significance:We show the value of DNA metabarcoding in investigating plant–pollinator interactions. It allows the systematic investigation of pollination networks, from individual insects to whole communities. Our results show how generalised networks can emerge from the short-term specialisation of individuals, thus reconciling generalised network structures with effective plant pollination. Treating hoverflies as a single functional group underestimates the range of pollination function within this ecologically diverse guild. This study is one of the first to use DNA metabarcoding to investigate a pollinator community and adds to our understanding of the role of hoverflies in pollen transport.