Quantitative plant-pollinator interaction webs have been used to describe and compare communities from both the pollinator and the plant perspectives. Inferences for pollination of plants made from the frequency of visits and the generality of their visitors may overestimate the extent of generalization among pollinators if individual foraging is more restricted than the species average. We hypothesized that incorporating data on the pollen carried on insect bodies would offer better insights into actual pollen delivery patterns. Insects were netted on seven flower species over two days in a 1 ha plot of restored prairie, yielding 486 specimens. We identified pollen grains on stigmas from flowers open during the insect census and from insect bodies. We compared observed proportions of conspecific pollen deposited on stigmas with predictions from two simple null models. The first predicted deposition from visitation frequency of insects weighted by the proportion of visits by each insect species to a plant species. The second predicted deposition from visitation frequency of insects weighted by the proportion of conspecific pollen on insect bodies for the most frequent insect visitor species to that plant. The proportion of conspecific pollen on stigmas (mean 92%, range 75 – 100%) was 1.72 (range 1.23 – 2.59) times and 1.24 (range 1.04 – 2.11) greater than that predicted by the first and second null models, respectively. The second model improved predictions for plant species whose visitors carried mostly conspecific pollen, suggesting that individual foraging was more restricted than species averages.
