Global change, especially landscape simplification, is a main driver of species loss that can alter ecological interaction networks, with potentially severe consequences to ecosystem functions. Therefore, understanding how landscape simplification affects the rate of loss of plant-pollinator interaction diversity (i.e., number of unique interactions) compared to species diversity alone, and the role of persisting abundant pollinators, is key to assess the consequences of landscape simplification on network stability and pollination services. We analysed 24 landscape-scale plant-pollinator networks from standardised transect walks along landscape simplification gradients in three countries. We compared the rates of species and interaction diversity loss along the landscape simplification gradient and then stepwise excluded the top 1-20% most abundant pollinators from the data set to evaluate their effect on interaction diversity, network robustness to secondary loss of species, and flower visitation frequencies in simplified landscapes. Interaction diversity was not more vulnerable than species diversity to landscape simplification, with pollinator and interaction diversity showing similar rates of erosion with landscape simplification. We found that 20% of both species and interactions are lost with an increase of arable crop cover from 30 to 80% in a landscape. The decrease in interaction diversity was partially buffered by persistent abundant generalist pollinators in simplified landscapes, which were nested subsets of pollinator communities in complex landscapes, while plants showed a high turnover in interactions across landscapes. The top 5% most abundant pollinator species also contributed to network robustness against secondary species loss, but could not prevent flowers from a loss of visits in simplified landscapes. Although persistent abundant pollinators buffered the decrease in interaction diversity in simplified landscapes and stabilised network robustness, flower visitation frequency was reduced, emphasising potentially severe consequences of further ongoing land-use change for pollination services.

Plant-pollinator interactions (bees: Hymenoptera: Anthophila; and hoverflies: Diptera: Syrphidae) were sampled along transects (1 km x 2 m) in different flowering habitat types (e.g., meadows, forest edges, hedgerows, flowering crops) in each surveyed agricultural landscape (24 landscapes in total, n=8 landscapes in Switzerland, Germany, and France). The transects were located within 500 m of the center of the landscapes, and 3 min were invested for sampling 25 m of the transect (stopping the clock when handling the pollinator or flower). Flowering plants were surveyed in quadrats of 1 m^2 along the same transects. Three sampling rounds were conducted in 2020 (the first in April, the second in May/June, and the third in July). Please see the methods section in the published manuscript for details about the sampling procedure.