Network modularity influences plant reproduction in a mosaic tropical agroecosystem

摘要

Biodiversity influences ecosystem function, but there is limited understanding of the mechanisms that support this relationship across different land use types in mosaic agroecosystems. Network approaches can help to understand how community structure influences ecosystem function across landscapes; however, in ecology, network analyses have largely focused on species–species interactions. Here, we use bipartite network analysis in a novel way: to link pollinator communities to sites in a tropical agricultural landscape. We used sentinel plants of Brassica rapa to examine how the structure of the community network influences plant reproduction. Diptera was the most common order of flower visitors at every site. Syrphidae visits were the strongest contributor to the number of fertilized pods, while visits by Syrphidae, Hymenoptera and Lepidoptera had the strongest effect on the number of seeds per pod. Sentinel pots at forest sites were visited by more unique species (i.e. species with higher d′) than sites in other land uses, and dairy sites had more visitors that were common across the network. Participation coefficients, which indicate how connected a single node is across network modules, were strong predictors of ecosystem function: plant reproduction increased at sites with higher participation coefficients. Flower visitor taxa with higher participation coefficients also had the strongest effect on plant reproduction. Hymenoptera visits were the best predictor for participation coefficients but an Allograpta sp. (Diptera: Syrphidae) was the most influential flower visitor species in the landscape network. A diverse insect community contributed to plant reproduction and connection among nodes in this system. Identifying the ‘keystone’ flower visitor species and sites that have a strong influence on network structure is a significant step forward to inform conservation priorities and decision-making in diverse agroecosystems.