Interindividual variations in plant and fruit traits affect the structure of a plant-frugivore network

Frugivores select their food in a hierarchical way, from plants to individual fruits, to meet their nutritional requirements. According to the optimal diet theory, finding, handling, and digesting fruits is costly, thus plant species that increase attractiveness and reward are usually preferred by frugivores. The same should be expected for individual plants of the same population, which differ from one another in traits related to frugivore attraction. We tested the hypothesis that plant traits that increase attractiveness and reward to frugivores would be strongly selected by birds in a population of Henriettea succosa (Melastomataceae). In 20 h of focal observation in 19 individual trees (380 h in total), we measured plant and fruit traits known to influence frugivore attraction and reward: plant height, fruit size, and fruit sugar content. In addition, we recorded bird behaviour during fruit consumption. We built two weighted networks of birds and individual plants: one monolayer network and a multilayer network with four layers, one for each type of behaviour. First, we evaluated three weighted descriptors of network structure: nestedness, modularity, and specialization. Then, we calculated metrics of centrality and correlated them with plant traits. We recorded 271 visits by 22 bird species of eight families. The network is modular and specialized, showing that subgroups of H. succosa trees with different trait combinations attract different subsets of bird species, in a way that specialist trees are not connected to a subset of the bird species that visit generalist trees. We also found that centrality metrics reached higher scores in plants with lower height, larger fruits, and intermediate sucrose content. Fruit handling was the predominant foraging behaviour in the multilayer network and represented 90 percent of the interactions. Downscaling a plant-frugivore network to its individual trees showed that the structure of the system is influenced by interindividual variations in the tree population, in which individuals with the best combination of traits occupied central positions in the network.