全球变化知识资源中心

Current understanding of mutualistic networks is grounded largely in data on interaction frequency, yet mutualistic network dynamics are also shaped by interaction quality-the functional outcomes of individual interactions on reproduction and survival. The difficulty of obtaining data on functional outcomes has resulted in limited understanding of functional variation among a network's pairwise species interactions, of the study designs that are necessary to capture major sources of functional variation, and of predictors of functional variation that may allow generalization across networks. In this community-scale study, we targeted a key functional outcome in plant-frugivore networks: the impact of frugivore gut passage on seed germination. We used captive frugivore feeding trials and germination experiments in an island ecosystem, attaining species-level coverage across all extant native frugivores and the plants they consume to (a) assess sources of functional variation, (b) separate effects of pulp removal from those of scarification via gut passage, and (c) test trait-based correlates of gut passage effect sizes. We found antagonistic seed predation effects of a frugivore previously assumed to be a seed disperser, highlighting the need to consider functional outcomes rather than interaction frequency alone. The other frugivores each exhibited similar impacts for individual plant species, with benefits primarily caused by pulp removal rather than scarification, supporting the use of animal functional groups in this context. In contrast, plant species varied widely in impacts of gut passage on germination. Species with smaller seeds and more frugivore partners had larger benefits of gut passage, showing promise for network metrics and functional traits to predict functional variation among plants. Synthesis. Combining network and demographic approaches, we assessed the degree and sources of variation in a key functional outcome of plant-frugivore interactions across an entire network. Using a detailed study design, our work shows how simpler study designs can capture primary sources of functional variation and that functional traits and network metrics may allow generalization across networks. Efficiently measuring and generalizing sources of functional variation within mutualistic networks will strengthen our ability to model network dynamics and predict mutualist responses to global change.