Progress in food web ecology relies on assessing the strengths, limitations and complementarity of different approaches to characterize food web structures. We determined whether the network approach, depicting the detailed architectures of nodes connected by trophic links, and the stable isotope approach, providing a biochemical representation of food webs using ecological tracers, deliver similar view of food web structures. We synthesized candidate metrics that characterize the vertical structure, clustering, complexity and trophic diversity of food webs in both approaches. We then combined databases of empirical freshwater stable isotope values and a niche model to simulate diverse but realistic networks and convert them into isotopic spaces. Finally, we evaluated the associations between metrics computed on networks and stable isotopes using redundancy analysis (RDA) and their variability depending on initial characteristics of food webs. Network and isotopic metrics were well associated (R² = 0.62), especially those describing vertical food web structure (e.g. food chain length). Network clustering metrics (e.g. number of modules) were associated with isotopic clustering (e.g. nearest neighbour distance) and isotopic diversity metrics (e.g. standard ellipse area). There were poor associations between complexity metrics in the network approach (e.g. connectance) and other isotopic metrics. Increasing node richness and food web compartmentalisation altered the associations between network and isotopic metrics by modifying delineation of energy pathways and trophic clusters. Our study highlights the conditions where network and stable isotope metrics are similar or bring complementary information to characterizing food web structures. We encourage continued exploration of the complementarity of both approaches under different environmental constraints and propose that isotopic data can represent biodiversity structure to test long‐standing hypotheses about food web structure complexity, including those related to ecosystem function.
Vagnon et al. (Tue,) studied this question.