Thu, Aug 18, 2022: 1:45 PM-2:00 PM
518A
Background/Question/Methods
What species interact with each other is a key question in community and network ecology. The concept of trait-matching implies that pairwise interactions are dictated by a match between consumer and resource species' traits. Should such rules prove widely applicable, then they would allow for general predictions of interaction structure rather than a tedious description of pairwise interactions one by one. Recently, a wide range of methods for establishing trait-matching have been proposed, each approaching the same phenomenon from a different perspective. Our objective was to evaluate the predictive power of theory-based vs machine-learning vs data-driven methods and to examine their potential synergies in terms of the ecological insight they provide. To do so, we applied five different methods to a species-rich tri-trophic Salix-galler-parasitoid network, exploring the structure of the two bipartite network elements of Salix-galler versus galler-parasitoid interactions. We enriched this data set with distinct traits for each trophic level, including body sizes, gall type (position on plant, structure of gall) and phenology, as well as phylogenetic proxies. We also compared traits to evaluate which were most important for driving network structure and whether traits or phylogeny was more successful for explaining the structure of each network.
Results/Conclusions
In the galler-parasitoid sub-network, approximately half of its structure (i.e. interactions) was explained by the species traits we used, whereas in the Salix-galler sub-network traits explained at most two-fifths. Gall type appeared to be the most important structuring trait in both networks. Depending on the method used, phylogeny explained as much, or substantially more than did traits. Overall we found that the Salix-galler network was built from highly specialized species while the galler-parasitoid network was more nested. As a result, different methods were more effective at capturing interactions and interaction structure in the different sub-networks. We used each method to predict all interactions within the networks and found that both networks were more modular, and the Salix-galler sub-network less nested, than predicted by any method. Thus, our analysis reveals how structuring impacts may vary even between levels within the same multitrophic network. As such, our study calls for comparative analyses of trait matching across a wide set of systems and methods. The insights gained by using a combination of tools are essential for predictions, estimating interactions likely missed and interactions resulting from new species invading a community or the interactions to expect within a community found in a new location.
What species interact with each other is a key question in community and network ecology. The concept of trait-matching implies that pairwise interactions are dictated by a match between consumer and resource species' traits. Should such rules prove widely applicable, then they would allow for general predictions of interaction structure rather than a tedious description of pairwise interactions one by one. Recently, a wide range of methods for establishing trait-matching have been proposed, each approaching the same phenomenon from a different perspective. Our objective was to evaluate the predictive power of theory-based vs machine-learning vs data-driven methods and to examine their potential synergies in terms of the ecological insight they provide. To do so, we applied five different methods to a species-rich tri-trophic Salix-galler-parasitoid network, exploring the structure of the two bipartite network elements of Salix-galler versus galler-parasitoid interactions. We enriched this data set with distinct traits for each trophic level, including body sizes, gall type (position on plant, structure of gall) and phenology, as well as phylogenetic proxies. We also compared traits to evaluate which were most important for driving network structure and whether traits or phylogeny was more successful for explaining the structure of each network.
Results/Conclusions
In the galler-parasitoid sub-network, approximately half of its structure (i.e. interactions) was explained by the species traits we used, whereas in the Salix-galler sub-network traits explained at most two-fifths. Gall type appeared to be the most important structuring trait in both networks. Depending on the method used, phylogeny explained as much, or substantially more than did traits. Overall we found that the Salix-galler network was built from highly specialized species while the galler-parasitoid network was more nested. As a result, different methods were more effective at capturing interactions and interaction structure in the different sub-networks. We used each method to predict all interactions within the networks and found that both networks were more modular, and the Salix-galler sub-network less nested, than predicted by any method. Thus, our analysis reveals how structuring impacts may vary even between levels within the same multitrophic network. As such, our study calls for comparative analyses of trait matching across a wide set of systems and methods. The insights gained by using a combination of tools are essential for predictions, estimating interactions likely missed and interactions resulting from new species invading a community or the interactions to expect within a community found in a new location.