A mechanistic perspective of ecological networks highlights the contribution of alternative interaction strategies
O_LISpecies traits mediate ecological interaction outcomes and community structure. It is important, therefore, to identify the minimum number of traits required to characterise observed networks, i.e. the minimum dimensionality. Existing methods for estimating minimum dimensionality often lack three features commonly associated with a higher numbers of traits: a mechanistic description of observed interactions, alternative interaction modes (e.g. different feeding strategies such as active vs sit-and-wait feeding), and trait-mediated forbidden links. Omitting these features can lead to underestimation of the trait numbers involved, and therefore, minimum dimensionality. Here, we develop a minimum mechanistic dimensionality measure, accounting for these three features. C_LIO_LIThe only input required by our method is the observed network of interaction outcomes. We then assume how traits are mechanistically involved in alternative interaction modes. These unidentified traits are contrasted using pairwise performance inequalities between the interacting species. E.g. if a predator feeds upon a prey species via a typical predation mode, in each step of the predation sequence the predators performance must be greater than the preys. We construct a system of inequalities from all observed outcomes, which we attempt to solve with mixed integer linear programming. The minimum number of traits required for a feasible system of inequalities provides our dimensionality estimate. C_LIO_LIWe applied our method to 658 published empirical ecological networks including animal dominance, predator-prey, primary consumption, pollination, parasitism and seed dispersal networks, to compare with minimum dimensionality estimates when the three focal features are missing. Minimum dimensionality was typically higher when including alternative interaction modes (54% of empirical networks), forbidden interactions as trait-mediated interaction outcomes (92%), or a mechanistic perspective (81%), compared to network dimensionality estimates missing these features. C_LIO_LIOur method can reduce the risk of omitting essential traits that are involved mechanistically, in different interaction modes, or in failure outcomes. More accurate estimates will allow us to parameterise models to generate theoretical networks with a more realistic structure at the interaction outcome level. Thus, we hope our method can improve predictions of community structure and structure-dependent dynamics. C_LI