1. Eco-evolutionary feedback loops arise when evolutionary change (i.e. heritable changes in the phenotype) is fast enough to drive ecological change (e.g. population dynamics), which in turn feeds back on phenotypic evolution through selection, thus closing the loop. 2. Quantifying feedback effects is challenging because they are highly context-dependent due to the high number of possible interactions between the ecological and evolutionary variables at play. Traditional parametric dynamical models often fail to capture this complexity as they are bound to represent an arbitrarily chosen subset of these interactions. 3. To address this, we introduce a novel approach, by combining the Geber method with neural ordinary differential equations (Geber-NODEs), which allows us to nonparametrically estimate the coupling between evolutionary and ecological variables, and hence reveal the most important pathways for eco-evolutionary feedbacks. By further embedding the Price equation in the Geber-NODE system (Price-Geber-NODEs), we are able to separate eco-evo feedbacks operating through selection and plasticity. 4. We illustrate our approach in various eco-evolutionary systems, including the rotifer-algae system and Darwin’s finches. Our results suggest that the strength of eco-evolutionary coupling varies substantially in time and across species, and that indirect feedbacks mediated by the impact of population dynamics on resource levels may be potentially more prevalent than direct feedbacks between population density and phenotypic change. This is a crucial step towards discovering eco-evolutionary dynamics where knowledge of the equations governing the underlying processes is limited.
