Merging empirical and mechanistic approaches to modeling aquatic visual foraging using a generalizable visual reaction distance model

Visual encounter distance models are important tools for predicting how light and water clarity mediate visual predator-prey interactions that affect the structure and function of aquatic ecosystems at multiple spatial, temporal, and organizational scales. The two main varieties of visual encounter distance models, mechanistic and empirical, are used for similar purposes but take fundamentally different approaches to model development and have different strengths and weaknesses in terms of predictive accuracy, physical and biological interpretability of parameters, ability to incorporate outside information, and utility for knowledge transfer. To overcome weaknesses of existing mechanistic and empirical models and bridge the gap between approaches, we developed a generalized visual reaction distance model that relaxes assumptions of a widely-used mechanistic model that are violated in real predator-prey interactions. We compared the performance of the generalized visual reaction distance model to a widely used mechanistic model and an empirical visual encounter distance model by fitting models to data from four predator-prey experiments. The generalized visual reaction distance model substantially outperformed the other models in all cases based on fit to reaction distance data and presents an attractive alternative to prior models based on comparatively high predictive accuracy, use of interpretable parameters, and ability to incorporate outside information—characteristics that facilitate knowledge transfer.