Discrimination of foraging paths produced by different search tactics
Higgins, Christopher L.
MetadataShow full item record
Search tactics are cognitive processes, or decision mechanisms, that organisms utilize to elicit a group of related and sequential behaviors that allow them to locate available resources such as food, mates, refugia, and habitats. However, our knowledge of the actual tactics that animals utilize while searching for resources is limited, and very little empirical evidence has been gathered. As a result, a suite of models of theoretical search tactics was developed to emulate the searching behaviors of mobile individuals so that inferences might be made about their decision mechanisms, and to determine whether individuals are capable of searching with a globally optimal solution. The theoretical tactics included four deterministic tactics (maximum-distance search, minimum-distance search, nearest-neighbor search, and trajectory-directed search) and five probabilistic tactics (reciprocal-distance search, inverse-squared-distance search, uniformly random search, Pearson random walk, and Levy random walk). These tactics included variations of random walks and systematic searches in an attempt to characterize biologically realistic searching behaviors. The theoretical search tactics were simulated with hypothetical predators finding and consuming the same sets of resources, assuming no satiation or learning, so that comparisons among the resulting search paths could be made. Correlated measures of path length, consistency of movement, path linearity and turn rate were used to characterize the geometry of the resulting search paths. Classieal multivariate analysis, including discriminant function analysis (DFA) and non-linear artificial neural networks (ANN), were used to discriminate among the theoretical tactics and to classify "unknown" search paths into one of the underlying search tactics, both based on the geometry of the resulting paths. Both linear and non-linear discriminatory anah ses revealed a strong overlap between the nearest-neighbor search and the minimum-distance search tactics. This overlap implies that it is possible for animals to utilize search tactics that mirror the globally optimal solution without having complete knowledge of the specific location of available resources. Furthermore, both methods of analyses demonstrated that inferences could be made about the actual tactics that animals utilize while searching for resources based on the geometry of the search paths, but with varying degrees of reliability.