Ant Search Strategies for Unknown Targets

Abstract

Search for resources of unknown locations is important for most mobile organisms, but we know relatively little about how animals perform it, compared to search for known locations and theoretical work or how agents should move. We know even less about it in collective systems like social insect colonies, where the individuals are expected to coordinate to maximize group success. In my PhD I investigate the search behavior of Temnothorax rugatulus ants in a large arena in the lab. Specifically, I show that1) ants do not merely walk in purely random fashion, as was the previous assumption but rather meander regularly in smooth left and right turns. This systematic element decreases the number of path self-crosses while not increasing the distance ants disperse from the nest, making their search patterns more space-filling. 2) The study species frequently relocate their nest, necessitating the foragers to become familiar with their new environment for effective navigation and search. In our experiments, over 3 days after being placed into a novel arena, colony-level exploration activity decreased, but dispersivity away from the nest increased, and individuals’ movements overall became straighter and faster over time, possibly reflecting a shift from familiarization or marking to searching. 3) Previous studies showed that ants walk straighter and faster on surfaces that are marked with chemical footprints of nestmates. We find the opposite effect. This can be explained by 3 factors: Chemical footprints causally make ants walk less straight; ants that move straighter move farther away, where there are fewer footprints; and some areas make ants walk less straight, thus depositing more footprints there. We also find no effect of ants turning away from areas with higher footprint concentrations. Quantitatively, our study shows that in a natural setting, environmental cues and interindividual variation are probably much more important than reactions to chemical footprints. The fourth chapter is devoted to an important methodological issue in the field of movement ecology. Studies on animal movement frequently need to resample the original track to retain only the ‘biologically significant’ turns between relatively straight path segments. Many methods of turn identification, or resampling, have been used in the movement ecology field, with very little validation of the results of the resampling step. We simulated a range of tracks and compared their characteristics with resampled tracks from four resampling methods. We found great inaccuracies, inconsistent results across all methods and ground truth track parameters, and high sensitivity to choosing the required resampling parameters. Results using different methods should thus be interpreted cautiously, and future studies should select the resampling method with a rigorous procedure.