Olfaction informs animal navigation for foraging, social interaction, and threat evasion. However, turbulent flow on the spatial scales of most animal navigation leads to intermittent odor information and presents a challenge to simple gradient-ascent navigation. Here we present two strategies for iterative gradient estimation and navigation via olfactory cues in 2D space: tropotaxis, spatial concentration comparison (i.e., instantaneous comparison between lateral olfactory sensors on a navigating animal) and klinotaxis, spatiotemporal concentration comparison (i.e., comparison between two subsequent concentration samples as the animal moves through space). We then construct a hybrid model that uses klinotaxis but utilizes tropotactic information to guide its spatial sampling strategy. We find that for certain body geometries in which bilateral sensors are closely-spaced (e.g., mammalian nares), klinotaxis outperforms tropotaxis; for widely-spaced sensors (e.g., arthropod antennae), tropotaxis outperforms klinotaxis. We find that both navigation strategies perform well on smooth odor gradients and are robust against noisy gradients represented by stochastic odor models and real turbulent flow data. In some parameter regimes, the hybrid model outperforms klinotaxis alone, but not tropotaxis.
Keywords: Animal navigation; Computational modeling; Klinotaxis; Olfaction; Tropotaxis.
Copyright © 2021 Elsevier Ltd. All rights reserved.