UC Berkeley

The antennules of many marine crustaceans enable them to rapidly locate sources of odorant in turbulent environmental flows. The antennules are typically flicked through the water, causing the animal to take spatially and temporally discrete odorant samples. A substantial gap in knowledge concerns how the physical interaction between chemosensory appendages and the chemical filaments forming a turbulent plume affects odorant detection and filters the information content of the plume. The research presented here is focused on using numerical models to simulate the flow of odorant-laden water around arrays of chemosensory aesthetascs, and the transport of odorant molecules to their surfaces, during a plume sampling event. The time-varying odorant flux signals generated during these events are of key interest, since they are the lens through which a plume tracking agent will perceive its odor environment.

Simulations of infinitely long arrays of sensory hairs indicate that there are likely to be design tradeoffs between maximizing the sharpness of an odorant flux signal versus the total amount of odorant mass detected. It is also clear that the duration of odorant flux during a sampling event similar to a flick downstroke is not long enough to enable detection of 0.6 mm wide odorant filaments by crustaceans such as lobsters, or by engineered chemical sensors currently available. This suggests that the return stroke and interflick pause may be critical if these animals are to detect fine-scale plume structures.

Models of hair arrays with a finite number of hairs reinforce the above conclusions, but because this more realistic geometry allows water to flow around the array of hairs in addition to through, it reveals additional behaviors that the infinite array simplification cannot capture. Fundamentally different trends in metrics describing the sharpness of the flux signal are observed for rapidly flicking, sparse arrays of hairs versus slowly moving, densely packed arrays. These surprising transitions are not well predicted by simple parameters describing only the fluid velocity field, pointing out the importance of explicitly modeling or observing odorant transport in addition to the flow of water.

A tomographic scan of a real aesthetasc array morphology, that of the spiny lobster P. argus, reveals substantial variation and deviation from any simple, idealized geometry, and 3D effects are likely to be important to its odorant sampling dynamics. Nonetheless, numerical simulations of an idealized version of P. argus morphology indicate that on average, odor-laden water is effectively channeled into the aesthetasc array as compared to a simple straight row of hairs. A simple row, however, still achieves greater odorant flux in most regards, again emphasizing differences in properties of the flow versus properties of passive scalar transport. Crustacean antennules might therefore best serve as starting iterations instead of optimal solutions for the design of engineered chemical sensor arrays for use on plume tracking robots.