Lobster Sniffing: Antennule Design and Hydrodynamic Filtering of Information in an Odor Plume

See allHide authors and affiliations

Science  30 Nov 2001:
Vol. 294, Issue 5548, pp. 1948-1951
DOI: 10.1126/science.1063724


The first step in processing olfactory information, before neural filtering, is the physical capture of odor molecules from the surrounding fluid. Many animals capture odors from turbulent water currents or wind using antennae that bear chemosensory hairs. We used planar laser–induced fluorescence to reveal how lobster olfactory antennules hydrodynamically alter the spatiotemporal patterns of concentration in turbulent odor plumes. As antennules flick, water penetrates their chemosensory hair array during the fast downstroke, carrying fine-scale patterns of concentration into the receptor area. This spatial pattern, blurred by flow along the antennule during the downstroke, is retained during the slower return stroke and is not shed until the next flick.

Many animals use chemical cues in the water or air around them to detect mates, competitors, food, predators, and suitable habitats (1–3). Large-scale turbulent flow in the environment carries odorants from a source to an animal's olfactory organ (such as an antenna or nose), while small-scale laminar flow near the organ's surface and molecular diffusion transport odorants to the receptors (2, 4). Turbulent fluid motion on a scale of meters to millimeters (5) determines the patchy intermittent structure of odor plumes in the environment (6); hence, chemical signals monitored at a point downstream from an odor source fluctuate in terrestrial (7, 8) and aquatic (9,10) habitats and in laboratory flumes (11–12). Recent attention has focused on the relation between the neural output of antennae and of the brain antennal lobe of moths in odor plumes (13) and on the neural processing of odorant pulses (14). We used lateral antennules of spiny lobsters, Panulirus argus, to analyze the critical first step in determining the spatial and temporal patterns of odor pulses arriving at receptors: the physical interaction of the olfactory organ with an odor plume. P. argus lateral antennules (Fig. 1A) bear rows of aesthetascs (hairs containing hundreds of chemoreceptor cells) flanked by larger guard hairs (15) and thus provide a system for investigating the design of hair-bearing olfactory antennae (16, 17).

Figure 1

Video frames of a lobster antennule flicking in a turbulent plume of fluorescent dye illuminated by a vertical sheet of laser light parallel to the flow direction in a flume. Image height, 31.4 mm. The aesthetasc-bearing lateral filament of the antennule is visible in (A) but not in (B throughD), in which an optical filter eliminates wavelengths other than the light fluoresced by the dye. The lighter the dye in (A), the higher the dye concentration. Concentration is coded by color in (B) through (D) (see scale in Fig. 2). The plume 1 m downstream from the source is composed of fine parcels of high dye concentration interspersed between layers of clean water. Recognizable features in the plume are carried downstream (that is, they move left) in sequence (B) through (D), which shows the antennule during a return stroke (B) and the next flick downstroke [(C) and (D)]. During the rapid flick [(A)and (C)], water carrying fine strips of dye penetrates into the array of hairs on the antennule. Because the ambient current transports dye along the antennule as it flicks, dye becomes spread along the antennule by the end of the downstroke (D). During the slower return stroke (B) and stationary pause, dye captured during the flick is trapped between the hairs, whereas dye in the ambient current does not penetrate into the array. Trapped water and dye are shed during the next rapid flick (C).

Fluid flow around a hair in an array depends on the relative importance of inertial and viscous forces, as represented by the Reynolds number (Re) (18). Because the fluid in contact with the surface of a moving object does not slip relative to the object, a velocity gradient develops in the flow around the object. The smaller or slower the object (that is, the lower its Re), the thicker this boundary layer of sheared fluid is relative to the size of the object. If the boundary layers around the hairs in an array are thick relative to the gaps between hairs, then little fluid leaks through the array. Hair arrays undergo a transition between nonleaky paddlelike behavior and leaky sievelike behavior as Re is increased (19–21). Although flow velocity has only a small effect on the rate of molecule interception by an isolated hair (4), our work with arrays of hairs showed that changes in velocity can have a profound impact on odorant encounter rates in the Re range in which the leakiness of the array is sensitive to speed (2, 22, 23).

Various crustaceans flick their olfactory antennules in the critical Re range for their particular aesthetasc spacing in which leakiness changes with velocity (17, 24, 25). High-speed kinematic analysis (25) and dynamically scaled physical modeling (17) of P. argusantennule flicking revealed that water flows through the aesthetasc array during the rapid flick downstroke but not during the slower return stroke. Flicking has been described as “sniffing”: a mechanism of enhancing odor penetration into the receptor area, as measured when isolated antennule preparations are hit with pulses of rapid flow (26–28). Although many modelers of odor tracking focus on the average properties of a turbulent plume, treating it as a diffusing concentration gradient (29), other investigators suggest that the fine-scale spatial patterns of concentration in a plume might provide information animals can use to locate the odor source (3, 8, 30). Whether animals have access to that information depends on how their olfactory organs disrupt plume structure. We studied how the hydrodynamics of antennule flicking in turbulent plumes physically alters the spatiotemporal patterns of concentration arriving at the aesethetascs.

We programmed a mechanical lobster to flick real P. argusantennules in a flume (31) in water flow with small-scale turbulence similar to that found in natural lobster habitats (25, 32). Fluorescent dye slowly released from the substratum 1 m upstream from the antennule mimicked odor leaching from a benthic source (33). High-speed videos of antennules flicking through a laser-illuminated plane in the dye plume (34) showed that the plume was composed of intermittent fine (∼1 mm thick) parcels of dye that only penetrated into the aesthetasc array during the flick downstroke (Fig. 1, A and C). By measuring in each video frame the dye concentration along a transect through the aesthetasc array parallel to the length of an antennule, a spatiotemporal map of the concentrations encountered by the aesthetascs along the antennule was constructed (Fig. 2). Such maps showed that the spatial pattern of concentration along an antennule changed during the leaky flick but not during the slow return stroke and stationary pause when the aesthetasc array was much less leaky. Thus, the antennule took a water sample during the flick downstoke and then retained it in the hair array (Fig. 1B) until the next flick (Fig. 1C). In addition, the maps revealed that high-frequency fluctuations in concentration were encountered in the hair array during the downstroke, whereas concentration remained relatively steady during the return and pause (Fig. 2 and Table 1).

Figure 2

Spatiotemporal map of dye concentration encountered by the aesethetasc-bearing portion of an antennule. The position along the antennule is plotted on the vertical axis, and time is plotted on the horizontal axis; concentration at each position at each time is indicated by color (yellow is 1% of concentration at the plume source; purple is 0%). A horizontal streak of color indicates that concentration at a position was not changing with time. When a rapid flick downstroke occurred (arrows), the spatial pattern of concentration along the antennule changed, but the new pattern was retained during the slower return stroke and the stationary pause.

Table 1

Mean temporal variance of concentration (% of source) measured at the midpoint of the aesthetasc-bearing section of an antennule for a duration of 44 ms midway during a flick, return, or pause (n = 83 flicks) (35).

View this table:

To test whether the temporal pattern in concentration variation encountered by the aesthetascs was due to changes in leakiness of the hair array, we compared the temporal variance of concentration at the midpoint of an antennule with that encountered by the midpoint of a virtual antennule that had no physical structure but that swept through the same portion of the plume at the same velocities as the antennule (34). Unlike the real antennule, the virtual antennule experienced high temporal variance in concentration during the return and pause as well as during the flick (Table 1). The temporal variance of the real and virtual antennules was similar during the leaky flick (35), indicating that the fine details of the plume structure did enter the hair array. In contrast, the temporal variance of the real antennule was much lower than that of the virtual during the nonleaky return and pause (35), indicating that the fluctuations in plume structure that occurred during those antennule behaviors were not sampled.

By the end of a flick, the hair-bearing lateral filaments of P. argus antennules had blurred the fine-scale structure of the odor plume. This blurring was due to shearing of the water in the aesthetasc array rather than to molecular diffusion (36). Even leaky arrays of hairs retain and shear some water near hair surfaces (2, 19–24). Because some dye was retained around the aesthetascs where parcels of dye flowed through the array, and because the ambient current transported those parcels of dye along the length of the antennule as it flicked (Fig. 1, B through D), dye from each parcel was spread along the antennule during the course of the downstroke. Therefore, by the end of the flick, the fine-scale spatial patterns of concentration in the plume had been smeared out in the sample within the aesthetasc array (Fig. 1D). The instantaneous spatial variance in concentration along the length of the antennule at mid-flick was much higher than at mid-return or mid-pause, whereas the spatial variance for the virtual antennule was always high (Table 2) (37).

Table 2

Mean spatial variance of concentration (% of source) measured at 210 evenly spaced points along the aesthetasc-bearing section of the antennule at the instant midway during a flick, return, or pause (n = 83 flicks) (37).

View this table:

Whether animals can use the fine-scale patterns of concentration in a turbulent plume to locate an odor source (3,7, 30) depends not only on neural filtering of that information but also on the fluid dynamics of their olfactory organs. The physical interaction of an olfactory antennule with the surrounding fluid is the first step in filtering the spatial and temporal patterns of concentration in the environment. Because P. argus flick their antennules in the Re range in which the leakiness of their aesthetasc array is sensitive to velocity, not only do they sniff (17, 25–28), but they also take samples of plume structure that are discrete in space and time. Fine-scale high-frequency structures of a plume enter the receptor area only during the rapid downstroke but become blurred by the end of the downstroke. The blurred sample is then retained between the aesthetascs until the next flick. The duration of the downstroke is ∼100 ms (25); although chemoreceptor cells in the anntenules of another species of lobster can respond to very brief (50-ms) pulses of odor, they must be exposed to a pulse for at least 200 ms in order to measure odorant concentration (38). Thus, although it is now clear how the hydrodynamics of an antennule filters the spatiotemporal map of concentration arriving at the receptors, which aspects of that information the lobsters use are still unknown.

  • * To whom correspondence should be addressed. E-mail: cnidaria{at}

  • Present address: Department of Civil, Environmental, and Architectural Engineering, University of Colorado, Boulder, CO 80309–0428, USA.


View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article