Odor identity influences tracking of temporally patterned plumes in Drosophila
© Krishnan et al; licensee BioMed Central Ltd. 2011
Received: 13 June 2011
Accepted: 27 June 2011
Published: 27 June 2011
Skip to main content
© Krishnan et al; licensee BioMed Central Ltd. 2011
Received: 13 June 2011
Accepted: 27 June 2011
Published: 27 June 2011
Turbulent fluid landscapes impose temporal patterning upon chemical signals, and the dynamical neuronal responses to patterned input vary across the olfactory receptor repertoire in flies, moths, and locusts. Sensory transformations exhibit low pass filtering that ultimately results in perceptual fusion of temporally transient sensory signals. For example, humans perceive a sufficiently fast flickering light as continuous, but the frequency threshold at which this fusion occurs varies with wavelength. Although the summed frequency sensitivity of the fly antenna has been examined to a considerable extent, it is unknown how intermittent odor signals are integrated to influence plume tracking behavior independent of wind cues, and whether temporal fusion for behavioral tracking might vary according to the odor encountered.
Here we have adopted a virtual reality flight simulator to study the dynamics of plume tracking under different experimental conditions. Flies tethered in a magnetic field actively track continuous (non-intermittent) plumes of vinegar, banana, or ethyl butyrate with equal precision. However, pulsing these plumes at varying frequency reveals that the threshold rate, above which flies track the plume as if it were continuous, is unique for each odorant tested. Thus, the capability of a fly to navigate an intermittent plume depends on the particular odorant being tracked during flight. Finally, we measured antennal field potential responses to an intermittent plume, found that receptor dynamics track the temporal pattern of the odor stimulus and therefore do not limit the observed behavioral temporal fusion limits.
This study explores the flies' ability to track odor plumes that are temporally intermittent. We were surprised to find that the perceptual critical fusion limit, determined behaviorally, is strongly dependent on odor identity. Antennal field potential recordings indicate that peripheral processing of temporal cues faithfully follow rapid odor transients above the rates that can be resolved behaviorally. These results indicate that (1) higher order circuits create a perceptually continuous signal from an intermittent sensory one, and that (2) this transformation varies with odorant rather than being constrained by sensory-motor integration, thus (3) offering an entry point for examining the mechanisms of rapid olfactory decision making in an ecological context.
Drosophila melanogaster is a cosmopolitan generalist that seeks the sources of diverse odorants [1, 2]. Tracking odors in flight is supported in part by bilateral spatial comparisons signal intensity across the antennae . However, due to fluid turbulence, the pattern of sensory exposure to odors in natural olfactory plumes also contains a high degree of temporal fluctuation [4, 5]. Olfactory receptor neurons (ORNs) and second order projection neurons (PNs) show temporal patterning which varies across receptors for a given odorant and across odors for a given receptor [[6–8]]. The identity of odors influences behavioral responses. During flight in Drosophila, behavioral responses to odors depend upon which specific ORNs are activated . The variation in temporal coding by ORNs coupled with the complex mapping of ORN input onto behavioral responses, would seem to pose a challenge to the tracking algorithm, yet might be mitigated in part by temporal fusion to enhance perceptual salience.
Neuronal mechanisms of peripheral olfactory coding are well studied in D. melanogaster . The odorant receptor (OR) repertoire selectively responds to a host of volatile chemical cues, which can be broadly classified as emerging from food, predators or a potential mating partner. Recent evidence suggests that these odorant receptors are comprised of heterodimeric cation channels, which could support the encoding of rapid odor transients . Electrophysiological studies in Drosophila have shown that ORs encode the temporal dynamics of an odor stimulus peaking between 1-10 Hz and falling off by approximately 100 Hz, thus defining an upper limit for the rate of information processing by the fly olfactory system . In locusts and moths, neuronal response dynamics mimic temporal aspects of the stimulus pattern [7, 13]. PNs in moth macroglomular complex follow stimulus pulses up to about 10 Hz . However, relative to spatial coding of the olfactory repertoire, virtually nothing is known about how the temporal dynamics of an olfactory signal are integrated into fly behavior.
What is the rate at which a temporally intermittent olfactory stimulus is perceived as continuous, and is therefore behaviorally tracked as if it were a steady stream? Temporal dynamics of sensory systems can be characterized by the Critical Flicker Fusion Frequency (CFFF), or the rate at which multiple discrete pulses of a stimulus are perceived as a continuous signal . In the case of lobster chemoreception the CFFF is 5 Hz, corresponding to a stimulus integration time of 200 ms . In the human visual system, the flicker-fusion frequency is wavelength-dependent, with longer wavelengths having lower CFFF .
We have sought to investigate how active tracking of a patterned odor plume varies with stimulus frequency and odor type. We used a virtual-reality plume simulator in which a magnetically tethered hungry fruit fly is free to rotate in the yaw plane of an arena capable of delivering precise visual and olfactory stimuli. By virtue of its design, the odor source in the arena is contained within a narrow low-velocity plume that does not elicit sensory adaptation, as there is no persistent odor field of constant concentration. In order to track the plume effectively, the fly must maintain its mean position within a narrow angular region of the arena, but makes small shifts in heading, back and forth. Although flies track a continuous plume of vinegar, banana, or ethyl butyrate equally well, pulsing these odors into intermittent plumes causes changes in tracking behavior that are unique for each odor. Thus, we have determined a CFFF threshold value for each odorant tested and found it to vary with odor identity in a manner qualitatively similar to the wavelength dependence of human visual flicker-fusion threshold.
To begin to examine the mechanism of the fly's CFFF, we made field potential recordings from the antennal funiculus in response to a temporally structured plume. Incidentally, this also served as a control for our stimulus delivery apparatus. We found that the relatively low behavioral flicker-fusion threshold values cannot be explained by the rather rapid olfactory transients that are encoded by the sensory system.
To determine the spatial and temporal properties of the odor plume we used a miniature photoionization detector (PID) capable of measuring volatile gas concentrations over a wide range of intensities with millisecond resolution. PIDs are used extensively in olfaction research to measure the fine scale temporal and spatial distribution of odor signals [[8, 12, 17]]. We sampled a spatial grid spanning the odor port in our behavioral setup to show that the area enveloping the head of the fly in flight receives the highest intensity of odor molecules (Figure 1B). We also used the PID to corroborate the fidelity of the solenoid switching system and confirmed that our apparatus is capable of delivering odor pulses at 1 Hz to 20 Hz that return to baseline intensity between each switching cycle (Figure 1C).
To facilitate robust tracking behavior, we delivered plumes at the highest gaseous concentration possible, from the headspace of undiluted solution. In odor field assays, particularly for apple cider vinegar, undiluted headspace vapor is highly aversive , yet in our apparatus, these headspace concentrations are always highly attractive. The apparent paradox is explained by the design of our olfactometer, which delivers a plume with a very low flow rate to isolate olfactory cues from mechanosensory upwind tracking reflexes common to insects such flies and moths [1, 20]. More importantly, the low flow rate also produces a spatially narrow plume that never envelopes both antennae with a constant high intensity signal, as experienced in odor field assays. Because the flow rate is low and the plume is narrow, we reason that the flies are therefore experiencing physiologically realistic odor concentrations, and if they weren't then we would expect to see some compromise to tracking or active repulsion as concentration is increased. Yet we have never observed improved tracking performance after reducing odor concentration maximum, but rather only degraded tracking which becomes not significantly different from water vapor control at a stimulus concentration diluted by two orders of magnitude (10-4) for all three odors tested (Figure 2D). The attractive feature of our design is that it elicits highly robust spatial tracking behavior - flies target the plume center with remarkable persistence and precision. The requisite compromise is that we do not produce odor field conditions and thus are unable to present typically attractive odors at concentrations sufficient to elicit repulsion.
Our results from the continuous plume show that each of the three test odorants (Figures 2, 3) influences the control of stable flight heading similarly. Having satisfied this crucial prerequisite, we next assessed the influence of temporally pulsing the different test odors. We first estimated the relevant test frequency range for each odorant by presenting a sweep of decreasing pulse frequency over the 25-second trial (additional file 1). We reasoned that as the plume became progressively more fragmented, the temporal integration of odor fluctuations would fail and tracking error would rise. We did this experiment only to coarsely estimate which frequency values to test in trials of fixed frequency.
From these data, we estimated the critical fusion frequency for each odorant, which represents the first stimulus frequency at which the mean deviation in heading differed significantly from the continuous plume. In other words, the critical fusion frequency for odor represents the first tested threshold pulse rate above which flies track the plume in a manner indistinguishable from the continuous stimulus. The critical fusion frequency for vinegar was 2 Hz, for banana was 0.5 Hz, and for ethyl butyrate was less than 0.15 Hz (Figure 4B). These results indicate that a flies' ability to integrate temporal intensity fluctuations within a plume varies with odor identity. Variation in the perceived intensity or vapor pressure of the different stimuli cannot explain these findings because the three odorants are tracked equally well when presented in a continuous stream.
We were intrigued by the extremely low-pass temporal characteristics of the ethyl butyrate tracking response (Figure 4A). Therefore, we next measured the influence of abruptly truncating the odor plume after three seconds of continuous exposure. We reasoned that an impulsive odor-off stimulus would reveal the dynamical behavioral latency to losing plume contact entirely.
Natural odor plumes are spatially and temporally intermittent [4, 17]. In this study, we used a magnetic tether system and a high performance olfactometer that presents an odor plume without mechanosensory wind cues to examine how fruit flies integrate a spatially restricted and temporally patterned odor stimulus for active plume tracking during flight (Figures 1, 2). We tested the hypothesis that tracking performance varies according to both the temporal frequency of exposure and odor type. We systematically varied the temporal pulse rate of each of three odors and, to our surprise, found that the precision with which a fly can track a patterned plume varies more across odor types than across temporal frequencies (Figure 4). One odorant, ethyl butyrate, produced tracking responses that far outlasted the stimulus (Figure 5). Active tracking of the 'phantom plume' cannot be attributed to the performance of the olfactometer (Figure 6), but can be attributed in part to olfactory enhanced optomotor responses that stabilize the fly's heading (Figure 7).
Since the navigating fly is likely to encounter odor blends rather than a single scent , a low-pass temporal filtering mechanism could act to 'smooth out' the unpredictable chaotic distribution of odor stimuli, and thus confer an adaptive edge to a navigating insect to fly straighter through a circuitous plume.
We were surprised by the finding that flies continue to orient toward the odor nozzle for many seconds after the ethyl butyrate plume is terminated (Figure 5). Several lines of evidence suggest that the representation of ethyl butyrate must persist in central rather than peripheral circuits. First, electrophysiological responses of olfactory receptor neurons sensitive to ethyl butyrate faithfully track rapid stimuli (Figure 6C). Thus, persistent tracking behavior cannot be explained by either residual flow from the olfactometer or receptor saturation. In many insects, neurons upstream of the primary receptors show long-lasting responses to olfactory stimuli. For example, similar long-lasting excitation effects have been observed in recordings from protocerebrum olfactory neurons in moths (Manduca sexta) where the response was found to persists 30 seconds beyond the stimulus offset , and in the same species where the duration of an excitatory burst from PNs to a pulse of pheromone exceeded the duration of the stimulus . Extracellular recordings made from the honey-bee brain have showed a rebound effect which lasts between 0.5 to several seconds after stimulus offset . This persistent neuronal response long after cessation of the stimulus has been proposed to be the mechanistic reason behind casting flight behavior - a long term pheromone mediated response which occurs after 400-500 ms after loss of odor in several moth species that persists when the moth moves into clean air . It is tempting to speculate that a similar mechanism could explain the persistency of the ethyl butyrate response after odor loss.
In addition, we have shown that during the tracking of the terminated ethyl butyrate plume, flies actively integrate visual cues, as their ability to track the plume is significantly diminished upon switching on a featureless uniform visual landscape (Figure 7A). These results are consistent with previous findings showing that a fly positioned within an active continuous plume requires rich visual feedback to maintain position at the plume . This results in part from odor enhanced sensitivity to wide-field yaw rotation . Here, we observe that the tracking persistence to the terminated ethyl butyrate plume is lost upon removal of yaw optic flow, suggesting that the same cross-modal mechanisms that are engaged to enable plume tracking persist beyond its termination. The fly continues to perceive and track a plume well after it has been terminated.
What would be the selective advantage of low versus high temporal fusion rates? A low fusion rate, such as is found for ethyl butyrate, could be advantageous when the distribution of odor signals is sparse and encountered infrequently, as would be the case far downwind from the source. In such a scenario, odor evoked changes in saccade rate and amplitude, coupled with upwind increase in flight velocity, would persist beyond the termination of the signal and combine to increase straight upwind flight and enhance the likelihood of encountering another odor packet in turbulent wind . The cost to sluggish temporal fusion is that signal loss is not detected for some time. Sensitivity to higher frequency intermittency, such as is found with apple cider vinegar, could be advantageous upon approaching the source of an odor where signals are densely distributed, and thus higher sensitivity to the rapid loss of a plume would rapidly modify saccade dynamics and reduce the likelihood of overshooting the source.
Ethyl butyrate has been shown to be released by ripe fruits that could offer potent cues for ovipositional preference . Thus, one could posit that flies might benefit from having a low flicker-fusion threshold (or a prolonged response after odor- loss) to ethyl butyrate thereby persevering within a sparse ethyl butyrate plume in order to maintain steady upwind heading and facilitate to reacquire a lost odor filament. It would be interesting to see if there are any differences in ovipositional preferences among the three odors tested. Positional repulsion to vinegar is mediated by the fly olfactory system [19, 34] whereas ovipositional attraction to acetic acid is mediated by the fly's gustatory system . Joesph et al, also postulate that fruit flies have an innate olfactory positional repulsion to acetic acid and that this repulsion is reversed only when flies need to oviposit . Thus, we reason that a high fusion rate for vinegar could inform a navigating fly about the proximity of a food resource (a rotting fruit for example), where the position aversional olfactory circuits have been overridden by the gustatory system in order for the fly to deposit eggs.
However, these evolutionary perspectives are entirely speculative. Our results conclusively show that for a temporally fluctuating signal, some odors are more easily tracked at a distance and others potentially better tracked near the source. Since the dynamical properties of fluid flow entirely determine the spatial distribution of components of a blend, odor signals are reduced in intensity as distance from the source increases, but the relative ratios of molecules within the blend do not vary . There is no physical mechanism by which some volatiles might be carried further from the source and therefore, there is no obvious way to explain a selective advantage of having differing behavioral sensitivity to the intermittency of different odors. Perhaps what we observe behaviorally bestows no clear evolutionary advantage, but rather results from the dynamical properties of olfactory coding that expand the perceptual odor space [7, 25], but at the same time place important constraints on the ability of flies to track natural plumes.
Our data provide new insights into the temporal processing of turbulent olfactory plumes of different fruit odors. Using a behavioral 'virtual-reality' assay, we have shown that the navigational ability of a fly to track a temporally intermittent plume is dependent on odor type, roughly analogous to human visual flicker fusion threshold which is dependent on the wavelength of the stimulus. Interestingly, flies display no innate preference between the specific odors tested under continuous odor conditions. We have defined a unique critical flicker fusion threshold value for each odor beyond which tracking behavior is no different than the tracking of a continuous plume of odor. Our study provides a framework for understanding the levels of processing of temporal information, as we show that downstream behavioral responses to intermittent odor stimuli are sluggish compared to peripheral processing as measured by antennal field potential responses.
Detailed procedures for constructing the components of the magnetic tether assay, odor delivery system, rearing of flies and acquiring video are described elsewhere in detail[[18, 35, 36]]. Briefly, adult female flies less than one week old were starved for 4-6 hours before being glued to a minutin pin (Fine Science Tools) and suspended between two magnets in such a way that allowed free rotation of the fly in the yaw plane. An electronic visual display panel system  consisting of LEDs enveloped the fly in azimuth, and occupied ±60 degrees in visual elevation with respect to the visual horizon. A gas multiplexer (Sable Systems) was used to switch mass flow regulated air precisely between two solenoid valves delivering odor or water vapor respectively at the rate of 7 ml per min. The frequency of odor pulses mentioned in the text and figures was controlled by custom code written in Matlab. The odor port was positioned less than 3 mm away from the head of the fly. A suction chamber (13 l/min) at the base of the arena ensured that the odor plume was drawn away quickly. Body heading of the fly in flight was illuminated with infrared and digitized using an infrared firewire camera (Fire-i) and analyzed offline using custom Matlab scripts. Odor stimuli consisted of apple cider vinegar (grocery store brand from Ralphs), natural ethanol extract of banana (Polarome Inc), and ethyl butyrate (99% pure, Sigma-Aldrich).
Each experimental trial began with the fly oriented within the plume. This was accomplished by projecting a 300 vertical bar on the LED display near the odor nozzle and oscillating the bar at 2 Hz for 10 seconds. Flies show a strong bar fixation reflex, and quickly respond to the bar motion by orienting directly toward the bar, at the odor nozzle (Figure 1a)
To measure the temporal properties of odor stimuli, we used a miniature photoionization detector (mini-PID Aurora Scientific, Ontario, Canada). The tip of the device inlet probe was placed at the location of the fly's head during flight, 2 mm from the odor nozzle. Ethanol, which has an ionization-potential of 10.62 eV, was used as a tracer gas.
Antennal Field Potential (AFP) recordings or electroantenograms (EAGs) were made from the third antennal segment of D.melanogaster as described previously . Briefly, a female fly was trapped in a truncated 200 μl micropipette tip in such a way that the olfactory organs were exposed. Recording and ground electrodes were filled with 0.17 M NaCl and the ground electrode was inserted into the head capsule. The recording electrode was brought into electrical contact with the dorso-medial surface of the funiculus and the odor delivery system consisted of solenoid valves controlled by custom written Matlab script that delivered ethyl butyrate pulses at the rate of 2.5 Hz alternating with water vapor. Signals were acquired at 10 kHz and filtered at 2 kHz using A-M systems amplifier (Model 3600).
MF is an Early Career Scientist with the Howard Hughes Medical Institute. This study was supported by NSF grant (IOS-0718325) to MF. We thank members of the Frye lab for helpful comments and suggestions. Omeed Paydar, Anita Chung and Daniel Niknam assisted with electrophysiological measurements.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.