Circadian Regulation of Olfactory Receptor Neurons in the Cockroach Antenna

Abstract

In the cockroach, olfactory sensitivity as measured by the amplitude of the electroantennogram (EAG) is regulated by the circadian system. We wished to determine how this rhythm in antennal response was reflected in the activity of individual olfactory receptor neurons. The amplitude of the electroantennogram (EAG) and the activity of olfactory receptor neurons (ORNs) in single olfactory sensilla were recorded simultaneously for 3–5 days in constant darkness from an antenna of the cockroach Leucophaea maderae. Both EAG amplitude and the spike frequency of the ORNs exhibited circadian rhythms with peak amplitude/activity occurring in the subjective day. The phases of the rhythms were dependent on the phase of the prior light cycle and thus were entrainable by light. Ablation of the optic lobes abolished the rhythm in EAG amplitude as has been previously reported. In contrast, the rhythm in ORN response persisted following surgery. These results indicated that a circadian clock outside the optic lobes can regulate the responses of olfactory receptor neurons and further that this modulation of the ORN response is not dependent on the circadian rhythm in EAG amplitude.

Introduction

Insect olfactory systems have been used extensively as a model to understand the cellular and molecular mechanisms of sensory transduction, sensory processing, olfactory behavior and learning and memory. Recently there has been an accumulation of evidence that shows the circadian system is involved in the regulation of insect olfaction. For example, in the fruit fly, Drosophila melanogaster (Krishnan et al., 1997; Tanoue et al., 2004), and in the cockroach, Leucophaea maderae (Page and Koelling, 2003; Rymer et al., 2007), measurements of the electroantennogram (EAG) over the course of the day have suggested the circadian system regulates olfactory sensitivity in the antennae. In addition, several studies have shown in several insect species including both fruit flies and cockroaches that behaviors that rely on olfactory cues exhibit circadian rhythms (e.g., Zhukovskaya, 1995; Liang and Schal, 1990; Kawada and Takagi, 2004; Barrozo et al., 2004; Zhou et al., 2005; Rymer et al., 2007) suggesting that the circadian modulation of olfactory receptors in the antennae may influence behaviors driven by olfactory cues. However, a critical problem for the interpretation of the physiological significance of EAG rhythms in insects is the lack of a clear understanding of the relationship between the EAG and the generation of the action potentials in the ORNs that relay olfactory information to the antennal lobes of the brain. While a variety of studies suggest that the ORN activity and EAG amplitude are correlated (e.g., Kapitskii and Gribakin, 1992; Mayer, 2001), there are indications that the EAG responses may, at least in some aspects of antennal receptor physiology, diverge from the ORN response as measured by the production of action potentials (e.g., Pophof, 2000; Grosmaitre, 2001; Zhukovskaya and Kapitskii, 2006). The picture is further complicated by the recent report that in D. melanogaster the amplitude of extracellularly recorded spontaneous action potentials in single sensilla exhibit a circadian rhythm, but neither spontaneous nor odor induced action potentials were found to vary in frequency over the circadian cycle even though there is a robust rhythm in EAG amplitude (Krishnan et al., 2008).

An additional reason to question the relationship between the EAG and ORN responses in cockroaches is that peak EAG amplitude was found to in the early subjective day, a time when the animal appears to be least active. Further, in the cockroach we have found the paradoxical result that although the EAG amplitude in response to stimulation with both food-related and sex pheromone components is at a minimum in the late subjective afternoon/early subjective night (Page and Koelling, 2003; Rymer et al., 2007), olfactory driven behaviors such as reproductive activity (Rymer et al., 2007) or olfactory learning (Decker et al., 2007) are restricted to this period of the day. The data suggest that interpretation of the physiological and behavioral significance of the circadian regulation of olfaction in insects will be critically dependent on understanding the role of the circadian system in the regulation of ORN responses. In the present paper we undertook long-term recording of individual olfactory sensilla in the cockroach antenna to determine the extent to which the temporal regulation of EAG amplitude by the circadian system is reflected in the activity of the olfactory receptor neurons. We show that the evoked response of the ORNs is regulated by the circadian system and that, like the EAG amplitude rhythm, the rhythm in ORN response peaks in the early subjective day. However, we also show that the circadian rhythm in ORN responses can be dissociated from the rhythm in EAG amplitude which appears to be independently regulated.

Materials and Methods

Animals

Experiments were performed using adult male cockroaches, Leucophaea maderae, raised in rearing boxes housed in an environmental chamber maintained at 25°C and in a light cycle consisting of 12h of light and 12h of dark (LD 12:12). Food (Nutro MAX, USA) and water were available ad lib.

Electrophysiology

Long-term recordings of odor induced electroantennogram (EAG) and olfactory receptor neurons (ORNs) responses were made simultaneously. Recordings were made from intact animals of two different light cycles, from animals following optic tract section and from animals whose compound eyes had been surgically ablated. To prepare for recordings, animals were anesthetized briefly with CO₂, placed them on a glass microscope slide (50mmx75mm) and held in position with tape that completely immobilized the animal. For recording the EAG, one of the antennae was placed under two Ag-AgCl wires, one at the base and the other near the tip (as in Page and Koelling, 2003). The wire attached to the base of the antenna also served as a reference electrode for ORN recordings. Electrical contacts with the antenna and recording wires were ensured using electrode cream (EC2, Grass Instrument). The slide was then fixed on the stage of a compound microscope (Olympus, SZX12) with antenna positioned below a 40X objective. The microscope was housed on a vibration isolation table (TMC Co.) in a light tight enclosure. A tungsten wire, electrolytically sharpened to a tip diameter of <0.5 μm, was used as recording electrode for ORNs response. The recording electrode was inserted at the base of the sensilla with the aid of a piezoelectric micromanipulator (Model PCS-5000, Burleigh Instruments, Inc.). Because of the difficulty of maintaining a stable recording from a single sensillum for several days, ORNs responses were all recorded from relatively large single-walled sensilla, which were blunt tipped, smooth and slightly curved, and 30–40 μm in length. Electrical signals from the electrodes were amplified (Gain: 100X, Bandpass: 300 Hz – 3 KHz), displayed on an oscilloscope and fed into a PC via a A–D converter (Model 1401 Plus, Cambridge Electronic Design). The preparation was maintained in constant darkness (DD) for 3–5 days. Data from animals in which both EAG and ORN recordings did not last at least 48 hours were not utilized. In general, recordings from intact animals persisted longer than those from animals that had been subjected to surgery. Within an hour or two after electrode insertion, spontaneous activity was virtually non-existent and all data analysis involved action potentials evoked in response to an odor stimulus.

Stimulus

Ethyl acetate (Sigma-Aldrich Chemical Co.) was found to be the most consistent odorant to induce EAG and ORNs responses and was used in all experiments as the stimulus odor. Stimulation was provided using an automated stimulus delivery system. The air stream was bubbled in a 1000 ml flask containing de-ionized water. After emerging from the tank the air stream was passed through a carbon filter and then into a solenoid valve that switched between two output ports. One port serves to provide a stream of clean air through a blank stimulus chamber (empty scintillation vial) during the inter-stimulus interval. On command from a programmable timer the solenoid was activated, directing the stream of air through the active stimulus chamber containing undiluted ethyl acetate. To direct the air stream/odor stimulus towards the antenna outputs from the blank and active stimulus chamber were combined to a single line into a 20 cm long glass pipette and positioned approximately 1 cm away from the antenna. The total flow rate was held constant at 200 ml/min. Odorant pulses 1.5 sec. in duration were given once per hour.

Surgery

To perform surgery, animals were anesthetized with CO2 and fixed on a specially devised plastic platform under a dissecting microscope. For optic tract section, a small square piece of head cuticle was cut open using a fractured razor blade to gain access to the brain. Special care was taken to avoid damage to the antennae, antennal nerves, and accessory antennal heart. After removing the ocellus and reflecting the trachea, the optic tract was visible and was cut with fine micro-scissors. Success of the surgery could be visually confirmed at the time of operation. After surgery, the head cuticle was replaced and sealed with low-melting point wax. After surgery animals were allowed to recover for 7–10days and then used for electrophysiology. For compound eye ablation (CEX), cuticle along the margin of the compound eye was cut and lifted to disconnect the optic nerves. Photoreceptors were carefully removed using forceps. After surgery the cuticle was replaced and sealed. CEX animals were either maintained in the original LD cycle or placed in a light cycle that was 12 hr out of phase from the light cycle they had been exposed before surgery.

Data analysis

Evaluation of EAG amplitudes and ORN responses were performed using Spike-2 software (Cambridge Electronic Design, Ltd.). For individual animals EAG amplitudes and number of spikes elicited by odor pulse was plotted against time. To examine averages activity, data were normalized as a fraction of the maximum value (EAG amplitude or number of spikes) over the entire recording period. Data were analyzed by Fast Fourier Transform-Nonlinear Least Squares (FFT-NLLS) method (Plautz et al., 1997, Izumo et al., 2006). FFT-NLLS permits assessment of the period and phase of the rhythm as well a measure of the degree of rhythmicity and is a particularly useful method for short, noisy data sets. Period values reported are those calculated for the periodic component with the lowest “relative amplitude error” (RAE) value in the circadian range (12–36 hours). RAE is a measure of the strength of the rhythm and can vary from a value of 0.0 for an infinitely precise rhythm to a value of 1.0 for a rhythm that is not statistically significant (Plautz et al., 1997). The determination of the presence or absence of rhythmicity was determined at a 95% confidence level. In order to evaluate the potential for false positives we reanalyzed EAG data from Page and Koelling (2003) obtained from animals in which the optic tracts had been severed. The EAG amplitude rhythm is abolished in these animals. Eight data sets (each 96 hours in duration) were analyzed by FFT-NLLS. In seven animals, no rhythmic component with a period in the circadian range (12–36 hours) was detected. In the remaining animal a rhythm was found with an RAE value of 0.955. All data for which we report a detectable rhythm in the current data returned RAE values of 0.8 or less and most (86%) were in the range of 0.2 to 0.7. In general rhythms that returned RAE values in the range of 0.2 – 0.6 were, subjectively, crystal clear (e.g., Figs. 2, 3). Independent estimates of rhythm phases were obtained by determining average of the time of peak activity on the second and third cycles of the rhythmic recordings. Because the durations of the recordings were relatively short and data points were collected once per hour, both the phase and period values should be considered as rough estimates. Graphical analysis was carried with SigmaPlot 10 software (Jandel Scientific) and data are presented as three-point moving averages. Statistical comparisons between groups were made using Sigma-Stat software (Jandel Scientific).

Fig. 2.

Circadian oscillations in the number of action potentials and EAG amplitude recorded from cockroach antennae in response to a standard pulse of ethyl acetate given once per hour. A and B plot the number of spikes recorded from a single olfactory sensillum and the EAG amplitude recorded from the same antennae. Animals were isolated in DD for the duration of the recording. The phase of the dark period of the prior light cycle (LD 12:12, lights on at 20:00 CST) is shown as a black bar on the first day. Both ORN activity and EAG amplitude exhibit circadian rhythms that freerun in constant darkness. C, D show the average, ORN activity (N=6) and EAG amplitude (N=5) following normalization with respect to maximum values. Error bars are standard errors that for clarity are only shown in the upward direction. Peak activity in both the ORNs and EAG occur in the early subjective day. E, shows the relative changes in mean amplitude of the largest spike in each hourly record over the first two days of recording. Over the 48 hour period, spike amplitude falls, on average, by about 20%, but there are no time-of-day effects.

Fig. 3.

Entrainment of ORN and EAG rhythms. Circadian oscillations in spike counts and EAG amplitude recorded in DD from animals that had been entrained to LD 12:12 (lights on at 08:00) shifted 12 hours from the light cycle used to entrain the animals from Fig. 2. A, B show data plotted from a single individual. C, D show average, normalized activity in ORNs (N=6) and EAGs (N=7) recorded from 7 animals. Peak activity is again in the early subjective day for both ORN and EAG rhythms.

Results

Circadian rhythm in olfactory receptor neurons

We recorded both the EAG and the ORN response from an individual olfactory sensillum in the same antenna of restrained animals maintained in constant darkness. Generally the ORN recordings initially appeared to be from one, two, or rarely, three cells. Fig. 1 shows traces of both EAG and spike activity from a single sensillum at 1 h, 48 h, and 96 h after the beginning of recording. As was typical, the amplitude of the action potentials tends to decline with time (likely due to changes in the recording electrode over time). Since the waveform of individual action potentials was not stable over the long duration of these recordings, it was not possible to sort out single units on the basis of amplitude and shape with any confidence, though in a few recordings it appeared there was a single action potential of large amplitude while others in the record were too near baseline noise to be reliably counted - e.g. Fig. 4E. Even in these cases, we could not unequivocally claim the recording was from a single unit throughout the duration of recording. Since we could not reliably identify single units, responses reported were based on counts of all action potentials that were clearly above baseline noise within each record.

Fig. 1.

Shows stability over time of recordings from a cockroach antenna. In each record, the top trace shows impulse activity from a single olfactory sensillum while the bottom trace records the EAG from the same antenna. A, is the recording at 1 hour; B, 48 hours; and C, 96 hours after the recording was begun. Stimulus in all cases was a 1.5 second pulse of ethyl acetate. The stimulus artifact (arrow) marks the onset of the stimulus.

Fig. 4.

Correlation between EAG amplitude and ORN response. A and C, plots of EAG amplitude (solid lines) and spikes (open circles) in response to hourly pulses of ethyl acetate. B and D, plots of the number of spikes as a function of EAG amplitude. Spikes and EAG are correlated, but there is significant variability. E, examples of ORN recordings at 48 hours and 96 hours for the antenna from C – only the large unit was counted.

We obtained long-term recordings from single sensilla in which we have recorded the response to a pulse of ethyl acetate given once per hour. Fig. 2A shows plots of impulse counts of multi-unit activity from a single sensillum from an animal isolated in constant darkness (DD) from LD 12:12 (lights on at 20:00 Central Standard Time; CST). Although the record is somewhat noisy, there is clear evidence of circadian rhythmicity with peaks and troughs occurring at about the same time each day. The phase of peak activity was at 23.5 CST, 3.5 hours after subjective dawn. The period was 23.9 h. Fig. 2B plots the EAG amplitude rhythm that was recorded simultaneously with the ORN activity. The period of the EAG rhythm was estimated at 24.2 h with the phase of peak amplitude at 23.5.

We obtained similar recordings that lasted at least 92 hours from six animals. ORN recordings from all six of the animals exhibited circadian rhythms (average RAE = 0.488 ± 0.140, mean ± SD) while EAG records showed statistically significant circadian rhythms in five of the six animals (RAE = 0.370 ± 0.172). Fig. 2 C, D shows the average spike count and EAG amplitude as a function of time. Values of spike counts were normalized with respect the peak value in each individual’s record. The average phase for the animals was 22.4 ± 1.35 CST (mean ± SD) for the ORN peak activity and 21.5 ± 2.97 CST for the EAG amplitude maximum. In three cases the ORN rhythm appeared to phase lead the ERG rhythm while in two other cases the ORN rhythm was phase lagging. For both ORN and EAG rhythms, freerunning periods were of the primary component were near 24 hours (ORN: 24.6 ± 0.90; EAG: 25.1 ± 1.67). Neither the phases nor the periods of the ORN and EAG rhythms were significantly different (p > 0.05, t-test). The data clearly indicate that the ORN activity, like the EAG amplitude (Page and Koelling, 2003), shows a freerunning circadian rhythm with both rhythms peaking in early subjective day. The amplitudes of the ORN rhythms were modest with activity peaks compared to preceding troughs in the first complete cycle representing, on average, a 39 % increase in activity.

In contrast to observations in Drosophila (Krishnan et al., 2008) robust rhythms in the amplitude extracellularly recorded ORN spike, there was no clear rhythm in the amplitudes of the extracellularly recorded ORN action potentials in the cockroach. Beginning at midnight of the first full day of recording, we measured the amplitude of the largest action potential in each of the six animals’ hourly records over the next 48 hours. With very few exceptions, the differences in individual spike amplitudes from the mean amplitude for either the first or second 24 h period was less than ± 20% (compared to a 2–3 fold daily excursion of amplitude found in Drosophila; Krishnan et al., 2008). Further, we were unable to detect any dependence of amplitude on time of day either by visual inspection of individual records or by statistical analysis (p > 0.50 for both the first and second day, ANOVA).

Entrainment

If the rhythms we recorded are truly circadian rhythms, they should be subject to entrainment by environmental light cycles. We obtained records of ORN activity and EAG amplitude in constant darkness from seven animals taken from a light cycle that was shifted by 12 hours (lights on at 08:00) from the light cycle used to entrain the animals shown in Fig. 2. Six of the seven animals exhibited circadian rhythms in spike activity (RAE = 0.481 ± 0.132) and EAG amplitude was rhythmic in all seven animals (RAE = 0.495 ± 0.198). The peak activity of the rhythms occurred in the early subjective day of the new LD cycle (Fig. 3). Average phase of maximal response of the ORN rhythms was 11.8 ± 0.93 CST and for the EAG rhythm it was 12.3 ± 2.29 CST. The differences in phases of the rhythms from the two light cycles (Fig. 2 and Fig. 3) were highly significant (p < 0.001). The results indicated that the phases of the freerunning ORN and EAG rhythms were entrained by the prior light cycle and show that the daily oscillations in spike activity and EAG amplitude were not a response to some unidentified artifact in the recording environment.

Relation between EAG amplitude and the number of action potentials

Impulse activity in the olfactory receptor neurons tracked the EAG amplitude reasonably well in most animals. Fig. 4 shows two examples from the group of animals from LD 12:12 (light on at 08:00) of ORN activity plotted overlaid with corresponding plots of EAG amplitude. The scales on the axes were adjusted to match the EAG amplitude values with the number of action potentials in the graph. In the top record (Fig. 4A) the recording from the sensillum, which was strongly rhythmic, involved at least two and probably three units while the bottom record in which the rhythmic pattern is less clear, appeared to be of a single unit (Fig. 4C, E). Regressions of the number of action potentials as a function of EAG amplitude for the two records are shown in Fig. 4B, D. In both cases the correlations were highly significant (as might be expected for any two periodic functions with similar periods). However, there is considerable scatter in data. Overall, the level of temporal correlation between the response of the olfactory receptors and the EAG in individual animals was not particularly strong, with a range of r² values of 0.002 to 0.380. There was a significant correlation in 7 of the 12 animals in which both the EAG and ORN were rhythmic, in five animals the correlation between EAG amplitude and ORN impulse activity did not reach the level of statistical significance. One potential explanation for the weakness of the correlation is that spike initiation and EAG amplitude are at least partially independent with regard to temporal regulation. This suggestion is confirmed by results reported in the next sections.

Effects of destruction of the circadian pacemaker in the optic lobe

Severing the optic tracts is known to abolish the rhythm in EAG amplitude and suggests that the optic lobe pacemaker is responsible for driving the rhythm in olfactory sensitivity in the antenna (Page and Koelling, 2003). This hypothesis leads to the prediction that optic tract section should also abolish the rhythm in ORN response recorded from single sensilla in the antenna if the ORN responses underlie the EAG. We recorded ORN activity and EAGs in DD from four animals in which the optic tracts had been severed bilaterally. Remarkably, while the rhythm in the EAG was abolished, the rhythm in ORN spike activity persisted in all four animals (RAE = 0.541 ± 0.208) (Fig. 5). The peak activity of the ORNs was in the subjective day and the rhythm had a similar amplitude and phase to the intact controls with peak activity occurring at 11.8 ± 1.56 CST. The results indicate that the ORN rhythm is driven by a circadian clock that is outside the optic lobes. The data further indicate that the circadian variation in nerve impulse activity recorded from a single sensillum can be dissociated from temporal variation in the EAG signal recorded from the whole antenna.

Fig. 5.

Effects of severing the optic tracts on rhythms in ORN response and EAG amplitude. Recordings were made in DD 7–10 days after surgery. A, B show data plotted from a single individual while C, D show average, normalized activity in ORNs recorded from 4 animals. Despite the loss of rhythmicity in EAG amplitude the response of the ORNs persisted after surgery. The results indicate that the circadian rhythms in EAG and ORN responses are controlled by anatomically separate circadian mechanisms.

Discussion

The results presented here show that the impulse activity of the olfactory receptors in the cockroach antennae is regulated by a circadian clock. In intact animals this activity tracks (temporally) the EAG amplitude reasonably well with the peak ORN activity occurring in the early subjective day. Surprisingly, however, we found that optic lobe ablation disrupts the EAG amplitude rhythm, without any clear effect on the ORN spike frequency rhythm. One potential explanation of this result is that the rhythms of the individual ORNs rapidly become desynchronized following destruction of the optic lobe pacemaker so that the population response (EAG) is no longer rhythmic. This seems unlikely since the ORN rhythms remained roughly in phase and average activity of the cells we recorded from is clearly periodic. However, we only recorded from a small subset of the population (e.g., cells innervating large, single-walled sensilla) and it is possible, though we think unlikely, that desynchronization among sensilla could account for the results.

The dissociation of the EAG from the ORN response raises questions about the relationships that exist among the EAG amplitude, ORN activity, and olfactory behavior that bear further investigation. A dissociation of the EAG amplitude from the number of spikes evoked in olfactory receptors is also evident in fruit flies where the EAG shows a robust circadian rhythm while the frequency of action potentials exhibits no significant dependence on circadian phase (Krishnan et al., 2008). One thing seems clear from the data on cockroaches, fruit flies (Krishnan et al., 2008), and moths (Pophof, 2000; Grosmaitre, 2001) is that the EAG amplitude is not a reliable indication of the vigor of the response in the individual ORNs and therefore is not necessarily informative about the olfactory information that is transmitted to the central nervous system. The explanation of the divergence of the amplitude of the EAG and action potentials evoked by an odor stimulus is not certain. It has been shown that the cockroach EAG is generated entirely in the antenna and with our method of stimulus and recording appears to be due exclusively to physiological processes associated with odor reception (antennal mechanoreceptors are not involved) (Page and Koelling, 2003). It also does not involve post-synaptic or other potentials in the CNS (Page and Koelling, 2003). However, the origin of the EAG in the antenna is not well understood. It is a field potential that likely involves elements of currents associated with ORN receptor potentials, ORN action potentials, and the transepithelial potential difference between the sensillar lymph and the hemolymph as well as the compartmental and antennal resistances associated with these currents (Kapitskii and Gribakin, 1992). Sorting out the relative impact of circadian modulation on these individual elements will be a complicated business to pursue using electrophysiological techniques and it seems likely that a molecular approach (e.g., Tanoue et al., 2008) will be much more fruitful for deciphering the targets and mechanism of circadian regulation of olfactory responses.

The fact that the ORN rhythms persists in animals without optic lobes also raises the question of where the circadian pacemaker(s) for the rhythms is located. The loss of the EAG rhythm with optic tract section confirms a previous report that suggested that the EAG rhythm was regulated by the optic-lobe pacemaker (Page and Koelling, 2003). It was further suggested that the control may be via a humoral link since there are (as yet) no known efferent connections from the central nervous system to the antenna. Clearly this hypothesis is not viable with regard to the ORN rhythm. An obvious alternative is that, like D. melanogaster (Tanoue et al., 2004), a circadian oscillation is generated locally in the antennae and this oscillation controls the responsiveness of the olfactory receptors (though apparently not the EAG). A homologue of the fruit fly period gene has been identified in another cockroach, Periplaneta americana (Reppert et al., 1994) and has been found by RT-PCR to be expressed in the P. americana antennae (Brown and Page, unpublished) lending some credence to the suggestion of an antennal clock in cockroaches, but additional investigation is needed to determine if the expression exhibits a persistent rhythm. Alternatively, it may also be possible to develop a culture system to record ORN responses in isolated antennae to determine if a rhythm persists in the antenna in vitro.

At this point the functional significance of the circadian rhythm in olfactory sensitivity remains uncertain. Thus far attempts to correlate temporally the olfactory rhythms in the antennae with rhythms in olfactory driven behaviors in cockroaches have not provided any useful clues. For example, the phase of the circadian rhythm in reproductive activity appears to be independent of the circadian rhythm in olfactory response to sex pheromones. A recent study has shown EAG amplitude in response to components of the sex pheromone of the male peak in the late subjective night/early subjective day while the copulatory activity is restricted to the late subjective day/early subjective night, and there was no evidence that the rhythm in EAG amplitude had any significant effect on the timing of reproductive activity (Rymer et al., 2007). These observations are consistent with the demonstration in the cockroaches Periplaneta americana (Zhukovskaya, 1995) and Supella longipalpa (Liang and Schal, 1990) that attraction of sex pheromone is highest in the early subjective night. Similarly the ability of L. maderae to learn to approach a non-preferred odor over a preferred one in a differential olfactory conditioning paradigm was restricted to the early subjective night, though there was no circadian phase-dependent difference in the naïve animal’s ability to distinguish between the two odors or to perform the task once it has been learned (Decker et al., 2007). The results suggest that the behavioral responses related to olfaction are largely independent of the rhythmic responses of the primary olfactory receptors to single odors. It is worth noting that olfactory behavior will be dependent in most cases on the combinatorial effects of multiple odors that will initially emerge during central processing in the antennal lobe (Lei and Vickers, 2008). Thus it may be naive to expect a straight-forward predictive relationship between the ORN response amplitude and behavior.

In contrast, in fruit flies it has been shown that the avoidance of aversive odors exhibits a circadian rhythm that follows the rhythm in EAG amplitude such that flies are more likely to avoid the odors in a T-maze test during the middle of the subjective night during the inactive portion of their circadian cycle when EAG responses in the antennae are highest (Zhou et al., 2005). In addition, the olfactory system appears to play a role in regulating the nocturnal courtship activity in male/females pairs (Fujii et al., 2007). Thus certain aspects of fly olfactory behavior appear to conform to expectation based on the temporal modulation of EAG amplitude. On the other hand, the discovery that the EAG amplitude rhythm is not reflected in the frequency of odor evoked action potentials from the ORNs in Drosophila (Krishnan et al., 2008) but is temporally related to a change in extracellularly recorded amplitude of spontaneous spiking activity (Krishnan et al., 2008) does not readily lend itself to a straight-forward interpretation of the link between EAG rhythms, ORN rhythms, and olfactory behavior. In summary, at this point the only thing that seems clear is that olfactory responses are regulated by the circadian system in, at least, these two divergent species of insect. What remains uncertain is how this might be related to the circadian modulation of olfactory behavior.