2.2.1. Antennae and ORNs

As in vertebrates, odors are first transformed into a neural signal by the ORNs. In contrast to the vertebrate nasal epithelium, insect ORNs are not randomly distributed but are housed in morphofunctional units, the olfactory sensilla. The olfactory sensilla can take different shapes, appearing as long or short hairs, plates, pegs, or cavities. Odor molecules penetrate into the sensilla through numerous pores, a characteristic of the cuticular wall of the olfactory sensilla whatever their external morphology is. Insect ORNs are bipolar neurons whose distal processes (the dendrite) extend into the lumen of the hair or right below the surface plate, and whose axons project to the primary olfactory centers of the insect brain, the antennal lobes (ALs). According to their response spectra, insect ORNs can be sorted into functional classes that are housed in stereotyped combinations within functional types of olfactory sensilla.

Olfactory sensilla are concentrated on the main olfactory organs, the antennae, which may totalize several tens of thousands of ORNs. But antennae are not the unique site of olfaction on the insect body. Other head appendages, the labial and maxillary palps, can bear olfactory sensilla with various odor specificity depending on insect orders. The odor tuning of maxillary palp ORNs has been well described in mosquitoes (Syed and Leal 2007) and in Drosophila melanogaster (de Bruyne et al. 1999). Adult Lepidoptera possess reduced maxillary palps but their labial palps bear a labial pit organ housing neurons highly sensitive to CO₂ (Kent et al. 1986).

The challenges of detection of the huge chemical diversity of volatile molecules are met by a repertoire of ORs. Individual insect ORNs generally express a single functional type of OR (Couto et al. 2005) that is the determinant of their specificity. ORs vary in their degree of specificity, from narrowly to very broadly tuned, so even if ORNs express only one OR type they can show very different width of chemical tuning (Hallem and Carlson 2006; Hallem et al. 2004).

Insect olfactory receptors differ from those of vertebrates and belong to different families varying in their mode of coupling with transduction pathways. The first family of insect ORs that have been identified was thought to belong to the GPCR family. It is now admitted that insect ORs are composed of an odorant-sensitive protein belonging to the G-protein coupled receptor family (ORx), dimerized with a protein akin to GPCRs (Orco, formerly named Or83 in Drosophila) that have ion channel properties. Like ORx, Orco has inverted orientation in the cell membrane. This molecular complex serves both as a GPCR and an ion channel, allowing very rapid detection of high odor concentrations by means of the ionotropic pathway, and slower but prolonged and highly sensitive odor detection via the G-protein-mediated signal amplification (Wicher et al. 2008). A second family of receptors, first described in D. melanogaster, has been called ionotropic receptors (IRs). IRs are odor-gated ion channels that might have derived from ionotropic glutamate receptors (iGluRs) (Benton et al. 2009). IRs and ORs possess distinct ligand spectra in D. melanogaster (Silbering et al. 2011).

The sensillum lymph that bathes the ORN dendrites contains odorant binding proteins (OBPs), a class of extracellular proteins postulated to carry the lipophilic odorant molecules in a hydrophilic medium. A subclass of OBPs, the pheromone binding proteins (PBPs) bind specifically the pheromone components. Apart from acting as carriers, OBPs and PBPs might also contribute to the specificity of the detection by their selective binding of a fraction of the volatile compounds that enter the sensillum. The olfactory tissues also contain odorant degrading enzymes (ODEs) that catabolize the odor and other volatile molecules.

The transduction of the chemical signal into a neural code involves sequential steps. The binding of odorant molecules to their corresponding type of ORs leads to a change in the dendritic transmembrane potential called the receptor potential. The membrane depolarization opens voltage-gated ion channels involved in the generation of action potentials whose firing frequency constitutes the neural code of the chemical signal. Although several of the molecular actors and ionic mechanisms underlying this complex transduction process remain elusive or controversial, a reasonably consistent picture emerges from modeling (Gu et al. 2009; Kaissling 2001). The dynamic properties of this complex process are critical to the functioning of the system for its sensitivity as well as its capacity to follow rapidly changing odor stimuli. For more details, see Chapter 4.

2.2.2. Antennal Lobes, Primary Olfactory Centers

The axons of the ORNs form an antennal nerve that enters into a specific part of the insect deutocerebrum, the ALs (Homberg et al. 1989). The ALs are constituted of spherical condensed neuropil structures called glomeruli. The numbers of glomeruli vary between species. Each AL of the silk moth, Bombyx mori (Kazawa et al. 2009) and the sphingid Manduca sexta (Rospars and Hildebrand 1992, 2000) for instance, contains about 60 ordinary glomeruli, while circa 50 glomeruli have been identified in D. melanogaster (Laissue et al. 1999). Within a species, glomeruli have been shown to be invariant, with each glomerulus having the same shape, size, and location across individuals of same species, sex, and developmental stage (Rospars 1988). Investigations of the anatomical features and molecular receptive ranges of the central neurons associated with identified glomeruli in different species, and particularly in M. sexta, have established that glomeruli show odor tuning and are functionally significant structures (Hildebrand 1996). The functional specialization of glomeruli is striking in moths, whose ALs present a strong sexual dimorphism correlated with the high specialization of the male olfactory system in sex pheromone perception. As in other moth species, ALs of male M. sexta possess a cluster of identified glomeruli forming the macro glomerulus complex (MGC) in which project exclusively the ORNs tuned to the components of the sex pheromone blend (Hildebrand 1995, 1996; Rössler et al. 1998). A sexually dimorphic olfactory glomerulus has also been identified in female ALs (Rössler et al. 1998) and the neurons it contains are activated when the ipsilateral antenna is stimulated with a general odorant, linalool (King et al. 2000). Electrophysiological and imaging data indicate that the glomeruli present in males and females ALs (ordinary glomeruli (OG)) receive inputs from general odorant ORNs (Carlsson et al. 2002). OR expression and ORN targeting maps have furthermore established in D. melanogaster that the axons of the ORNs expressing the same OR type typically converge onto the same glomerulus (Couto et al. 2005; Laissue and Vosshall 2008).

Within glomeruli ORNs synapse with projection neurones (PNs) that transmit odor-evoked neural activity into higher regions of the brain, as well as with local neurones (LNs) whose arborizations do not leave the AL (Hansson and Anton 2000; Hansson et al. 1992; Homberg et al. 1989). The AL neurons are far less numerous than the ORNs. The total number of PNs in B. mori, for instance, is estimated to be about 430 for a total of 710 LNs (Namiki and Kanzaki 2011a).

In M. sexta dendritic arborization of the AL PNs are usually restricted either to OG or to MGCs showing maintained segregation between pheromone and general odor information. PNs innervating OG have arborization in a single glomerulus or are multiglomerular (Kanzaki et al. 1989). Most PNs are cholinergic (Homberg et al. 1995; Waldrop and Hildebrand 1989). In B. mori, most PNs innervating OG have extraglomerular ramifications (Namiki and Kanzaki 2011a), while those innervating sexually dimorphic glomeruli have little or no extraglomerular ramifications, suggesting different coding strategies for pheromone and general odors. Although a majority (75.6%) of them are uniglomerular in B. mori, PNs show diversity in the number of innervated glomeruli, their innervation pattern in the AL, their soma position, and the antennocerebral tract through which they project to higher olfactory centers (Namiki and Kanzaki 2011a). The PN and LN somata are clustered. There are two major cell clusters in B. mori, the medial and the lateral clusters, and one small cluster, the anterior cluster (Namiki and Kanzaki 2011a).

LNs branch exclusively within the ALs and interconnect glomeruli. In bees, homo-LNs uniformly innervate many glomeruli but hetero-LNs innervate one glomerulus densely and several sparsely (Meyer and Galizia 2012). LNs modulate the ORN input and the PN output activities within and between glomeruli. LNs are generally broadly tuned and morphologically diverse, innervating several glomeruli; 61% of the 360 LNs of M. sexta were identified as GABA-immunoreactive, (Reisenman et al. 2011). Most of the LNs in the AL of B. mori are also GABA-ergic (Seki and Kanzaki 2008). The dominant model of interglomerular interaction has long been “lateral inhibition,” serving a putative narrowing of the PN tuning (Olsen and Wilson 2008) and presynaptic gain control (Root et al. 2008). However, the dense AL network might be even more complex. Non-GABA-ergic LNs have been found in the ALs (Iwano and Kanzaki 2005). Accordingly, subsequent electrophysiological studies have established the existence of lateral excitation (Tootoonian and Laurent 2010). Presynaptic peptidergic modulation of ORNs has also been found in the ALs of Drosophila (Ignell et al. 2009) and the ORNs express a Drosophila tachykinin receptor allowing presynaptic inhibitory feedback from peptidergic LNs.

The olfactory information processed in the ALs is sent to the protocerebral areas via several fiber tracts. Although insects share a common plan of the antennoprotocerebral tracts (APTs), there are notable variations among taxa (Martin et al. 2011). There are five APTs in Lepidoptera (Homberg et al. 1988; Kanzaki et al. 2003). In turn, some protocerebral neurons project into the ALs (Homberg et al. 1988).

2.2.3. Second-Order Olfactory Areas

The second-order olfactory areas are situated in the insect protorocebrum and comprise the mushroom bodies (MBs), the lateral and superior protocerebrum (LP and SP), and the lateral horn (LH) (Galizia and Rössler 2010; Kanzaki et al. 1991).

The insect MBs have been particularly studied, notably in a comparative context (Strausfeld et al. 2009). In the MBs, PNs synapse with hundreds of thousands of intrinsic neurons, the Kenyon cells (KCs), in the calyces. The MB calyces are not homogeneous structures but contain multiple subsystems (Martin et al. 2011). The KCs, far more numerous compared to AL neurons, are targeted by AL PNs. KCs project to two regions called the alpha and beta lobes where they synapse onto small populations of “extrinsic” neurons. Behavioral and genetic experiments in D. melanogaster and the honeybee have revealed that the MBs are critical for associative learning of odor stimuli (Heisenberg 2003). Detailed analyses of the fine organization of the MB calyces and targeting of PNs in D. melanogaster (Jefferis et al. 2002, 2007) suggest that classes of KCs integrate information from a small group of PNs (Turner et al. 2008). Detailed data on the fine organization of the MB calyces as comparable to those in D. melanogaster are not yet available in other insect groups. In moths, functional domains are also visible in the MBs of B. mori (Namiki et al. 2013) with some segregation of pheromone information (Kanzaki et al. 2003). However, although the axons of the PNs that innervate pheromonal and nonpheromonal glomeruli are organized into concentric zones and show regional localization, the dendritic fields of KCs mixed pheromonal and nonpheromonal inputs (Namiki et al. 2013).

The LH of the PC appears in to be a diffuse aglomerular neuropil. However, PN inputs in the LH are stereotyped and segregated according to their originating glomerulus in D. melanogaster. Fruit odors are represented mostly in the posterior dorsal LH, whereas pheromone-responsive PNs project to the anterior ventral LH (Jefferis et al. 2007; Marin et al. 2002) showing persistent spatial organization in higher olfactory centers. Some regionalization can still be observed for the projections of pheromone-responsive PNs in moth LHs (Seki et al. 2005).

2.3.1. Quality Coding of Pheromones and General Odors

Female moths produce and release a sex pheromone that stimulates and attracts males from a distance. The sex pheromone is a relatively simple blend comprising a small number of components released in stable ratios. Each pheromone component is individually recognized by specialist ORNs (Pher-ORNs). Pher-ORNs are narrowly tuned to one component, which means that this component triggers excitatory responses at very low concentration although higher concentrations of related compounds also activate them. Pher-ORNs are housed in male-specific types of sensilla: long hair sensilla or trichoid sensilla. Pheromone sensilla generally house two Pher-ORNs, each tuned to one different component. The functional types of sensilla defined according to the chemical tuning of the ORNs they house are constant within a species. The distribution of the types of sensilla on the antennal segments is also constant within species. When stimulated with the proper pheromone component, Pher-ORNs show a phasic-tonic excitatory response, increasing their firing frequency in a concentration-dependent way. Pher-ORNs project into a specialized area of the male AL, the MGC (Christensen et al. 1995; Hildebrand 1996). The moth MGC comprises several subunits. Generally a larger cumulus is accompanied by several smaller glomeruli that vary in number and shape according to species and are called the toroid and the horseshoe in B. mori (Koontz and Schneider 1987). Pher-ORNs expressing different ORs project to different subparts of the MGC showing a spatial organization called odotopy (Hansson and Anton 2000; Hansson et al. 1992; Hildebrand 1996). In some moth species, a type of Pher-ORNs is specifically tuned to one component of the pheromone blend of a related species that, when added to the proper blend, inhibits behavioral responses. The detection of such interspecific antagonist reinforces sex pheromone specificity in sympatric species sharing some of their pheromone components. In the noctuids Heliothis virescens and Helicoverpa zea, these antagonists evoke excitatory firing activity in PNs restricted to a third glomerulus in the MGCs that is clearly distinct from the two glomeruli in which the attractive binary blend is represented in both species (Vickers et al. 1998). With its well-individualized lines, coding of sex pheromones is a near-perfect example of a labeled line type of coding.

How individual pheromone components and blends are represented in the MBs has not yet been fully elucidated. The spatial distribution of input and output neurons in the MB calyx was recently investigated in B. mori (Namiki et al. 2013). The authors compared the distribution of the presynaptic buttons of AL-PNs, which transfer odor information from the antennal lobe to the calyx. PNs for pheromone components and plant odors enter the calyx in a concentric fashion and they are read out by the dendritic fields of KCs. The axons of PNs that innervate the MGC are confined to a relatively small area within the calyx. In contrast, the axons of PNs innervating nonpheromonal glomeruli are more widely distributed. PN axons for the minor pheromone components cover a larger area than those for the major pheromone component and partially overlap with those innervating nonpheromonal glomeruli, suggesting the integration of the minor pheromone component with plant odors by KCs.

Host plant volatiles and other general odors comprise generally a higher number of components compared to pheromones. How are their components detected individually? Although less narrowly tuned than Pher-ORNs, GO-ORNs still show selectivity and are not as broad “generalists” as most vertebrate ORNs. Their molecular receptive ranges generally allow sorting of odorants according to their functional groups (Carlsson and Hansson 2006). Inside functional groups selective ORNs can be found that respond selectively to a more limited range of molecules (Shields and Hildebrand 2001). The general principles by which olfactory stimuli are represented in the ALs are nevertheless not so different in the pheromone and general odorant subsystems (Christensen and Hildebrand 2002). Functional types of GO-ORNs project into specific glomeruli in the AL. Calcium imaging has shown in several different species, including moths (Carlsson et al. 2002; Galizia et al. 2000), that more than one glomerulus is activated by a general odorant and the same glomerulus is activated by several odorants. All odorants, however, evoke unique patterns of glomerular activity that are reproducible at repeated stimulation within an individual. Intracellular recordings from the PNs specifically associated to one glomerulus are difficult to obtain and scarce. The PNs arborizing in one identified glomerulus of the AL of M. sexta were found to have a narrow receptive range, responding to cis-3-hexenyl acetate, while the propionate and butyrate homologues needed higher doses to elicit responses in the same glomerulus (Reisenman et al. 2005). In turn, these PNs were hyperpolarized by linalool, a compound that excited PNs in an adjacent glomerulus, providing evidence for lateral inhibitory interactions between glomeruli. Across fiber mode of coding and network functioning in the CNS are determinant in recognition of general odors. Disruption of GABA_A receptors in the AL increases discrimination thresholds in M. sexta (Mwilaria et al. 2008). By analogy with other sensory modalities, it has been suggested that lateral inhibition narrows the tuning of PNs in the ALs. However, PNs are more broadly tuned compared to their ORN input in Drosophila showing nonlinear transformation of the olfactory code (Bhandawat et al. 2007; Wilson et al. 2004). In various insect species odors are represented by very small assemblies of highly specific KCs realizing a sparse coding, while odor codes are compact into the ALs (Perez-Orive et al. 2002; Szyszka et al. 2005) indicating that integration of the olfactory information is far from being linear from the periphery to the higher centers. Sparse coding seems to be the rule in moth MBs too. KCs in M. sexta, show very low prestimulus activity; a defined olfactory stimulus triggers the firing of a few spikes mainly at stimulus onset, sometimes at its offset, from a few cells (Ito et al. 2008). Most of the KCs respond to only a few of the 21 compounds that were tested, although a small subset of them show broader tuning. Responses to general odor in the MBs are spatially distributed and vary over the stimulus course.

2.3.2. How Does Inhibition Contribute to Quality Coding?

In various insect species, certain odorants elicit an increase in the firing activity in some ORNs but a decrease in the firing frequency in other ORNs. Reciprocally, the firing activity of some ORNs is strongly increased by certain odorants but decreased by others. For instance, among a panel of 102 general odorants presented to the antennae of M. sexta females, some compounds elicited excitatory responses (measured as an increase of firing frequency) in some sensilla, but inhibitory (decrease in firing frequency) responses in other (Shields and Hildebrand 2001).

To which extent the inhibition of ORNs involves a specific inhibition pathway or results from the modulation of the excitatory pathway is still not clear. In insects, Dubin and Harris (1997) found that odor-modulated conductances differed among cells in semi-intact D. melanogaster antennae. Pupal and adult ORNs responded with increased as well as decreased action potential frequency and hyperpolarization with a concomitant decrease in conductance was observed (Dubin and Harris 1997). Odorant-induced suppression of inward transduction or voltage-dependent currents have been frequently observed in vitro with ORNs isolated from different vertebrate species. However, whether such effects play a role in natural odor coding may be questionable due to their generally poor specificity.

Inhibitions by odorants were observed in D. melanogaster after prolonged stimulation of ORNs with low concentration of their best ligand to increase their firing (de Bruyne et al. 2001). Linalool, but not citronellol, inhibited responses to sustained (25 sec) stimulations by ethyl acetate, for example. Due to the conditioning of cells with low concentrations of their excitatory odorant, the hypothesis that this type of inhibition results from molecular masking rather than an inhibitory pathway cannot be ruled out. An intracellular mechanism for inhibition of the response in mixtures has been described in rodents but not yet in insects. Certain odorants can activate the phosphatidylinositol 3-kinase (PI3K) pathway and induce the generation of phosphatidyl-inositol 3,4,5-triphosphate (PIP3). PIP3 is then able to inhibit primary signal transduction and calcium signaling, leading to inhibition of response to excitatory olfactory stimuli when two odorants are presented together (Wetzel et al. 2005). Alternatively, cases of inhibition by general odorants of the response to pheromone could be attributed to molecular interactions in several moth species. First, linalool and other terpenoids inhibit the firing response of Pher-ORNs in Spodoptera littoralis without hyperpolarizing the ORN when presented alone (Party et al. 2009). Second, plant odorants significantly inhibited the binding of pheromone to HR13, a pheromone receptor of H. virescens, expressed in a cell line (Pregitzer et al. 2012).

2.3.3. Intensity Coding

Odor intensity (i.e., the concentration of odorant molecules in the air) is important in many respects for insect olfactory communication. Intensity may not only provide information on the proximity and value of the source but also alter the specificity of detection, or it may even modify the valence of the stimulus. The dose behavioral-response curves to aggregation pheromones of mites are convex, no aggregation is observed at low doses, aggregation is observed at the optimal midconcentration, and pheromone compounds like neral elicit alarm and escape at the highest doses (Kuwahara 2004).

The intensity of the concentration of an odor decreases rapidly with the distance from the source and under stable conditions and short distance concentration gradients could provide orientation cues to a moving insect. Navigation strategies to olfactory cues may be different among species, and vary within a species between circumstances (Gaudry et al. 2012). However, due to the turbulent regime of dissemination of airborne signals, uniform gradients of concentration rapidly break into discontinuous filaments of odorized air at a relatively short distance from the source (Cardé and Willis 2008). Thus, the orientation strategy of flying male moths is based on the dynamic perception of events of finding or loss of the target odor filaments and seems well adapted to the intermittent nature of the pheromone plume (Cardé and Willis 2008).

The capacity of insect ORNs to code the concentration of stimulating molecules into a spike firing frequency over a surprisingly large range of concentrations is well documented. Dose response curves in ORNs present a typical sigmoid curve. Above threshold, the firing frequency increases linearly as the log of the concentration, and it reaches a plateau. Central neurons in the MGC of Agrotis ipsilon male moths respond to the pheromone blend with a biphasic excitatory-inhibitory response (Jarriault et al. 2009b). Pheromone stimulus intensity is encoded by the firing frequency, the number of spikes, and the latency of the excitatory phase. In turn intensity has no effect on the duration of the response. MGC neurons present a lower threshold compared to ORNs. This lowering of the threshold is generally considered as the direct consequence of the convergence of many ORNs onto a lower number of PNs.

Encoding odor intensity, however, interferes with that of stimulus identity. Specificity and intensity are indeed interdependent as chemical specificity decreases when concentrations of odorants increase. High concentrations of a poorly efficient compound may trigger the same level of firing activity in ORNs as a low concentration of a very efficient compound. Globally, the total afferent input to the AL increases with odor concentration. Broadening of glomerular activation with increasing doses has been shown in several species. In A. ipsilon, for example, more and more AL glomeruli show calcium activity with increasing doses of a specific plant volatile, heptanal, causing a loss of specificity in the representation of this odor (Deisig et al. 2012).

2.3.4. Coding of Signal Temporality

Because of the nature of the stimulus, olfaction is not a fast communication channel compared to either audition or vision. The timing of the stimulus largely depends on the transport of molecules by air movement and its turbulence (Murlis and Jones 1981), although sometimes emission might be pulsed at the source. Still, time is a more important parameter for olfaction than it could appear at a first glance. First, as underlined in the previous section, fast detection of the plume finding and loss events is crucial to moth orientation to pheromone sources. Second, temporal patterns of firing activity within neuronal networks are a component of the coding of odor quality.

The intermittent character of airborne odors due to the turbulence in carrying air and its importance for male moths flying in pheromone plumes have been acknowledged for a long time (see, for instance, Mafra-Neto and Cardé 1994; Willis and Baker 1984). Numerous studies in the wind tunnel with different species resulted in building a generally admitted strategy for orientation of male moths to sex pheromone in flight that involves a succession of upwind surges when encountering pheromone filaments and zigzag flight when losing odor plume (Cardé and Willis 2008). During flight, moth ORNs must be able to follow these fast changes in pheromone stimulus. The characterization of the dynamic of ORNs has primarily concentrated on their capacity to discriminate pulsed stimuli (Barrozo and Kaissling 2002; Bau et al. 2002). The upper experimental limits to repetition rate resolution are probably limited by the strong hysteresis in the olfactometers used to deliver olfactory pulses at high rates and the methods used to analyze the resulting ORN activity. Dedicated statistical methods showed that the neuronal activity in male antennae was able to follow repetition rates up to 33 pulses/sec (Bau et al. 2002). The intermittency of the stimulus is reproduced by the bursting activity of PNs in the ALs (Chaffiol et al. 2012; Lei et al. 2009). Central neurons in the AL of M. sexta can track pulsed odor up to 30 Hz (Tripathy et al. 2010). Disruption of the GABA inhibition in the AL suppressed the ability of PNs to produce bursting responses reproducing the intermittency of pulsed stimulation (Lei et al. 2009). Accordingly, GABA inhibition blocked male orientation to the pheromone in the wind tunnel.

The temporal performances of insect olfactory systems are also important for a fast discrimination of odors. Concurrent airborne odorants intermingle and fluctuate at fast time scales and the olfactory system needs to segregate odors from independent sources as different odor objects. Honeybees can segregate two odors presented with a 6-ms temporal difference showing that the temporal resolution of the olfactory system of insects is faster than previously thought (Szyszka et al. 2012).

To maintain these dynamic performances the odor molecules must be cleared off the antennae by odorant ODEs The enzymatic activity directly impacts the dynamic of the detection. For instance, ORN firing responses to cis-vaccenyl acetate were both stronger and longer in mutant D. melanogaster flies lacking the gene for an extracellular carboxylesterase, esterase-6 (Chertemps et al. 2012). In turn, responses to heptanone were not altered in mutants. ODEs should not only quickly inactivate the molecules of the specific signal (the half-life of a pheromone molecule has been estimated to be 15 ms), but also reduce the concentration of general odorants or xenobiotics that otherwise would alter the detection or be toxic to neurons with broad-spectrum biotransformation enzymes. Four categories of ODEs have been recognized (Vogt 2005): (1) soluble extracellular ODEs are present in the sensillum lumen (esterases, aldehyde oxydases, alcohol dehydrogenases, etc.) and show a diversity corresponding to the diverse functional groups of the pheromone components, (2) membrane-bound ODEs (epoxyde hydrolase), (3) cytosolic ODEs are multifunctional enzymes involved in the biotransformation of potentially toxic chemicals that can also attack odor molecules (for instance cytochrome P450 oxygenases, glutathionine-S-transferase), and (4) cuticular ODEs can degrade airborne pheromone and other odors while adsorbed on the waxy insect cuticle. Since these adsorbed molecules constitute second sites of odor release, their surface degradation might significantly reduce the chemical noise.

Insect ORNs show quickly decreasing response to prolonged or repetitive stimulation (Zack-Strausfeld and Kaissling 1986). The concentration of odorants probably stay at a high level for a long time in the antennae so that sensory adaptation might constitute a protection mechanism against overstimulation and improve response dynamics of ORNs. Adaptation is a form of plasticity at the neurone level that results from different cellular mechanisms affecting the transduction pathway, each involved in different types of adaptation (Zufall and Leinders-Zufall 1997, 2000). Although adaptation has been proven in laboratory conditions, how it contributes to the temporal shaping of the response to the low concentration levels normally faced in natural conditions is still unknown.

2.3.5. Temporal Codes

Temporality in the olfactory system is not only important for detection of the fluctuations of olfactory stimuli, but convergent data from different species show that temporal patterns of neuronal activities contribute to quality coding. Temporal coding means that the spike firing fluctuation carries specific information on odor identity beyond merely reproducing the temporal pattern of the stimulus (Wilson and Mainen 2006).

Compared to ORNs, the responses of PNs in the ALs present far more complex temporal patterns in the different insect species studied so far. The PNs show change in firing behavior with successive increases and decreases in the instantaneous firing rates, which often greatly outlast the stimulus duration. Many PN responses are inhibitory, with an inhibitory period followed by a period of increased firing. Furthermore, the time pattern of this offset activity is different among responses to different odors so it could contribute to odor coding in B. mori (Namiki and Kanzaki 2011b).

Odor-evoked synchronized oscillations of PN ensembles have been observed in several insect species as local field potential (LFP) oscillations (Laurent and Davidowitz 1994; MacLeod et al. 1998; Wehr and Laurent 1996). Synchronous firing among neurons is believed to facilitate the integration of signals in the olfactory neural network by creating a consistent representation of a given odor in the insect brain. Such oscillatory synchrony is an emergent property of the neural network and must be distinguished from a synchrony that simply reflects the coactivation of neurons by the same stimulus (Wilson and Mainen 2006). There is evidence that emergent synchronatory oscillations arise from dendro-dendritic connections between PNs and LNs in the locust ALs (MacLeod and Laurent 1996). In the absence of odors, moth PNs typically fire action potentials sporadically and their firing is asynchronous. In M. sexta, simultaneous recordings of the responses evoked by pheromone components from pairs of PNs innervating the same or different glomeruli within the MGC showed that PNs that branched in the same glomerulus and were activated by the same pheromone component also showed the strongest synchronization of their firing activity. Stimulation with a two-component blend evoked increased synchrony between intraglomerular pairs of PNs, supporting the idea that lateral inhibition between glomeruli shape the representation of pheromone blend (Lei et al. 2002).

The synchronized input from PNs is further processed in the protocerebron. There, contrasting to the high spontaneous activity and high probability of response of the AL-PNs, KC responses to odors in the locust protocerebron are rare and made of a small number of spikes lacking the temporal patterning of PNs (Perez-Orive et al. 2002). KCs integrate incoming spikes over brief, 50-ms time windows that are periodically reset by the oscillation cycle and are particularly sensitive to phase-locked spikes. Perez-Orive and colleagues concluded that KCs act as selective coincidence detectors from the fraction of AL-PNs that show synchronized activity over the staggered activity of PNs from which they receive input.

2.4.1. Measuring Interactions

Because of the interactions between its components, the amplitude of the response (response being either the firing of a single ORN or a very integrated behavior) to an odor mixture is not entirely predictable by the responses to the components presented individually. It can be less than (suppression) or greater than (enhancement) the value inferred from the responses to the individual components (Laing et al. 1984). The first methodological problem is to properly define how much is less or greater. Interactions may be synergistic or antagonistic. Synergy means broadly “working together” and antagonism means “working against each other.” However, these generally understood meanings imply quantitative criteria that do not receive general agreement (Berenbaum 1989). A review of the methods aiming to quantify synergy or antagonism is out the scope of this chapter. Since it is a general problem in pharmacology, biochemistry, or even ecotoxicology, useful information for olfaction may be found in papers from these research fields (Berenbaum 1989; Jonker et al. 2005). This difficulty in quantification is reflected by the great diversity of terms appearing in the literature. Many definitions have been proposed, such as hypoadditivity, complete additivity, hyperadditivity; synergy and inhibition; agonism, partial agonism, complete agonism, synergistic agonism. Because the relation between individual stimulus intensity and response is not linear and its parameters depend on the compound, calculating the response to a blend requires computing the equation from a dose-response function after measuring its main parameters (threshold, maximum response, slope), a very strict condition that is not often reached in most experimental studies (Duchamp-Viret et al. 2003).

It is worth a reminder that the purely physical mixture interactions arising at the source well upstream should not be neglected. Hygrometry, for instance, changes the vapor pressure of organic volatile compounds. Preevaporative effects have been reported for interactions with DEET (Syed and Leal 2008). Such effects must be taken into account when a precise quantification of the response to mixture is needed.

2.4.2. Mixture Effects at the ORN Level

Mixture effects are commonly observed at the level of single ORNs (for inhibitory interactions, see also Section 2.3.2). ORs are often differentially sensitive to a variety of odorants. Interaction is said to be competitive when two or more agonist odorants bind to the main receptor site and trigger receptor activation, although only one can be bound at a time. Noncompetitive effects may result from different mechanisms including one of the odorants binding to another site, which modifies the receptor properties at the main binding site. A high frequency of noncompetitive interactions were found in rat ORNs (Rospars et al. 2008), suggesting that such interactions play a major role in the perception of natural odorant mixtures. Competitive interactions may be discriminated from noncompetitive interactions because they usually can be counteracted by increasing the amount of agonist relative to that of the competitor. Recording the responses of ORNs to single compounds and their blend revealed a majority of mixture suppressions in various insect species (Carlsson and Hansson 2002; De Jong and Visser 1988; Hillier and Vickers 2011). Prevalence of inhibition was also found in response to binary mixtures compared to pure odorants in the cockroach Periplaneta americana during testing a series of aliphatic alcohols on the ORNs found in one type of sensilla housing generalist neurones (Getz and Akers 1997).

A second mode of interactions at the peripheral level results from the specific organization of the insect antenna compared to the olfactory epithelium of vertebrates. In insects, two or more ORNs with different odor tunings are cohoused into single functional units, the sensilla. This colocalization of neurones with different rather than similar sensitivities was postulated to be highly adaptative, serving a more precise ratio discrimination of pheromone blends (Baker 2009; Vermeulen and Rospars 2004). This seducing hypothesis was lacking experimental support for interactions between neighbor ORNs, however. Such interactions by edaphic coupling have been demonstrated only recently in D. melanogaster. A conjugation of electrophysiological and molecular biology approaches showed that the sustained response of one ORN is inhibited by transient activation of its neighboring ORN within the same sensillum (Su et al. 2012).

2.4.3. Recognition of Complex Plant Bouquets

Phytophagous insects localize their host plants through emitted odors (Bruce et al. 2005). Volatile blends differ between plant species both quantitatively and qualitatively. The specific combinations of compounds in these blends, many of which are ubiquitous, as well as their ratios, are assumed to drive host plant finding. Natural plant odors are complex blends of tens of components, most of them common to several plant species. The composition of these blends may greatly vary between individuals within a plant species according to its metabolism, physiological state, and developmental stage. Thus, to recognize and efficiently locate their host plants, herbivorous insects are faced with three problems: the complexity of the aroma, the ubiquity of components, and the variability of the composition, these last two compromising the specificity of the odor cues. About 120 different compounds are present in the volatile emissions of one major host plant of M. sexta, and female antennae possess ORNs able to detect 60% of them (Spathe et al. 2012). The majority of these ORNs are broadly tuned to a number of the host volatiles, but some ORNs respond to compounds specific to only one of the host plants. This indicates that their sensory equipment enable females to select plant species and plant quality for oviposition sites on the basis of olfactory cues (Spathe et al. 2012).

Adults of many moth species visit flowers for feeding on nectar, and flower volatiles may trigger innate behavioral responses. The floral bouquet from Datura wrightii flowers evokes foraging behavior in naïve M. sexta (Raguso and Willis 2002). To find out how behaviorally relevant floral mixtures are encoded in the olfactory system, Riffel et al. used a combination of chemical analysis of flower aroma, wind tunnel experiments, and multielectrode extracellular recordings in the ALs (Riffell et al. 2009a,b). Mixtures were more efficient than individual compounds in eliciting behavioral response, but although the whole floral bouquet comprised more than 60 compounds that varied in identity and concentration, moths responded with an equal frequency to natural flowers and a blend of only nine odorants. The behavioral responses to mixtures were consistent over a 1000-fold range in concentration. The ensemble analysis of the multiunit responses revealed a strong preference for only a small subset of the compounds emitted by D. wrightii flowers. Many single units were activated by one or more of the nine behaviorally active odorants. Single-unit responses to the blend were found to be suppression (11%), hypoadditivity (42%), or synergy (7%). The proportion of units exhibiting synergy or responding to mixtures was higher at high concentrations compared to responses to single odorants, but not at low concentrations. Spatial distribution, correlation coefficient, and Euclidian distance were calculated to compare the ensemble representations between different odor mixtures and single compounds. The representations for a mixture at lower concentrations were not statistically dissimilar to those of the single odorants, suggesting that the spatial distribution pattern of ensemble responses alone does not fully explain the behavioral activity of the mixtures relative to the single odorant or their consistency across concentrations. Neither the percentage of neurons that exhibited synergistic responses to the mixture, nor the percentage of responsive units alone could explain the behavioral consistency of the attractive mixture across the concentration range. In turn, stimulation with mixtures greatly enhanced the synchronous firing between pairs of neurons to form a distinct temporal activity that did not change over concentration range, suggesting that neural synchrony between certain cells encodes the mixtures in a spatiotemporal representation.

2.4.4. Interactions between Odorant Signals Released from Different Sources

The detection of a specific signal and its perception may be altered by odorants from the odorant background. Pheromones and plant odors provide a good model example for such interactions between signal and environment. The responses of male moths to sex pheromone are innate. However, it has been repeatedly observed that male moths are attracted in larger numbers to pheromone traps baited with a blend of synthetic pheromone plus some plant-related volatiles compared to pheromone alone (Light et al. 1993; Meagher 2001). They also fly more readily in the wind tunnel in response to such blends (Deng et al. 2004; Schmidt-Büsser et al. 2009; Yang et al. 2004). Their increased attraction to blends has been interpreted as evolutionary adaptive because of the high probability of finding sexually receptive females on host plants so that host odor provides supplementary cues to find a mate. Correspondingly, in Eupoecilia ambiguella, the synergistic effects were observed mainly at low or high dosages of the pheromone when orientation to males becomes more difficult (Schmidt-Büsser et al. 2009). The mechanisms of these interactions remain not well understood but several levels of the olfactory system are involved. Several authors have recorded and analyzed the firing responses of Pher-ORNs to mixtures of pheromone and plant odors with contradictory results. The Pher-ORNs of A. segetum were found to respond to high doses of plant compounds (Hansson et al. 1989). Synergistic effects, the firing response being increased in response to the main pheromone component plus linalool or cis-3-hexenyl acetate, have been observed only in the noctuid moth Helicoverpa zea (Ochieng et al. 2002). Most studies, however, conducted on other moth species using linalool or other plant compounds revealed a prevalence of antagonistic interactions (Kaissling et al. 1989; Party et al. 2009; Van der Pers et al. 1980). Stimulation of Pher-ORNs of the noctuid moth, Heliothis virescens, with mixtures comprised of the cognate pheromone component and either another pheromone component or a host plant volatile resulted most frequently in attenuation of the firing response (mixture suppression) (Hillier and Vickers 2011). The antagonism between linalool and pheromone was recently convincingly proven to arise at the OR levels in H. virescens (Pregitzer et al. 2012). In Spodoptera littoralis, antagonism between linalool and the main pheromone compound resulted in an increase in the capacity of Pher-ORNs to follow pulsed stimuli and the background was postulated to increase the capacity of the olfactory system to follow the fast changes in pheromone concentration during the moth flight (Rouyar et al. 2011).

Because pheromone and general odors interact mostly antagonistically in the antennae, the mechanisms to explain the increased behavioral responses to mixtures must be searched for in their integration in the brain. While the marked spatial separation between both types of inputs in the moth ALs has long been considered a sign that intraspecific (pheromone) and interspecific (plant volatiles) odor signals were processed separately, there is now evidence for some convergence in the ALs of different species (Kanzaki et al. 1989; Namiki et al. 2008; Trona et al. 2010). Among the multiglomerular PNs that were found in M. sexta, one of them arborized within the MGC (the pheromone-specific structure) and the OGs and showed responses to pheromone components and a plant volatile (Kanzaki et al. 1989). Calcium imaging (Deisig et al. 2012) and electrophysiological recording of pheromone-sensitive projection neurones in the MGC of male A. ipsilon (Chaffiol et al. 2012) showed that the response to the mixture pheromone + heptanal was generally weaker than to the pheromone alone, indicating a suppressive effect of heptanal. However, these neurones responded with a better resolution to pulsed stimuli. Conversely in B. mori, bombykol responses in PNs of the MGC were enhanced in the presence of Z3-hexenol, a host plant odor (Namiki et al. 2008). In contrast, the responses of PNs innervating ordinary glomeruli to Z3-hexenol were unaffected when bombykol was applied along with the plant odor.

2.5.1. Changes in Responsiveness to Olfactory Signals According to Physiological Status

It is advantageous to insects to reach maximal responsiveness to olfactory signals at the right time. For instance, the maximal responsiveness of a male moth to the sex pheromone should coincide to the development of its reproductive organs and the presence of receptive females (Gadenne et al. 2001). Similarly, female moths should be more sensitive to larval host-plant odors after mating, when they are physiologically ready to oviposit. Male A. ipsilon copulate only once in a single scotophase and mating induces a transient inhibition of the attraction to the sex pheromone in newly mated males (Gadenne et al. 2001), but does not affect their responses to plant odor (Barrozo and Gadenne 2010; Barrozo et al. 2010). In turn, mating turns on attraction to larval host plant-odor in female Lobesia botrana (Masante-Roca et al. 2007) or S. littoralis (Saveer et al. 2012). These changes in behavioral responses are correlated to changes in the responsiveness of the AL neurons to sex pheromones or plant odors (Anton et al. 2007). There is a reduction in the sensitivity to pheromone of AL neurons in male A. ipsilon; however, their sensitivity is not completely shut off. Newly mated males not only do not respond any more to pheromone and still fly toward a linden flower extract, but are also not attracted to mixtures of pheromone and flower extract at a high ratio of the pheromone (Barrozo et al. 2010), indicating a change in hedonic valence of the pheromone.

Besides these mating effects, an age-dependent plasticity has been described in moths. In spite of the conservative organization of the ALs, a certain degree of variability has been observed in the shape, volume, or location of their glomeruli in M. sexta (Couton et al. 2009; Huetteroth and Schachtner 2005) or S. littoralis (Couton et al. 2009), whose functional signification remains unclear (Guerrieri et al. 2012) but could result from the morphological changes that occur epigenetically in the olfactory centers of adult holometabolous insects. Persistent neurogenesis has been evidenced in the MBs of adult males and females, but not in their ALs (Dufour and Gadenne 2006). Hormones and neuromodulators are also involved in the modulation of the responsiveness during adult maturation. The sensitivity of AL neurons to pheromone increases with age and is dependent on the levels of juvenile hormone (JH) and octopamine in A. ipsilon (Jarriault et al. 2009a).

While in moths the regulation of olfactory sensitivity seems to operate mainly at a central level whereas the sensitivity of the periphery remains stable, female mosquitoes show a reduced sensitivity of their lactic acid ORNs after a blood meal (Davis 1984, 1986; Qiu et al. 2006). Correspondingly, a downregulation of a putative OR gene (Fox et al. 2001) or other olfactory genes (OBPs) has been observed (Biessmann et al. 2005). The sensitivity of functional classes of ORNs involved in the detection of cues for oviposition sites was not downregulated after a blood meal (Siju et al. 2010).

2.5.2. Plurimodality Interactions

While communication between sexes in moths and several other insect groups is largely moderated by olfaction as a dominant sensory modality, communication between sexes in other groups involves multimodality. Pentatomid bugs integrate both pheromone and vibratory signals in their mating communication system. In the green stink bug Nezara viridula long-range attraction has been generally attributed to the male emitted pheromone (Brézot et al. 1994), while short-range mate localization relies on substrate-borne vibration (Cokl et al. 1999). However, the emission of pheromones by the male is modulated by the female song (Miklas et al. 2003b). Bugs respond to a large range of component ratios of the pheromone blend so that integrating vibratory songs and pheromones contribute to increase the species-specificity of the communication between sexes (Brézot et al. 1994; Miklas et al. 2003a).

Pollinators show innate responses to colors and learn more rapidly to associate a reward with a color they innately prefer. Orientation to flowers in M. sexta combines visual and olfactory stimuli (Balkenius and Dacke 2010). Calcium imaging of responses to bimodal stimuli consisting of odor and color in M. sexta showed that color could either enhance or suppress odor-induced responses in the MBs. A blue stimulus suppressed the response to phenylacetaldehyde, a general flower scent (Balkenius et al. 2009). By contrast, it enhanced the response to a green leaf volatile (1-octanol). Hawk moths can learn combinations of stimuli, which suggests that they are capable of configurational learning (Balkenius and Kelber 2006).

Using multiple modalities offers several advantages to the insect. It can increase the detection of relevant objects against background noise and reduce the risks of confusion in a changing environment. It can also decrease the reaction time and response threshold. More generally, integrating information from different sensory modalities might indeed improve the reliability of olfactory signaling, a global context being often more specific than individual cues.

2.5.3. Plasticity in Response: Better Facing the Risks of Communication?

The context in which olfactory communication takes place also involves risks from predation or deception by concurrent organisms so that a certain degree in plasticity in the response may be advantageous. Innate and stereotyped communication systems may be subject to exploitation of perceptual biases by other organisms that will take advantage of the strong attractivity of an odor to the receiver for attracting it to their own and only advantage (deceptive communication). Plants in the Araceae and Orchidaceae families are known to use deceptive pollination by which they attract specific pollinators (Dafni 1984) by mimicking sex pheromones or food and prey signals (Brodmann et al. 2009; Stokl et al. 2010, 2011). Predators can also use sensory traps. Bolas spiders, for instance, produce volatiles mimicking a sex pheromone to lure and catch male moths (Stowe et al. 1987).

Communication makes animals conspicuous to their predators. Flying to pheromone sources exposes male moths to bat predation. Male moths react to bat echolocation sounds by escape maneuvers such as diving to the soil (Svensson et al. 2007). Males of A. segetum and Plodia interpunctella orienting towards a sex pheromone source in a flight tunnel were exposed to ultrasound mimicking the echolocation calls of a bat (i.e., high predation risk). Males of both species accepted the predation risk when attracted to a pheromone source of high quality (a female gland extract or the complete synthetic blend at high dose). In contrast, a lower proportion of ultrasound-exposed males than unexposed ones located the pheromone source when it was of low quality (an incomplete synthetic blend or the complete blend at low doses) (Svensson et al. 2004). Thus, when confronted with various contexts male moths make a trade-off between reproduction and predator avoidance depending on the relative strength of the perceived conflicting stimuli showing behavioral plasticity (Svensson et al. 2007).

2.6.1. Dealing with the Ubiquity of Some Chemical Signals: Context Dependence

Many chemicals are released by a large variety of organisms so that their presence alone would not allow an insect to reliably identify or even locate a resource. Yet, several of these ubiquitous compounds have proven to play key roles as resource cues, even by specialist insects. Carbon dioxide is naturally present in the atmosphere as a background, but nevertheless it plays multiple roles in insect foraging behavior (Guerenstein and Hildebrand 2008). Carbon dioxide is released into the atmosphere from various sources, for instance, respiration of living organisms or decomposition of plant matter. Variations in the rate of plant respiration lead to large intraday fluctuations of its atmospheric concentrations from 350 ppm during the day to up to 1000 ppm in dense vegetation at night. Atmospheric turbulences also modify the pattern of CO₂ concentrations. Nevertheless, 10 to 30 ppm bursts of CO₂ could be detected in natural habitats by a gas analyzer several tens of meters downwind from natural and artificial sources (Zöllner et al. 2004). Insects possess receptor cells highly sensitive to such changes of CO₂ concentrations in their antennae or their labial or maxillary palps. The CO₂ receptor cells of moths for instance can encode fast increases or decreases in the CO₂ concentration (Guerenstein et al. 2004). Upwind orientation in airstreams enriched with CO₂ have been demonstrated in different hematophagous insects, like mosquitoes, tsetse flies, and triatomine bugs, as well as in moths (Guerenstein and Hildebrand 2008). Differences in concentrations as small as 50 ppm above the ambient level are sufficient to trigger tsetse fly oriented flight (Evans and Gooding 2002).

Because CO₂ is everywhere, high sensitivity to CO₂ is not enough to make it a reliable resource indicator. The capacity to orient toward CO₂ enriched airflows is synergistically enhanced by organic volatile compounds released by the hosts in triatomine bugs (Barrozo 2004) and mosquitoes (Takken and Knols 1999). Sensitivity of female yellow fever mosquitoes to human skin odors increases significantly after a brief exposure to a plume of 4% CO₂ (Dekker et al. 2005). Dependence of the response to CO₂ on the odorant context has been established also in phytophagous moths and flies. Female M. sexta showed significant bias in approach of artificial flowers with above-ambient CO₂ only in the presence of volatiles of the leaves of their host plant, the tomato (Goyret et al. 2008). Experienced M. sexta when offered surrogate flowers without nectar do not maintain their spontaneous preference for flowers associated with high CO₂ level (Thom et al. 2004). Interestingly, the context might even modify the valence of one odorant compound for the insect. As a potential toxic, CO₂ has an inherent negative hedonic value to D. melanogaster flies that spontaneously avoid it. In turn, cider vinegar has an inherent positive value (attraction) (Faucher et al. 2006). But in a four-field olfactometer in which the flies could smell CO₂ in one field and some other odors or pure air in the other fields, their responses to CO₂ were found to depend on the odors presented in the other fields (Faucher et al. 2006).

2.6.2. Dealing with Variable Odors: Generalization

Showing a narrow specificity of responses to odors may be disadvantageous when the composition of the signal being not strictly controlled by the emitter varies unpredictably. Tolerance for ratio changes in host plant effluvia has been observed in moths. Mated females of the oriental fruit moth, Cydia molesta, are innately attracted to synthetic mixtures mimicking the natural blend of their host plants. Female attraction is maintained while the ratio of one of the constituents, benzonitrile, is increased up to 100 times (Najar-Rodriguez et al. 2010). Calcium imaging showed that odor evoked responses in one glomerulus of the female ALs mirror the behavioral effects of the manipulation of the benzonitrile ratio. A second glomerulus responded to changes in benzonitrile ratios, but small levels of benzonitrile in the mixture were inhibitory. These differences between glomeruli show that although the blend representation starts in the AL, final processing of the ratio must take place in higher-order olfactory centers. This relatively broad tolerance to the constituent ratios could offer the moth the advantage of recognizing its host plants across phenological stages in spite of quantitative and qualitative significant fluctuations in their volatile emissions.

The ability for an organism to learn that perceptually distinct olfactory stimuli lead to common outcomes is called generalization. Honeybees conditioned to individual compounds or to mixtures generalize their responses to stimuli belonging to the same chemical classes or with the same functional value (Sandoz et al. 2001). Components of alarm pheromones induce more generalization than floral compounds (Sandoz et al. 2001). Bees trained to respond to a binary odor mixture by being rewarded with sugar respond to novel proportions of the same compounds even when the interblend differences are substantial (Wright et al. 2008). The resulting generalization depends on the rewarding paradigm and the variability of the conditioned stimulus. This suggests that this outcome-dependent generalization is cognitive rather than perceptual. The bee brain can construct perceptual qualities of odors that depend, at least in part, on previous experiences. This capacity is necessary for maintaining sensitivity to interodor differences in complex olfactory scenes and it is not restricted to social hymenoptera. M. sexta conditioned to 2-hexanone or 1-decanol also responded to other alcohols and ketones and the generalization of the conditioned response decreased as a function of the chain length and functional group (Daly et al. 2001).

2.6.3. Sensitization and Learning

The basic knowledge on learning in insect and its neural basis comes first from the honeybee. The past decades have generated a wealth of novel research on the cognitive capabilities of bees and other insect species (see, for instance, reviews in Giurfa 2013; Menzel and Giurfa 2006; Menzel et al. 2006). Here I will focus on examples showing that experience can also modulate olfactory behavior generally considered as highly determined in nonsocial insects and in which structures in the olfactory systems are involved. Lepidoptera may be innately attracted to floral odors from flowers they feed on, but in addition they can learn new odorant-reward associations. Helicoverpa armigera moths trained to feed on flowers providing sugar sources that were odor-enhanced using phenylacetaldehyde or alpha-pinene showed a significant preference for the flower odor type on which they were trained (Cunningham et al. 2004). In turn, moths conditioned on flowers that were not odor-enhanced showed no preference for either of the odor-enhanced types. These results imply that moths not only show learned responses to floral volatiles, but they can also discriminate among odor profiles of individual flowers from the same species.

Like bees, moths respond to the presentation of sucrose by extending their mouth parts, which is called the proboscis extension reflex (PER). Effective associative conditioning of the PER occurs when the unconditioned stimulus (US), a sucrose reward, is proposed a few seconds after the onset of a sustained odor pulse, the conditioned stimulus (CS). Interestingly, associative learning has revealed experience-dependent plasticity in the ALs of M. sexta (Daly et al. 2004). PER was monitored by recording the feeding muscle activity by electromyography while recording neuronal activity in the ALs. More and more responsive neural units were found in the AL during conditioning and this neuronal recruitment persisted after conditioning. Recruitment occurred when odor reliably predicted food (CS+, not CS−, so that sensitization can be excluded). Conversely when odor did not predict food, a loss of responsive units was observed. This demonstrates that odor representations in the first olfactory centers show experience-dependent plasticity and are involved in olfactory memory.

Nonassociative learning, or nonassociative plasticity, is also largely present in insect olfactory-driven behaviors although its neuronal bases have received less attention than those of classical conditioning. The experience of a type of larval food at emergence influences flight orientation and oviposition choices of females in two species, Plodia interpunctella and Ephestia cautella, that do not feed as adults, (Olsson et al. 2005). Plasticity in the development of the olfactory system in relation to olfactory experience has been reported. For instance, deprivation from antennal sensory input causes changes of AL volume (Sanes and Hildebrand 1976). Males of S. littoralis briefly exposed to the sex pheromone are more responsive to it than naïve ones 27 hours after the preexposure. This increase of behavioral responsiveness is correlated with an increased sensitivity of AL neurons to pheromones (Anderson et al. 2003, 2007). Preexposure to linalool and geraniol, two volatile plant compounds, and to attractive and repulsive gustatory stimuli also modified the response level of males to pheromones (Minoli et al. 2012). The increased behavioral response to pheromone following brief exposure to a stimulus mimicking echo-locating sounds from predatory bats is accompanied by an increase in the sensitivity of AL neurons (Anton et al. 2011). These cross-modality effects have been attributed to a general sensitization (Minoli et al. 2012).

Before more recent studies showed that ALs are also involved in olfactory learning, their areas of projection, the MBs, have long been linked to memory and associative learning. Many gene products with roles in learning are expressed at high levels in fly MB neurons (Turner et al. 2008). MBs are required for acquisition, storage, and recall of olfactory memories in Drosophila and different lobes of the MBs probably have different roles in memory (Krashes et al. 2007). However, we are still far from a comprehensive view of the neural circuit involved in olfactory learning. In M. sexta stimulus time-dependent plasticity in the KCs does not constitute the odor representation that coincides with reinforcement (Ito et al. 2008). Odor presentations that supported associative conditioning elicited only one or two spikes on the odor’s onset (and sometimes offset) in a small fraction of KCs. Moths learned to associate individual odors with the reward even when unconditioned stimulus were temporally separated from the time window of KC firing by several seconds. These spikings ended well before the reinforcement was delivered.