In order to discriminate odor blends having slightly different ratios of similar odorants it is crucial that receivers bear sensory cells specifically tuned to each one of the components of the blends. Male moths have pheromone receptor neurons (PRNs) specifically tuned to each one of the two or three compounds that typically comprise the species-specific sex pheromone blend released by conspecific females . If females had to detect their own sex pheromone and discriminate it from similar ones, they would presumably require a pheromone detection system similar to that of males. However, males use the sex pheromone to locate females, whereas females presumably use it to detect the presence of other females , so the pheromone detection system of males and females may not need to be alike. We tested this hypothesis in the Oriental fruit moth G. molesta by means of electroantennography (EAG) and single sensillum recordings (SSR).
At the EAG level, which represents the combined activity of ORN receptor potentials across the antenna [5, 22, 23], we found that males were approximately 20 and 5 times more sensitive than females to the major (Z8-12:Ac) and minor (E8-12:Ac) pheromone components, respectively. This indicated that either the female antenna was equipped with a lower proportion of pheromone-responding cells than males, that these cells were not as sensitive as those of males, or both. Furthermore, the male antenna was about 10-fold more sensitive to Z8-12:Ac than to E8-12:Ac, whereas the female antenna had similar sensitivity to the two isomers, suggesting additional sex differences in pheromone detection.
In order to uncover the cellular mechanisms behind sex differences in pheromone perception, we tested the response of male and female ORNs to the sex pheromone, plant volatiles and the male-produced hair-pencil pheromone. Hierarchical cluster analysis grouped ORNs in 5 neatly separated response classes. ORN responses agreed with the EAGs because in males most of the cells sampled were major-component sex pheromone specialists, and correspondingly the male EAG was about 10-fold more sensitive to the major than to the minor pheromone component. In contrast, female ORNs responded very weakly and unspecifically to sex pheromone, and thus EAG responses of female antennae to sex pheromone were very small and did not discriminate between the two sex pheromone isomers. Thus, the smaller EAG response of female G. molesta to sex pheromone is not due to a smaller number of male-equivalent PRN than in males, but to the summed response of many unspecific ORNs with weak pheromone responses.
ORN responses to sex pheromone have been reported in females of several moth species . Two of them, Spodoptera littoralis (Boisduval) and Heliothis virescens (F.), have been studied in substantial detail. S. littoralis have long and short sensilla trichodea, but while the former occur abundantly throughout the male flagellomere (80 in each one), in females they are and are located exclusively on the lateral edges of the flagellomere, six on each side . In males about 80% (in a sample of 125 sensilla) of the long sensilla trichodea house a single ORN that responds specifically to the major pheromone component (Z9,E11-14:Ac), and the remaining 20% respond to minor pheromone components . In females 52 to 98% (in samples of 40 to 120 sensilla) of their few long sensilla trichodea house one cell that responds mainly to the major pheromone component, and dose-response curves show that it is as sensitive as the male PRNs [24, 25]. Molecular analysis supports physiological observations, showing a 50-fold higher expression receptor protein of the major pheromone component SlitOR5 in male than in female antennae , and also higher expression of receptor genes for minor pheromone components in male than in female antennae [26, 27]. By comparison, H. virescens males have many more long than short sensilla trichodea, and females only bear the short type . The majority of the long trichoid sensilla of males (80% in a sample of 403) house a neuron that responds specifically to the major pheromone component (Z11-16:Ald), whereas 3% of the sampled neurons respond specifically to the minor component (Z9-16:Ald). The remaining 16% respond to heterospecific pheromone compounds with somewhat lower specificity than the PRNs [29–31]. A small percentage of male ORNs in short sensilla trichodea (15% in a sample of 202) are major component specialists, and 38% respond to minor pheromone components with diverse specificity . In females, however, only 2% of the cells out of 184 recordings from the short sensilla trichodea responded to the major compound and 37% of the cells responded to the minor compound or heterospecific sex pheromone components, and in both cases the responses were weak and not too specific . Corresponding with the physiological data of H. virescens the expression of the olfactory receptor genes HvirOR13 and HvirOR6 associated with sex pheromone components is significantly larger in male than in female antennae .
Other moth species have been studied in less detail. A sample of 75 female Agrotis segetum (Denis & Schiffermüller) ORNs showed no pheromone responses , and in Manduca sexta (L) similar pheromone ORN dose-response curves were observed in both sexes, but it was to a minor component and only 8 cells out of 200 were of this type . Weak response to the major pheromone component 9Z,12E-14:Ac in one out of 36 female neurons of Cactoblastis cactorum (Berg) were recorded , and five Trichoplusia ni (Hübner) female sensilla harboring a neuron with similar threshold responses to the major pheromone component as male neurons have been reported, but the number of cells sampled was relatively small .
Altogether, S. littoralis (and perhaps T. ni) emerge as species where females appear to have true PRN equivalent to those of males, while in the other species studied, including G. molesta, the evidence of PRN in females is scarce or lacking, and in every case pheromone-responding ORNs occur in much smaller numbers in females than in males. It could be argued that females do not need as many, or as sensitive, PRNs as males to detect the presence of other females. However, females do not appear to have PRNs specifically tuned to each ingredient of the pheromone blend, and thus they should not be able to discriminate their own pheromone blend from similar ones, as males do. For example, S. littoralis females have an ORN type which sensitivity, as per dose-response curves, is equivalent to that of the major component PRN of males , but ORNs responding to the minor components, Z9-14:Ac and E11-14:Ac, have not been detected in females . Furthermore, female moths lack the behavioral antagonist ORNs that males of some species have in order to prevent cross-attraction between species that produce very similar pheromone blends . Thus, the overall evidence available today indicates that although females of some moth species appear to have major component PRNs, no females are endowed with ORNs tuned individually to both, the major and minor pheromone components, or the pheromone antagonists, and therefore females, unlike males, are probably unable to discriminate their own pheromone from those of related species. So, how can we explain the substantial evidence of pheromone autodetection in female moths? 
We believe that to unambiguously establish pheromone autodetection it should be first demonstrated that females are able to behaviorally discriminate their own sex pheromone blend from similar ones, otherwise they may just be detecting pheromone components irrespective of their taxonomic relationship. This procedure is standard in sex pheromone identification where males are presented with different blend ratios and often prefer the one of conspecific females , but as far as we know this approach has not been applied yet in pheromone autodetection tests . Secondly, many olfactory autodetection tests employ unnaturally high pheromone doses for prolonged periods of time, which may result in altered odor perception due to sensory adaptation or stimulation of non-target receptors. For instance, the behavioral tests that demonstrated pheromone autodetection in G. molesta employed 10 to 100 µg pheromone loads on rubber septa for up to 10 h on caged females [19, 20, 41]. These experimental conditions were chosen to emulate the environment that theses insects experience under mating disruption in the field. Because EAG responses (and the number of matings) were reduced 24 h after exposure [19, 20, 41], it is very likely that females experienced acute sensory adaptation and so their capacity to sense or discriminate sex pheromone was probably altered.
Most ORN dose-response curves typically show that, at natural concentration ranges, a given odorant stimulates a narrow spectrum of receptor neuron types, but at abnormally high concentrations additional receptor types (that would not typically respond to that odorant at natural doses) may also be excited [7, 42]. The displacement of the dose-response curve along the concentration axis represents the affinity between odorant and receptor (estimated with the ED50), while the height of the curve (i.e, the firing rate) indicates the maximum number of ORs activated in that ORN . A cell that is very specific for a given ligand will have a very low ED50 to it relative to other ligands, independently of the firing rate. For example, the dose-response curves of male G. molesta PRNs show that the ED50 of the Z- and E-cells is lower to their respective ligands than to the other one, but as the curves approach saturation the cells fire at similar rates and thus become less specific at the highest doses (Figure 2 and ). The same occurs in ORNs tuned to non-pheromone stimuli , so it is possible that autodetection experiments that use unnaturally high pheromone doses may stimulate non-pheromone neurons (like the unspecific neurons of G. molesta that we describe in here). This unspecific activation could falsely signal the brain the presence of non-pheromone odors which could then trigger behavioral responses that the sex pheromone at natural concentrations would never do. Furthermore, at high stimulus concentrations small impurities present in synthetic compounds may cause false positives, especially if they are more volatile than the target compound .
Calling females would be a more suitable source of pheromone stimulus in autodetection experiments than the synthetic blends, since they release a natural concentration of the optimal blend. Indeed, the calling period of target females is affected by neighboring calling females [45–47]. This change in behavior is very reasonable evidence of pheromone autodetection. Yet, as with synthetic stimuli, if conspecific and heterospecific calling females are not compared alongside then there is no definitive demonstration that females can discriminate their own sex pheromone from similar ones, and thus pheromone autodetection cannot be demonstrated.
Our study shows that G. molesta females have very sensitive and specific ORNs to stimuli other than their own sex pheromone. Testing compounds other than the sex pheromone in an adequate number of cells in both males and females was decisive in separating cell types with cluster analysis. Testing male and female ORNs together was also crucial considering the notorious methodological variation across olfactory setups , and it also allowed direct comparison of the established response of male ORNs to sex pheromone with the poorly characterized response of female ORNs. The clear separation of ORNs by stimulus type in both sexes indicates that the absence of sex-pheromone specific cells in females did not result from an inadequate sample size or faulty test stimuli. In male G. molesta 29% of the sensilla trichodea do not respond to pheromone , which is not far from the 38% that we report here. The presence of plant-volatile sensitive ORNs in males is believed to increase their probability of finding females, since these odorants have a synergic effect on male response to pheromone [14, 16, 17, 49, 50]. The low number of plant ORNs found in females probably resulted from testing only 3 plant compounds. Testing further ecologically relevant compounds will probably increase the number of cells responding to host odorants [14, 16]. Females also have a higher number of auricilic sensilla than males , and recordings from these sensilla would probably increase the number of plant responding ORNs in G. molesta males and females .
Our SSR experiment further revealed that 3% of the female sensilla are specifically tuned to ethyl trans-cinnamate, a close-range volatile that males release during courtship . In the noctuid moth H. virescens 33% out of 184 female ORNs responded to the male hair-pencil pheromone and to an interspecific behavioral antagonist . Interestingly, ORNs of a similar physiological type were also reported in males of H. virescens (19% out of 202 cells) [32, 52]. Conversely, we found no ORNs tuned to the male hair-pencil pheromone, ethyl trans-cinnamate, in male G. molesta. Male G. molesta produce additional hair-pencil pheromone components  that may disclose additional ORN types when tested.
Our study shows that the organization of the olfactory sensory system of male and female G. molesta appears to be shaped by the ecological function of each sex. Females and males seem to be relatively unable to detect their own pheromones but have ORNs that respond to the pheromone of the opposite sex and to ecologically relevant plant stimuli. Although PRNs have been described in females of other species, close scrutiny reveals that their physiological and biological relevance appears to be overrated. However, there is field evidence that female behavior can be altered under mating disruption conditions where the environmental level of pheromone surpasses the natural concentration that they themselves release . There are also indications that mating disruption delays the age at which females mate and this could reduce their reproductive output . Further studies on the effect of sex pheromone on female behavior are wanted to establish the effect of mating disruption on females, even if this effect is not mediated by pheromone autodetection sensu stricto.