Unique neural coding of crucial versus irrelevant plant odors in a hawkmoth

  1. Sonja Bisch-Knaden  Is a corresponding author
  2. Michelle A Rafter
  3. Markus Knaden
  4. Bill S Hansson
  1. Max-Planck-Institute for Chemical Ecology, Department of Evolutionary Neuroethology, Germany
  2. CSIRO Health and Biosecurity, Australia

Abstract

The sense of smell is pivotal for nocturnal moths to locate feeding and oviposition sites. However, these crucial resources are often rare and their bouquets are intermingled with volatiles emanating from surrounding ‘background’ plants. Here, we asked if the olfactory system of female hawkmoths, Manduca sexta, could differentiate between crucial and background cues. To answer this question, we collected nocturnal headspaces of numerous plants in a natural habitat of M. sexta. We analyzed the chemical composition of these headspaces and used them as stimuli in physiological experiments at the antenna and in the brain. The intense odors of floral nectar sources evoked strong responses in virgin and mated female moths, most likely enabling the localization of profitable flowers at a distance. Bouquets of larval host plants and most background plants, in contrast, were subtle, thus potentially complicating host identification. However, despite being subtle, antennal responses and brain activation patterns evoked by the smell of larval host plants were clearly different from those evoked by other plants. Interestingly, this difference was even more pronounced in the antennal lobe of mated females, revealing a status-dependent tuning of their olfactory system towards oviposition sites. Our study suggests that female moths possess unique neural coding strategies to find not only conspicuous floral cues but also inconspicuous bouquets of larval host plants within a complex olfactory landscape.

Editor's evaluation

This article is of particular interest to researchers in the fields of neuroecology of insect olfaction and of insect–plant interactions in general. The authors investigate the olfactory signals that guide the specialist hawkmoth Manduca sexta towards plants that are used for oviposition and for nectar-feeding in a natural setting. How insects distinguish useful information from irrelevant information is an important question. The authors use elegant chemical ecology techniques and recordings of neuronal activity to ask how female moths (M. sexta) could discriminate co-occurring behaviorally relevant versus irrelevant plant and floral volatiles.

https://doi.org/10.7554/eLife.77429.sa0

Introduction

Nocturnal insects largely rely on their sense of smell to locate food sources and oviposition sites. However, preferred nectar sources or suitable host plants are often rare, and odors emitted by these essential plants are mixed with bouquets released by neighboring plants (Bruce et al., 2005). Plants that depend on nocturnal pollination, for example, by hawkmoths, advertise their nectar-providing flowers with bright colors, and most notably by a strong scent to attract their nectar-feeding pollinators (Raguso et al., 2003a; Raguso et al., 2003b). Evolution has thus formed a system that greatly facilitates the location of nectar sources by foraging insects. The situation is very different when gravid females search for a suitable host plant for oviposition. Vegetative parts of plants, probably to remain cryptic to herbivores, emit only trace quantities of volatiles that might be difficult to identify against the olfactory background provided by other plants (Turlings et al., 1995). At the same time, leaf damage by insect herbivores leads to an increased emission of volatiles, sometimes attracting parasitoids and predators of these insects (Pare and Tumlinson, 1999). Herbivore-induced plant volatiles are often similar across plant species (Mumm and Dicke, 2010), and herbivores are omnipresent in natural environments. Together, these facts suggest that female moths searching for an oviposition site encounter either undamaged, olfactorily unremarkable plants or damaged plants with a more conspicuous volatile profile, components of which are shared across many non-host plants. In any case, gravid moths have to identify suitable host plants against a vast odor scenery provided by the background vegetation. For this purpose, insects might depend on taxon-specific volatiles released by host plants but not by surrounding plants. An example is the cabbage moth Plutella xylostella that uses host plant-specific isothiocyanates to locate cruciferous hosts (Liu et al., 2020). Most insects, however, seem to identify host plants by blends of ubiquitous components that are present in plant-specific ratios. This has been shown in experiments where small changes in the composition and ratios of crucial odor blends had a huge impact on the behavior of insects (Cha et al., 2008; Kárpáti et al., 2013; Riffell et al., 2014; Visser and Avé, 1978; Webster et al., 2010). Furthermore, the chemical composition of crucial blends and the capability of the insect’s antenna to detect specific components of these blends have been studied in detail (Conchou et al., 2017; Fraser et al., 2003; Tasin et al., 2010). At the level of the first olfactory processing center in the insect brain, the antennal lobe, previous studies have investigated the temporal coding patterns evoked by floral bouquets. By impaling a restricted region of the antennal lobe with a multiunit probe, simultaneous recordings from a mixed population of local interneurons and projection neurons in this area were performed (Riffell et al., 2009a; Riffell et al., 2009b). Using the same approach, inhibitory interactions between these neurons could be studied in addition (Lei et al., 2004). However, this recording technique does not allow assigning functional significance to individual olfactory glomeruli, which are the morphological and functional subunits of the antennal lobe (Gao et al., 2000; Hansson et al., 1992). An analysis of the spatial coding patterns, however, is possible via functional calcium imaging. Although this technique does not inform about temporal coding patterns or inhibitory interactions, it was used in different insect species to provide a detailed insight into the spatial representation of natural odor blends across the glomerular array (Burger et al., 2021; Lahondère et al., 2020; Saveer et al., 2012; Schubert et al., 2014; Zhao et al., 2020). Usually, only odor blends that are known to be essential in the ecology of the insect were tested as the aim of those studies was to reveal how crucial blends are coded in the brain. However, it remains unclear how the olfactory systems of insects can differentiate between crucial and irrelevant blends, that is, how peripheral detection and central representation allow the identification of food sources and oviposition sites within a complex olfactory environment.

In our study, we collected headspaces of focal plants and background vegetation in the habitat of the tobacco hawkmoth Manduca sexta in Southern Arizona. Volatiles of hawkmoth-visited plants, like of most vegetation, differ between day and night both regarding floral (Hoballah et al., 2005; Raguso et al., 2003b) and leaf emissions (De Moraes et al., 2001). We collected plant headspace only during the night as we were interested in how the nocturnal M. sexta would detect and process these olfactory cues.

The primary nectar sources for M. sexta in the Southwestern United States are flowers of Agave palmeri and Datura wrightii. Volatile emissions of these two species are strong but very different from each other (Raguso, 2004; Raguso et al., 2003a; Riffell et al., 2008), and their pollen together account for 90% of the pollen load on the proboscis of M. sexta (Alarcón et al., 2008), a measurement that can be used as a proxy for flower visitation. Presence of pollen from Mirabilis longiflora and Mimosa dysocarpa on the moth’s proboscis and nighttime observations reveal that these plants are additional secondary nectar sources in the same habitat (Alarcón et al., 2008; Grant and Grant, 1983). Datura, in addition to being a valuable nectar source for M. sexta, is one of its two larval host plants in the area. Datura plants thus have to interact with an insect that is at the same time an important pollinator and a damaging herbivore (Bronstein et al., 2009), enabling the moth to find an oviposition site by navigating towards the scent of nectar-providing flowers emitted by the same plant (Reisenman et al., 2010). The other local host plant of M. sexta larvae is Proboscidea spp., the only known host belonging to a non-solanaceous family (Mechaber and Hildebrand, 2000). Flowers of Proboscidea, however, are not visited by foraging hawkmoths, that is, Proboscidea plants are suffering from leaf consumption by M. sexta larvae but do not profit from pollination by ovipositing moths. Furthermore, we sampled odors from another 11 native, frequent plants in the direct neighborhood of M. sexta’s focal plants. These background plants have no documented relevance for M. sexta; they included flowering herbaceous plants, nonflowering woody shrubs or trees and tufts of grass. Three of the background plants, the desert willow Chilopsis linearis, the sunflower Helianthus annuus, and the wild grape Vitis arizonica, are larval hosts of other sympatric hawkmoth species (Table 1).

Table 1
Headspace collections from plants at the Santa Rita Experimental Range in Arizona (US).
Plant species (plant family), common nameType of sampleNectar source for adult M. sextaHost plant for M. sexta larvaeLarval host plant for sympatric hawkmothsNocturnal pollination
Agave palmeri (Asparagaceae), Palmer’s century plantFlowerXX
Datura wrightii (Solanaceae), Sacred daturaFlowerBranchXXX*X
Mimosa dysocarpa (Fabaceae), VelvetpodFlowering branchXX
Mirabilis longiflora (Nyctaginaceae), Sweet four o'clockFlowering branchXX
Proboscidea parviflora (Martyniaceae), Devil’s clawFlowering plantX
Chilopsis linearis (Bignoniaceae), Desert willowBranch with seedsX
Helianthus annuus (Asteraceae), Common sunflowerFlowering plantXX
Vitis arizonica (Vitaceae), Wild grapeBranchX§
Amaranthus palmeri (Amaranthaceae), CarelessweedFlowering plant
Argemone pleiacantha (Papaveraceae), Prickly poppyFlowering branch
Baccharis salicifolia (Asteraceae), SeepwillowBranch with buds
Gutierrezia sarothrae (Asteraceae), SnakeweedFlowering plant
Poaceae spp,, GrassTuft of grass
Prosopis velutina (Fabaceae), Velvet mesquiteBranch
Quercus emoryi (Fagaceae), Emory oakBranch
Senna hirsuta v glaberrima (Fabaceae), Woolly SennaFlowering plant
  1. *

    M. quinquemaculata.

  2. M. rustica, M. florestan.

  3. M. muscosa.

  4. §

    Eumorpha achemon.

After collecting all nocturnal plant headspaces in situ in the field, we proceeded to analyze this comprehensive chemical database. We then used the plants’ headspaces as stimuli in physiological experiments with female M. sexta. Specifically, we investigated which components of the volatile blends the moth’s antenna can detect, and how the glomerular array of the antennal lobe is coding these complex odor bouquets. An insect’s reaction to olfactory cues is known to be plastic in relation to its physiological condition and experience (Gadenne et al., 2016). The moths tested in our study were laboratory-reared on artificial diet, naïve to plant odors, not fed, and tested only once, as we were interested in the insects’ innate neuronal responses. However, M. sexta has been demonstrated to differentially respond to plant odors depending on its mating status (Mechaber et al., 2002). Underlying this differential response is a state-dependent modulation of the olfactory system, which may take place at the level of the antenna, the brain, or at both levels (Gadenne et al., 2016; Saveer et al., 2012). Therefore, we investigated the peripheral detection of plant headspace and the central representation of this olfactory information in both virgin and mated M. sexta females.

Our results revealed that the olfactory system of female moths responds strongly to odors related to nectar sources. Suitable oviposition substrates elicited much weaker but specific responses, a specificity that was most pronounced in gravid females. Evolution thus seems to have shaped an olfactory system that allows efficient feeding at all stages and that enables the mated female to pinpoint an optimal home for her offspring.

Results

Nocturnal emissions of plants in the habitat of M. sexta

We collected the nocturnal headspaces of 17 plant species at the Santa Rita Experimental Range, our study site in Southern Arizona (Figure 1A, Table 1). Headspace samples were analyzed chemically by gas chromatography coupled with mass spectrometry (GC-MS). We first evaluated the number of GC-peaks per sample as a proxy for the number of volatile compounds present. In 10 of the 17 plant samples, the number of emitted compounds was in the range of blank control collections (Figure 1B, gray area). The richest volatile bouquets, on the other hand, were emitted by the sunflower Helianthus, and by M. sexta’s nectar sources Datura flower, Agave flower, and Mirabilis. When we considered not only the number of GC-peaks but also their chemical identity, the same four bouquets revealed distinct chemical profiles. Headspaces of the remaining plants were statistically distinctive but largely overlapping due to low emission rates and shared volatiles, which were also present in the blank control samples (Figure 1C; one-way ANOSIM, R = 0.67, p<0.0001; Bray–Curtis similarity index).

Chemical analysis of nocturnal headspaces collected from plants in the habitat of M. sexta in Southern Arizona.

(A) Representative photographs (left) and chromatographs (right) of each headspace collection. x-axis of chromatographs, retention time; y-axis, abundance, same scale for all headspaces, maximum abundance indicated in Datura flower headspace; gray bar, internal standard (5 ng 1-bromodecane). (B) Number of GC-peaks. Squares, average values of 3–5 individual plant samples; whiskers, range; dotted line and gray area, average and range of control values obtained from nocturnal collections in the same habitat with empty bags (n = 2), and with unused filter material (n = 1); open squares, within control range; filled squares, outside control range. (C) Non-metric multidimensional scaling plot (Bray–Curtis, 2D stress: 0.09) based on a nontargeted analysis (https://xcmsonline.scripps.edu; Tautenhahn et al., 2012) of 69 chromatograms (Figure 1—source data 1). Color code of plant samples as in (B).

What does the moth detect?

So far, our analysis considered the chemistry of nocturnal plant emissions. However, M. sexta might still be able to detect plant volatiles occurring only in trace amounts but having a high biological significance, for example, to identify an appropriate oviposition site. Therefore, we performed GC-coupled electro-antennographic detection (GC-EAD) using the antennae of female M. sexta as biological detectors. This technique allows successive presentation of headspace compounds in naturally occurring concentrations to the moth antenna and, in parallel, recording of the pooled response of all antennal olfactory sensory neurons (Figure 2A). We first evaluated the number of EAD-active fractions in the effluent of the GC for each sample type (Figure 2B). With the exception of two background plants (Argemone, Gutierrezia), all plant bouquets contained EAD-active fractions. The nectar sources Agave flower and Datura flower emitted the highest number of compounds (on average, 20 and 17 active fractions, respectively), followed by the bouquets of host plants of sympatric hawkmoths (Chilopsis, Helianthus, Vitis) and a background tree (Prosopis) (11–14 active fractions). The two larval host plants of M. sexta, on the other hand, contained only 4–7 active compounds.

Antennal responses of M. sexta females to nocturnal headspaces of plants.

(A) Examples of gas chromatography-coupled electro-antennographic detection (GC-EAD) recordings after stimulation with four plant headspaces representing nectar sources (Agave flower), host plants (Proboscidea), host plants of sympatric hawkmoths (Vitis), and background plants (Prosopis). Upper traces, gas chromatograph-coupled flame ionization detection (GC-FID); lower traces, electro-antennographic detection (EAD) of female M. sexta. Letters indicate EAD-active GC-peaks (labeled in C) that evoked a response in at least three animals. Arrows, internal standard: 5 ng 1-bromodecane; in Agave flower, the internal standard co-eluted with GC-peak ‘z,’ and GC-peaks ‘v’ and ‘x’ are cropped. (B) Number of EAD-active GC-peaks per plant species. We stimulated the antennae (4–7 moths/headspace) with the same representative sample per headspace type. Filled squares, average values; whiskers, range; open squares, no active GC-peaks detected in three moths. Each moth was tested only once. (C) Antennal responses towards GC-peaks (rows) present in headspace (columns). Each cell in the heat map represents the median EAD amplitude of on average five moths (range: 4–7) per headspace. Rows are sorted by EAD amplitude (Figure 2—source data 1); magnitude of response is coded by shades of gray (see inset at top); empty cells, no response/GC fraction not present. Color code of compounds according to chemical class (see inset at bottom). Numbers next to ethyl sorbate and propyl sorbate label different enantiomers present in Agave flower and depict their order by retention time; DMNT, (E)–4,8-dimethyl-1,4,7-nonatriene. Numbers to the right of the heat map depict how often a given compound was present; rows without numbers indicate compounds found only in one headspace. (D) Effectiveness of the strongest antennal stimulants. x-axis, concentration of compounds derived from their peak area (logarithmic scale); y-axis, median EAD amplitudes ≥ 1 mV; gray vertical bar, range of peak areas of the internal standard 1-bromodecane (5 ng). For compounds present in more than one plant species, the lowest concentration eliciting a median EAD amplitude ≥ 1 mV was chosen; letters indicate compounds as in (C); *α-copaene; #(E)-β-farnesene. Peak area of ethyl sorbate2 (‘v’) shows lower limit of concentration as the GC seemed overloaded with this odor.

Figure 2—source data 1

Related to Figure 2C.

Gas chromatography-coupled electro-antennographic detection (GC-EAD) results from 80 antennae.

https://cdn.elifesciences.org/articles/77429/elife-77429-fig2-data1-v1.xlsx

Across all headspaces, we found 77 EAD-active compounds (Figure 2C) and could tentatively identify 69 of them. These compounds mainly belonged to three chemical classes: terpenes, aliphatic esters, and aromatics. The most potent antennal stimulants (n = 16) elicited median EAD amplitudes > 1.0 mV. Eight of these strongly activating odors were aliphatic esters present exclusively in the bouquet of Agave flowers; three more odors were present in the headspace of nectar sources (Agave flower, Datura flower, and/or Mirabilis) but not in other sample types. The remaining strongly activating odors each occurred in at least five plant species from all sample types and included the most common volatiles in our collections: (Z)-3-hexenyl acetate (11 plants) and β-ocimene (10 plants). When we plotted the concentration of the most activating GC fractions versus the EAD amplitude they evoked, we found that α-copaene, (Z)-3-hexenyl acetate, and β-ocimene were the most active odors at concentrations below 5 ng in 12 hr of odor collection (Figure 2D).

The antenna of M. sexta was in addition reacting with a weaker response towards many more plant-released volatiles in a species-specific manner. Furthermore, two-thirds of all EAD-active GC fractions (51 out of 77) were restricted to one of the plant species (Figure 2C). Thus, beyond the impression received from the chemical analysis (Figure 1C), the moths’ antennae seemed to be well suited to distinguish between plant bouquets even when they had low volatile concentrations and inconspicuous chemical profiles, like the two larval host plants of M. sexta and most background plants. The mating status of the moth had no impact on its detection capabilities at the level of the antenna (two-way ANOSIM, mating status: R = −0.06, p=0.756, plant species: R = 0.97, p<0.0001; Bray–Curtis similarity index).

How is plant headspace represented in the moth’s antennal lobe?

In in vivo calcium imaging experiments, we successively stimulated the antennae of female M. sexta with puffs of the plant bouquets collected in Arizona and recorded the odor-evoked neural activity among the olfactory glomeruli of their antennal lobe. Olfactory glomeruli are functional subunits occurring in species-specific numbers. Female M. sexta possess 70 glomeruli arranged in a monolayer around a central neuropil (Grosse-Wilde et al., 2011). Activity patterns of glomeruli in the dorsal-frontal part of the antennal lobe can be monitored using in vivo calcium imaging (Hansson et al., 2003; Sachse et al., 1999). To enable comparison of headspace-evoked activation patterns among different animals, we identified 23 glomeruli in each moth using diagnostic, monomolecular odorants (Figure 3A and B, Bisch-Knaden et al., 2018). We found that plant headspace activated these 23 identified glomeruli (Figure 3C). However, two glomeruli (22 and 23) responded exceptionally weak. They are tuned to acids and amines (Bisch-Knaden et al., 2018), chemical classes that were functionally absent in the tested plant bouquets (Figure 2C). Next, we tested which responses were true headspace-evoked responses, that is, which responses were different from the response towards stimulations with the eluent dichloromethane (Figure 3D), and normalized the fluorescent signals of headspace-evoked responses for each glomerulus and animal (Figure 3E).

Headspace-evoked activity patterns in the antennal lobe of female M. sexta.

(A) Schematic of 23 olfactory glomeruli at the dorsal surface of the right antennal lobe. Entrance of the antennal nerve is in the upper-left corner. Numbers, glomeruli identification as in Bisch-Knaden et al., 2018. (B) Examples of in vivo calcium imaging recordings after stimulation with plant headspaces representing nectar sources (Agave flower, Datura flower), host plants (Datura foliage, Proboscidea), host plants of sympatric hawkmoths (Vitis), background plants (Prosopis), and the eluent dichloromethane (first and second stimulations at the beginning and end of the experiment). False-color-coded imaging results of the right antennal lobe in a virgin (left column) and a mated female (right column) normalized to their highest response (see color bar). Top row, schematic of individual antennal lobes, colors as in (A). (C) Maximum increase of fluorescence in 23 identified glomeruli. Graph depicts for each glomerulus (color code as in A) the average maximum responses (bars) and 1 standard deviation (whiskers) of 10 virgin and 10 mated females after stimulation with plant headspaces. In 69% of 460 cases (20 maximum values in 23 glomeruli), Datura flower was the headspace eliciting the maximum response, and in 17% it was Agave flower. (D) Number of activated glomeruli in the antennal lobe depending on female mating status. A glomerulus was scored as activated if its headspace-evoked response was different from the averaged response to the two stimulations with the eluent dichloromethane (p<0.01, Friedman test with Dunn’s multiple-comparisons test). For the identity of glomeruli activated by each plant headspace, see Table 2. (E) Activity levels evoked by plant headspace in individual glomeruli in the antennal lobe. Colored dots represent median normalized responses of activated glomeruli in 10 virgin (top) and 10 mated (bottom) females; color code of glomeruli as in (A). Only values of activated glomeruli are shown (small circles, p<0.01, large circles, p<0.001, Friedman test with Dunn’s multiple comparisons test, Figure 3—source data 1).

Figure 3—source code 1

Custom-written software for processing calcium imaging data in IDL (L3Harris Geospatial).

https://cdn.elifesciences.org/articles/77429/elife-77429-fig3-code1-v1.zip
Figure 3—source data 1

Calcium imaging results from 10 virgin and 10 mated females (Figure 3E).

https://cdn.elifesciences.org/articles/77429/elife-77429-fig3-data1-v1.xlsx

Consistent with the results from GC-EAD experiments, Datura and Agave bouquets were again unique regarding not only the number of activated glomeruli (Figure 3D) but also the strength of response (Figure 3E). Datura flower scent evoked the maximal response recorded in all but two glomeruli in virgin females (glomeruli 12 and 21), and in all but one glomerulus in mated females (glomerulus 12). Agave flower scent was the best activator for these remaining glomeruli. Apart from the weak activation levels of glomeruli 22 and 23, this exceptional representation of M. sexta’s two primary nectar sources was independent of the females’ mating status (Table 2), at least among the 23 glomeruli imaged in this study.

Table 2
Headspace-activated glomeruli independent and dependent of mating status.
GlomerulusResponse independent of mating statusResponse only before matingResponse only after mating
1*Agave, DaturaMirabilis, Helianthus, Gutierrezia
2Agave, Datura, Mirabilis, HelianthusMimosa, Vitis, Gutierrezia
3Agave, Datura, Mirabilis, HelianthusMimosa, Chilopsis, Vitis, Gutierrezia
4*Agave, Datura, Mimosa, Mirabilis, Datura foliage, Chilopsis, Helianthus, VitisGutierrezia
5Agave, Datura, Helianthus, VitisMimosa, Chilopsis, Gutierrezia, ProsopisMirabilis
6*Agave, Datura, Mimosa, Mirabilis, Chilopsis, Helianthus, VitisGutierrezia, Prosopis
7Agave, Datura, HelianthusMimosa, Vitis, Gutierrezia
8Agave, Datura, Mimosa, Chilopsis, Helianthus, VitisGutierrezia
9Agave, Datura, Mimosa, Helianthus, VitisChilopsis, Gutierrezia
10Agave, Datura
11Agave, Datura
12*Agave, Datura, MimosaAmaranthusMirabilis, Proboscidea, Chilopsis, Helianthus, Vitis
13*Agave, Datura, MimosaProsopisMirabilis, Chilopsis, Helianthus, Vitis
14Agave, Datura, Mimosa, AmaranthusProsopis, SennaChilopsis, Helianthus
15Agave, Datura, Mimosa, Chilopsis, Helianthus, Vitis, Prosopis, SennaProboscidea, AmaranthusMirabilis
16Agave, Datura, Mimosa, Chilopsis, Helianthus, ProsopisAmaranthus, BaccharisVitis
17*Agave, Datura, Mimosa, Helianthus, Vitis, SennaProsopisMirabilis, Chilopsis
18*Agave, Datura, MimosaChilopsis, Helianthus
19*Agave, Datura, Mimosa, HelianthusProsopisChilopsis, Vitis
20*Agave, Datura
21*Agave, Datura
22Agave
23Datura
  1. Font format depicts type of plant headspace: nectar source of M. sexta, host plant of M. sexta, host plant of sympatric hawk moths, background plant.

  2. *

    Glomerulus whose activation level is positively correlated with odor-guided behavior of virgin females in wind tunnel experiments (Bisch-Knaden et al., 2018).

Volatiles emitted by M. sexta’s larval host plants each activated only a single glomerulus out of the 23 imaged in the antennal lobe of females (Figure 3D). Glomerulus 4 responded to the bouquet of Datura foliage irrespective of the female’s mating status; Proboscidea headspace, however, activated a different single glomerulus in virgin (glomerulus 15) than in mated females (glomerulus 12) (Figure 3E, Table 2). Furthermore, we observed a notable effect of the mating status on the representation of plants that are oviposition sites of sympatric hawkmoths (Chilopsis, Helianthus, Vitis). These plants evoked a major response in the antennal lobe of virgin M. sexta females (8–13 activated glomeruli) and an even stronger response in mated females (11–16 activated glomeruli) (Figure 3D and E). In contrast, females became almost anosmic towards the headspace of background plants following mating as these plants activated on average 3.1 glomeruli (range: 0–9) in virgin, but only 0.6 glomeruli (range: 0–2) in mated females (Figure 3D). The few glomeruli still responding towards background bouquets were different from the two host plant-activated glomeruli (Figure 3E, Table 2).

A multivariate analysis confirmed that mating status as well as plant species had a significant effect on overall activation patterns across glomeruli in the antennal lobe (two-way ANOSIM, mating status: R = 0.71, p=0.0001, plant species: R = 0.83, p=0.0001; Bray–Curtis similarity index).

Discussion

For a female moth, two plant-based resources are of overriding importance: flowers providing nectar for sustenance of the animal itself and plants providing suitable oviposition sites and thereby food for the offspring. Here, we studied how the olfactory system of the female hawkmoth, M. sexta, has evolved to allow unambiguous identification of these resources based on their emissions of volatile molecules.

As could be expected, plants that predominately (Datura, Mirabilis) or at least partly (Agave) rely on nocturnal pollination by hawkmoths (Alarcón et al., 2008; Emiliano Trejo-Salazar et al., 2015) sent a clear and distinctive chemical signal in the night. Our results regarding the number and identity of compounds emitted by Agave and Datura flowers and the dissimilarity between both floral bouquets confirm earlier studies (Raguso, 2004; Raguso et al., 2003a; Riffell et al., 2008). In addition, the sunflower Helianthus emitted a strong and distinct scent, illustrating that its around-the-clock open flowers depend not only on diurnal but also on nocturnal pollinators. Although M. sexta is not described as a pollinator of Helianthus (Torretta et al., 2009), unspecified pollen from the sunflower family was found on the proboscis of M. sexta and other hawkmoths, indicating that these moths occasionally feed also on sunflowers (Alarcón et al., 2008). Low nighttime emissions observed in the remaining samples might reflect the plants’ independence of nocturnal pollination (in the case of flowering plants) or an avoidance strategy against herbivory (for collections from nonflowering branches).

When we tested the antenna of female M. sexta with plant headspaces using GC-EAD, we found that the moths in most of the cases detect at least some compounds, even in wind-pollinated background vegetation like grass or careless weed, plants that, based on the chemical analysis, had a weak smell consisting of a small number of components. Female moths, both virgin and mated, therefore seem to be equipped with the sensory capability to distinguish not only strong and complex scents emitted by nectar sources but also bouquets of host plants and surrounding background vegetation. However, the highest number of active compounds was present in the floral headspace of Agave and Datura. In particular, the bouquet of Agave contained 9 of the 16 strongest antennal stimulants, 8 of them being aliphatic esters. These esters are signature compounds of Agave flowers (Raguso, 2004) as they are rarely found in other floral headspace investigated in almost 1000 plant species from 90 families (Knudsen et al., 2006). In particular, this chemical class is lacking in typical hawkmoth-pollinated flowers (Raguso et al., 2003a; Raguso et al., 2003b). Two enantiomers of the Agave-characteristic ester ethyl sorbate elicited the strongest antennal response of all active GC-peaks (median EAG amplitudes: 2.6 mV and 2.9 mV, stimulus concentrations: ~0.5 µg and >>0.5 µg, Figure 2D). These responses are higher than the response of a male antenna when stimulated with bombykal, the main component of M. sexta’s sex pheromone (1.9 mV, stimulus concentration: ~100 µg; Fandino et al., 2019). Even if we consider that bombykal has a lower vapor pressure than ethyl sorbate and that less odor might reach the antenna in EAG than in GC-EAD experiments, this comparison indicates that the female antenna is at least as sensitive to promising floral, that is, nectar-indicating volatiles, as the male antenna by the female sex-pheromone.

EAD activity might correlate with behavior (Liu et al., 2020; Zhu et al., 1993), but a strong antennal response towards an odor does not always imply a strong behavioral response to this odor molecule (Honda et al., 1998; Suckling et al., 1996). In M. sexta, a comparison of physiological and behavioral data is possible as almost half of the active and identified GC-peaks in this study were previously tested in a wind tunnel assay (Bisch-Knaden et al., 2018). EAD responses evoked by these 31 shared odors belonging to seven chemical classes are indeed positively correlated with the duration a female moth shows feeding behavior when encountering the same odors (EAD amplitude versus duration of proboscis contacts with a scented filter paper, Pearson correlation coefficient r = 0.41, p=0.023). In contrast, no correlation was found between EAD activity and the duration of abdomen curling behavior, that is, behavior related to oviposition (r = −0.03, p=0.9). Hence, no conclusions from GC-EAD results can be drawn regarding an odor’s relevance in connection with an oviposition site, whereas odors that evoked a strong response at the antenna of female M. sexta often are attractive in the context of feeding.

In previous GC-coupled single-sensillum recordings (GC-SSR), 60% of randomly chosen olfactory sensilla on the antenna of female M. sexta reacted to the aliphatic ester (Z)-3-hexenyl acetate when stimulated with the scent of herbivore-damaged Datura foliage (Spaethe et al., 2013b). If the olfactory sensory neurons housed in these sensilla would not only detect (Z)-3-hexenyl acetate but aliphatic esters in general, this could explain the prominent response towards typical Agave esters in our GC-EAD experiments. In addition, the antenna might harbor narrowly tuned olfactory sensory neurons strongly responding only to Agave esters. Hawkmoth-pollinated flowers like Datura, Nicotiana, and Petunia emit oxygenated aromatics that are especially attractive to foraging hawkmoths and elicit a strong response from the antenna of M. sexta, and in its antennal lobe, respectively (Bisch-Knaden et al., 2018; Hoballah et al., 2005; Kessler et al., 2008; Riffell et al., 2013). This study shows that M. sexta females in addition exhibit a robust physiological response towards the bouquet collected from Agave flowers, reflecting the significant role this copious nectar source — releasing a very different smell than typical hawkmoth flowers — plays in M. sexta’s foraging behavior. Other hawkmoth species feeding on nectar from Agave flowers might share similar olfactory detection and processing abilities (Alarcón et al., 2008; Emiliano Trejo-Salazar et al., 2015).

Three volatiles, α-copaene, (Z)-3-hexenyl acetate, and β-ocimene, stood out as particularly strong activators of the female M. sexta antenna, although they were present in very low concentrations. The high sensitivity towards these odors might indicate that they act as long-distance cues guiding the moth to places with vegetation (Webster and Cardé, 2017). Furthermore, these odors have meanings that are more specific: α-copaene is involved in the oviposition decision process of M. sexta (Zhang et al., 2022) and in addition might indicate rewarding nectar sources as it was functionally present, that is, EAD-active, only in the headspace of Datura flower and Mirabilis. (Z)-3-hexenyl acetate and β-ocimene, on the other hand, are typical herbivore-induced volatiles and are released by herbivore-damaged Datura leaves (Allmann et al., 2013; Hare and Sun, 2011; Zhang et al., 2022). Interestingly, some projection neurons in the female antennal lobe targeting an identified glomerulus were reported to specifically respond to low concentrations of (Z)-3-hexenyl acetate (Reisenman et al., 2005). This odor might thus inform a M. sexta female searching for oviposition sites about the presence of potential larval competitors and predators already at a distance.

An earlier GC-EAD study with female M. sexta reported that Datura and Proboscidea foliage each emit 10 identified EAD-active compounds and share 8 of them (Fraser et al., 2003). Our work, in contrast, shows that both host plants emit only nine and four active compounds, respectively, and have no active compounds in common. In the case of Proboscidea headspace, none of its EAD-active compounds found in our experiments was identified in the former study, and vice versa. Many methodological factors could have contributed to this discrepancy in the results. In detail, Fraser et al. collected headspace from single, potted, undamaged plants with buds, flowers, and seeds removed; and a cultivar of Proboscidea was used, not the wild type. In this study, we collected headspace of local, mostly herbivore-damaged plants growing in a natural plant community and did not remove any parts of the plant. In addition, our odor collection lasted 12 hr during the natural dark phase versus 24 hr of artificially induced scotophase in Fraser et al., 2003. Conditions like growth in mixed plant populations or in monocultures (Kigathi et al., 2019), light deprivation (He et al., 2021), herbivore attack, and other stress factors (Holopainen and Gershenzon, 2010) influence both composition and quantity of plant-emitted volatiles. Therefore, the observed differences between the studies could be expected and emphasize the significance of odor collections in the field.

Bath application of a fluorescent calcium sensor allows monitoring of odor-induced neural activity in the brain. Each neuron type in the treated brain region might take up the marker molecules. However, as each glomerulus in the antennal lobe receives input from 4000 to 5000 olfactory sensory neurons (Oland and Tolbert, 1988), and is targeted by only 4–5 projection, that is, output neurons (Homberg et al., 1988), odor-evoked activation patterns in calcium imaging experiments can be assumed to reflect mainly the activity of input neurons. Additionally, about 360 local interneurons per antennal lobe (Homberg et al., 1988) with inhibitory and/or excitatory functions (Reisenman et al., 2011) might synapse back onto the sensory neurons, thus modulating their activity and accordingly the observed calcium signal. Although most of these interneurons arborize in many, if not all, glomeruli, some interneurons have a more restricted innervation pattern and connect only a few glomeruli (Christensen et al., 1993). This type of interneuron seems predisposed to play a role in the coding of complex odor blends released by plants. Interestingly, patchy interneurons are present mainly in female M. sexta (Matsumoto and Hildebrand, 1981). In the vinegar fly Drosophila melanogaster, patchy interneurons are responsible for nonlinear processing of binary odor mixtures (Mohamed et al., 2019). For some glomeruli in D. melanogaster, this modulation occurred already at the presynaptic level, that is, at the level we monitored in our calcium imaging experiments. To estimate if nonlinear interactions might occur in the antennal lobe of M. sexta, we compared headspace-evoked activation patterns with activation patterns evoked by EAD-active, single compounds that were present in the respective headspace. From the bouquet of Datura foliage, for example, (Z)-3-hexenyl acetate elicited the strongest antennal response (Figure 2C) and activates mainly four glomeruli (glomeruli 6, 13, 16, and 12) when tested on its own (Bisch-Knaden et al., 2018). After stimulation with the complex headspace, however, none of these glomeruli was responding (Figure 3E, Table 2). The second best antennal activator in the headspace of Datura foliage, geraniol, activates mainly three glomeruli (glomeruli 6, 4, and 5, Bisch-Knaden et al., 2018). Of these, glomerulus 4 was the only one responding towards stimulation with the headspace (Figure 3E, Table 2). We thus conclude that there are indications of local inhibition as we otherwise would have expected to observe more activated glomeruli after stimulation with the complex headspace. A similar inhibition of glomeruli in mixtures of odors was reported in a calcium imaging study in honey bees, where the inhibitory effect was stronger in ternary than in binary mixtures (Joerges et al., 1997). As the plant bouquets tested in our study contained up to 20 EAD-active components, and as local interneurons in M. sexta, like in most insects, are mainly inhibitory (Christensen et al., 1993), the observed inhibitory mixture interactions after stimulation with complex blends seem plausible.

However, we also revealed coding characteristics that were similar at the periphery and in the brain, especially after stimulation with feeding-related odor bouquets. Representation of the essential nectar sources Datura and Agave flowers in the antennal lobe was outstanding as in the vast majority of all glomerulus-headspace combinations these two floral scents elicited the highest response. In former studies using a different approach, only a small fraction of the compounds present in the two floral bouquets (Agave: 10%; Datura: 15%) activated neurons in the antennal lobe of male M. sexta (Riffell et al., 2009a; Riffell et al., 2009b). However, the recording technique used by Riffell and colleagues targets about 10 glomeruli and part of the adjacent neuropil in the lateral part of the male antennal lobe, while we recorded from 23 identified glomeruli in the dorsal part of the female lobe. In addition, we used a puff of the floral headspace as stimulus, that is, the antennal lobe was activated by the full floral blend, whereas in previous studies a GC-coupled stimulus was applied, that is, the antennal lobe was activated by temporally separated single compounds present in the floral blend. The differing results might thus be due to these different approaches and potential sex-specific coding differences. However, compounds that were identified to activate neurons in the male antennal lobe, like ethyl sorbate (Agave) and benzyl alcohol (Datura), were also EAD-active in this study (Figure 2C) and evoked responses in some glomeruli of the female antennal lobe in a previous imaging study (Bisch-Knaden et al., 2018). Interestingly, lab experiments in a wind tunnel suggest that even such a strong scent as the one from Datura flowers can become less attractive to male M. sexta when presented in the olfactory background of a selected nonhost plant (Riffell et al., 2014).

Olfactory coding can change depending on the mating status of an insect (Gadenne et al., 2016). However, the neural response of female M. sexta towards volatiles from its main nectar sources was not altered following mating as it was reported for the noctuid moth Spodoptera littoralis (Saveer et al., 2012). Different life history traits of noctuid and sphingid moths might explain this different result: noctuid moths are generalists and lay their eggs in clusters on a wide range of acceptable host plants. Sphingid moths like M. sexta, on the other hand, are usually specialized on a few host plant families and females lay only a few single eggs on a given plant. Thus, hawkmoths need to refill their energy reservoir between oviposition bouts at host plants that are rare in the habitat and therefore require long flights between them (Alarcón et al., 2008; Raguso et al., 2003a). Moreover, female hawkmoths benefit from nectar feeding following mating as they live longer and produce more mature eggs compared to starved females (Sasaki and Riddiford, 1984; von Arx et al., 2013). The energy demand of hawkmoths is in addition especially high as both feeding and oviposition usually occur while the moth is hovering in front of the plant (Stöckl and Kelber, 2019). Taken together, the prominent and mating status-independent representation of floral bouquets at the antenna and in the antennal lobe of female M. sexta is in accordance with the moths’ ecology.

While the coding of flower volatiles in nectar-feeding moths is probably independent of sex, odors indicating oviposition sites should be of special importance for female moths after mating. Two enlarged, female-specific glomeruli that are located at the entrance of the antennal nerve into the female antennal lobe — at the same position as the sex pheromone-processing macroglomerular complex in males (Matsumoto and Hildebrand, 1981; Rössler et al., 1998) — seem predisposed to be involved in oviposition choice. This hypothesis is supported by the fact that output neurons targeting both glomeruli respond to headspace of tomato leaves, another host plant for M. sexta (King et al., 2000; Reisenman et al., 2009). On the other hand, the two host plant bouquets tested in our imaging experiments did not activate these glomeruli (glomeruli 1 and 2, Table 2), confirming results of a study using vegetative headspace from the hosts Datura, Nicotiana, and tomato. These scents failed to evoke a response in sensilla targeting mainly the two female-specific glomeruli (Shields and Hildebrand, 2001). Therefore, the question if these glomeruli might be involved in identifying an oviposition site is still open.

In contrast to the wide and strong activation of antennal lobe glomeruli by flower odors, M. sexta’s host plant bouquets each activated only a single glomerulus of the 23 glomeruli under investigation. While the responding glomerulus towards Datura foliage was independent of the female’s mating status, Proboscidea activated a different glomerulus in virgin than in mated females. This result is in line with the fact that the ecological meaning of Datura foliage does not change after mating as its smell indicates both a suitable host plant and a profitable nectar source (Kárpáti et al., 2013). Proboscidea, on the other hand, does not provide nectar for hawkmoths and is therefore interesting for the female moth only after mating. Many EAD-active compounds were tested in a previous calcium imaging study using monomolecular odorants as stimuli (Bisch-Knaden et al., 2018), allowing a comparison between these data and our imaging results obtained with natural mixtures. Some compounds present in the headspace of Proboscidea and Datura foliage, for example, when tested alone activated most strongly glomerulus 6, a glomerulus that was not activated after stimulation with the complete headspaces, again indicating nonlinear processing and robust presynaptic inhibitory interactions between glomeruli (Joerges et al., 1997; Mohamed et al., 2019). The two host plant-activated glomeruli in mated females were as well responding to nectar sources and hosts of sympatric hawkmoths. However, host plants exclusively activated one of these glomeruli, whereas the other sources activated additional glomeruli. Even if nonhost plants as well as host plants would activate more glomeruli in areas of the antennal lobe that were inaccessible in our imaging study, the resulting neural representation of nonhost plants in the antennal lobe of mated females would remain different from the pattern evoked by host plants.

Interestingly, virgin and mated females differed markedly in their response to the odor of background plants: of the 23 identified glomeruli, these plants activated only a small number in virgin females (range: 0–9) and even less glomeruli (range: 0–2) in mated females. The few background-activated glomeruli did not include the two host plant-activated glomeruli. The moths’ reduced responses to background plants but not to host plants after mating, together with the finding that the host plant Proboscidea activates a different glomerulus in virgin and mated females, indicate that the olfactory system of M. sexta females becomes tuned towards host plants following mating. Mechanisms mediating these post-mating changes in moth olfactory processing seem to be independent of neurotransmitters like octopamine and serotonin (Barrozo et al., 2010) but might include neuropeptides as in the vinegar fly, D. melanogaster (Hussain et al., 2016), and regulation of chemosensory-related genes like in Drosophila suzukii (Crava et al., 2019). Our data suggest that mated females could potentially be able to identify suitable oviposition sites by the relative activity of one or two glomeruli compared with the activity of other glomeruli. A similar sparse coding strategy was recently described for the discrimination of differentially attractive body odors by mosquitoes (Zhao et al., 2020). In this case, the relative activity of a single, human-odor-activated glomerulus versus a broadly tuned glomerulus has been proposed to enable the mosquito to identify its preferred human host.

M. sexta females intersperse feeding and oviposition bouts when visiting a flowering Datura (Raguso et al., 2003a), and lay more eggs on flowering than on nonflowering plants (Reisenman et al., 2010). However, Datura foliage alone attracts egg-laying females in the field (personal observation) and the lab (Nataraj et al., 2021; Spaethe et al., 2013b). We therefore tested leaves of Datura separately and compared their headspace with that of Proboscidea, the only other host plant in our study area. As the single activated glomerulus was different after stimulation with the two host plant bouquets, which also did not share any EAD-active compounds, M. sexta females should be able to distinguish the two plants based on olfaction alone. Field observations and experiments show that females lay more eggs on Proboscidea than on solanaceous hosts, although the plants grow next to each other and have a similar leaf surface (Diamond and Kingsolver, 2010; Mechaber and Hildebrand, 2000). These findings indicate that M. sexta can indeed discriminate between host plants belonging to different plant families, although visual and tactile cues might play a role in combination with olfactory cues. The observed low overall activity across the antennal lobe evoked by host plant odors corresponds to the preference of M. sexta for plants with a faint smell when searching for oviposition sites. Inbred horse nettle (Solanum carolinense) exhibits much lower nocturnal volatile emissions than outbred horse nettle, a solanaceous host plant of M. sexta in the southeastern US. Correspondingly, female moths spend more time hovering near inbred plants and lay more eggs there than on outbred plants. This preference is governed by olfactory cues alone as it persists in the absence of visual and contact cues (Kariyat et al., 2013). Furthermore, when given the choice between headspaces of two solanaceous host plant species with different total volatile concentration, M. sexta clearly prefers the weaker smelling plant. Diluting the headspace of the more intensely smelling plant leads to a reduction in this preference (Spaethe et al., 2013a). These findings show again that female moths consistently favor host plants with low volatile emissions, probably because high emission of specific volatiles are signs of active plant defense mechanisms, indicating the presence of larval competitors (De Moraes et al., 2001), and leading to impaired larval growth (Delphia et al., 2009). High levels of these herbivore-induced volatiles also attract predators and parasitoids (Kessler and Baldwin, 2001; Turlings et al., 1995), and egg-laying moths therefore avoid these sites (De Moraes et al., 2001; Li et al., 2018). Conversely, when M. sexta has to choose between flowering tobacco plants from populations that differ in their flower volatile concentration, the moths clearly prefer to forage at flowers with a stronger smell (Haverkamp et al., 2018). Hence, M. sexta pursues different strategies when searching for oviposition or feeding sites as the moths favor weakly or strongly scented sources, respectively.

In contrast to M. sexta’s host plants, the bouquets of host plants of sympatric hawkmoths activated many glomeruli in virgins, and even more glomeruli in mated females. Two glomeruli contributed mostly to this effect as they were responding to all three nonhost bouquets only after mating (glomeruli 12 and 13; Table 2). Odor-induced activation levels of these two glomeruli were positively correlated to odor-induced behavior in wind tunnel experiments using monomolecular odorants (Bisch-Knaden et al., 2018). However, only virgin females were included in this study, so conclusions regarding the behavior of mated females cannot be drawn.

The strong activation of antennal lobe glomeruli by host plants of other hawkmoths living in the same habitat was in contrast to weak but specific activation of single glomeruli (among those imaged) by host plants of M. sexta. The conspicuous activation patterns evoked by host plants of sympatric hawkmoths might serve as a stop signal for M. sexta during their search for a suitable oviposition site and therefore might help gravid females to avoid inappropriate hosts at a distance. It would be interesting to compare headspace-evoked activation patterns in the antennal lobe of co-occurring hawkmoths upon stimulation with odors of their own and of other species’ host plants to test if this might be a general coding policy. Examples of olfaction-based avoidance of nonhost plants were also reported for example in bark beetles (Huber et al., 2000). Antennae of these insects respond strongly to many volatiles released by nonhost trees. Like in the case of M. sexta, some compounds are present in the bouquet of both host and nonhost plants, corroborating the hypothesis that odor-guided choice of host plants relies on blends of ubiquitous compounds in a specific ratio (Bruce and Pickett, 2011) and concentration (Spaethe et al., 2013a) rather than on the detection of host-exclusive odors.

By using ecologically relevant odors collected in the actual habitat of our model animal, M. sexta, we revealed olfactory coding strategies both for odors emanating from crucial resources but also for those emitted by substrates that should be avoided. We also show how the female mating status affects olfactory processing but, interestingly, in a way well adapted to the specific life history traits of the species under investigation. In a broader perspective, our study contributes to understanding innate neural representation of natural odor mixtures in the brain and coding strategies enabling animals to distinguish crucial resources from background noise.

Materials and methods

Headspace collection in the field

Request a detailed protocol

We collected the headspace of plants in a habitat of M. sexta at the Santa Rita Experimental Range, 40 km south of Tucson, Arizona, at the foot of the Santa Rita Mountains (31°78′ N, 110°82′ W). All plant species sampled (Table 1) are native to the habitat and belong to the regular desert grassland vegetation at Santa Rita Experimental Range (Medina, 2003). We sampled from flowering plants or flowering branches if the respective plant was blooming during the experimental nights. Otherwise, nonflowering branches were sampled (Table 1). Agave is a succulent plant with a basal rosette of sharp-edged leaves, each with a length of c. 1 m and long spines at the tip. These leaves did not fit in our collection bags (see below). We therefore only collected headspace from Agave flowers, which appear in umbels at the end of a long bloom stalk without leaves (about 5–6 m above the basal rosette of leaves). In the case of Datura, the flower is a valuable nectar source and the leaves are an oviposition substrate for M. sexta. We therefore tested flowers and leaves of Datura separately.

At sunset, we carefully enclosed plants in polyethylene terephthalate bags (Toppits, Germany). Charcoal-filtered, environmental air was pumped into the bag through a silicone tube connected to a custom-made portable pump. Air was pumped out of the bag through a second silicone tube passing a volatile collection trap (Porapak-Q 25 mg, https://www.volatilecollectiontrap.com). Shortly after sunrise, we unpacked the plants, removed the volatile collection traps, and stored them at −20°C. We collected the headspaces of plants on nine consecutive nights (August 19–27, 2018). In the first and last nights, we made a control collection with an empty bag placed on the ground close to the collection sites of plant headspaces. One volatile collection trap not used but treated in the same way as headspace-collecting traps served as a handling control. In Jena, Germany, all volatile collection traps were eluted with 4 × 100 µl dichloromethane containing 5 ng/µl bromodecane as an internal standard.

Chemical analysis

Request a detailed protocol

Headspace samples were analyzed by GC-MS (7890B GC System, 5977A MSD, Agilent Technologies, https://www.agilent.com) equipped with a polar column (HP-INNOWAX, 30 m long, 0.25 mm inner diameter, 25 µm film thickness; Agilent) with helium as carrier gas. The inlet temperature was set to 240°C. The temperature of the GC oven was held at 40°C for 3 min, and then increased by 10°C per min to 260°C. This final temperature was held for 15 min. The MS transferline was held at 260°C, the MS source at 230°C, and the MS quad at 150°C. Mass spectra were taken in electron ionization mode (70 eV) in the range from m/z 29 to 350. GC-MS data were processed with the MDS-ChemStation Enhanced Data Analysis software (Agilent).

Breeding of M. sexta

Request a detailed protocol

M. sexta larvae were reared in the laboratory on artificial diet (Grosse-Wilde et al., 2011). Female pupae were kept in a climate chamber (25°C, 70% relative humidity) with a reversed light cycle (8 hr dark/16 hr light), and moths were tested during their scotophase on days 2–4 after hatching (GC-EAD), or at day 3 after hatching (calcium imaging). Moths were unfed and had no experience with plant-derived volatiles. To obtain mated females, we placed them in a cage with an equal number of males 1 day before an experiment was planned. We checked the cage 3–4 hr later and removed all animals that were not mating.

GC-EAD recordings

Request a detailed protocol

We used GC with flame-ionization detection (GC-FID) coupled with electro-antennographic detection (EAD) to identify compounds in headspace collections that can be sensed by M. sexta. One antenna of a female moth (virgin and mated in equal numbers), age 2–4 days, was cut and connected to two glass electrodes filled with physiological saline solution (Christensen and Hildebrand, 1987). The reference electrode was inserted into the basal segment of the antenna, and the recording electrode was brought in contact with the tip of the antennae. The EAD signal (transferred via Ag-AgCl wires) was pre-amplified (10×) with a probe connected to a high-impedance DC amplifier (EAG-probe, Syntech, https://www.syntech.nl) and digitally converted (IDAC-4 USB, Syntech), visualized, and recorded on a PC using the software Autospike (Syntech). For each run, 2 µl of a headspace sample (a lower amount of 1 µl for Agave flower and Datura flower) was injected into a GC-FID (6890N, Agilent) equipped with a polar column (HP-INNOWAX, 30 m long, 0.32 mm inner diameter, 0.25 µm film thickness, Agilent) with helium as carrier gas. The inlet temperature was set to 250°C. The temperature of the GC oven was held at 40°C for 2 min, and then increased by 10°C per min to 260°C. This final temperature was held for 10 min. The GC was equipped with an effluent splitter (Gerstel) at the end of the analytical column, with a GC:antenna split ratio of 1:10 and helium as carrier gas. One arm was connected with the FID of the GC, and the other arm entered a heated (270°C) GC-EAD interface (Syntech) that was connected to a bent glass tube (diameter: 12 mm). The antenna-directed GC effluent was mixed with a humidified, charcoal-filtered air stream (1 l/min) to cool the effluent down and guide it to the antenna. Signals from the moth’s antenna and the FID were recorded simultaneously. Sample size was 4–7 antenna per sample type; if a given sample type did not elicit a single response in three different animals, it was not tested any further (Argemone, Gutierrezia, empty bag). A GC-peak was scored as EAD-active if it induced a response at the same retention time in at least three antennae and if this GC fraction was present in at least one other headspace of the same type.

Identification of compounds

Request a detailed protocol

On both GC instruments (GC-MS and GC-FID), we ran a series of 20 n-alkanes and matched retention times of EAD-active peaks and peaks obtained with the GC-MS using their Kovats retention indices. EAD-active peaks were tentatively identified by comparison of their mass spectra and Kovats retention indices with those from a reference library (National Institute of Standards and Technologies), and a database built in our laboratory with synthetic standards using the same GC-MS instrument. Compounds yielding a match of mass spectra above 90% were rated as tentatively identified. The fragmentation pattern of some EAD-active peaks could not be clearly matched to any library compound and was labeled as unidentified (see Figure 2C).

Preparation for calcium imaging experiments

Request a detailed protocol

Female moths were tested; they were either virgin or mated on day 1 after emergence. On day 2, moths were positioned in a 15 ml plastic tube with the tip cut open. The head was protruding at the narrow end and was fixed in this position with dental wax. Labial palps and proboscis were also fixed with wax to reduce movement artifacts during the experiments. A window was cut in the head capsule between the compound eyes, and the tissue covering the brain was removed until the antennal lobes were visible. We added 50 µl Pluronic F-127 (Invitrogen) to 50 μg of the membrane-permeant form of a fluorescent calcium indicator (Calcium Green-1 AM, Invitrogen) and sonicated the solution for 10 min. Then, we added 800 µl physiological saline solution (Christensen and Hildebrand, 1987) and sonicated again for 10 min. Then, 20 µl of this dye solution was applied to the exposed brain, and the preparation was incubated in a humid chamber for 45 min at room temperature. Then, we rinsed the brain several times with physiological saline solution to remove excess dye and stored the moths at 4°C overnight to calm them down and reduce their movements. Imaging experiments were performed the following day (day 3 after emergence).

Calcium imaging

Request a detailed protocol

The imaging setup consisted of a CCD camera (Olympus U-CMAD3) mounted to an upright microscope (Olympus BX51WI) equipped with a water immersion objective (Olympus, 10×/0.30). Calcium green-1 AM was excited at 475 nm (500 nm shortpass optical filter; xenon arc lamp, Polychrome V, Till Photonics), and fluorescence was detected at 490/515 nm (dichroic longpass/longpass). The setup was controlled by the software Tillvision version 4.6 (Till Photonics). Fourfold symmetrical binning resulted in image sizes of 344 × 260 pixels, with one pixel corresponding to an area of 4 µm × 4 µm.

Odor stimulation

Request a detailed protocol

To create a functional map of glomeruli in the antennal lobe, we first tested 19 diagnostic odors (Bisch-Knaden et al., 2018) in each animal. Then, we tested 17 headspaces of plants (Table 1) and 1 collection with an empty bag. The same samples as in GC-EAD experiments were used. Then, 10 µl of a diagnostic odor or an eluted headspace were applied onto a circular piece of filter paper (diameter: 12 mm, Whatman) that was inserted in a glass pipette; 10 µl of the solvent mineral oil (diagnostic odors) or the eluent dichloromethane (headspace) served as control stimuli. A pipet tip sealed with dental wax closed the pipettes until the start of the experiment. As dichloromethane alone evoked a response in the antennal lobe (Figure 3B), pipettes with headspace samples and two pipettes with dichloromethane were left open for 3–5 min before sealing them to allow evaporation of the eluent. Filter papers were renewed every experimental day (diagnostic odors) or the pipettes were stored at –20°C and used on up to three experimental days (headspace). The immobilized moth was placed upright under the microscope. A glass tube (diameter: 5 mm) was directed perpendicular to one antenna and delivered a constant stream of charcoal-filtered, moistened air (0.5 l/min). Two glass pipettes were inserted through small holes in the tube. One pipette (inserted 5.5 cm from end of tube) was empty and added clean air to the continuous airstream (0.5 l/min). This airstream could automatically be switched (Syntech Stimulus Controller CS-55) to the second pipette (inserted 3.5 cm from the end of tube) that contained an odor-laden filter paper. By this procedure, the airstream reaching the antenna was not altered during odor stimulation, thus reducing mechanical disturbances. One odor stimulation experiment lasted 10 s and was recorded with a sampling rate of 4 Hz corresponding to 40 frames. The time course of an odorant stimulation experiment was as follows: 2 s clean airstream (frames 1–8), 2 s odorous airstream (frames 9–16), and 6 s clean airstream (frames 17–40). Odors were presented with at least 1 min interstimulus interval to avoid adaptation. The sequence of headspace stimulations changed from animal to animal, and a control stimulus with dichloromethane was presented at the beginning and end of this sequence.

Processing of calcium imaging data

Request a detailed protocol

Stimulation experiments resulted in a series of 40 consecutive frames that were analyzed with custom-written software (IDL, L3Harris Geospatial; Figure 3—source code 1). Several processing steps were applied to enhance the signal-to-noise ratio: (1) background correction: background activity was defined as the average fluorescence (F) of frames 3–7 (i.e., before stimulus onset) and was subtracted from the fluorescence of each frame. This background-corrected value (deltaF) was divided by the background fluorescence to get the relative changes of fluorescence over background fluorescence for each frame (deltaF/F). (2) Bleaching correction: the fluorescent dye bleached slowly during the exposure to light, and therefore, we subtracted from each frame an exponential decay curve that was estimated from the bleaching course of frames 3–7 and frames 26–40 (i.e., before and after stimulus and response). (3) Median filtering: a spatial median filter with a width of 7 pixels was applied to remove outliers. (4) Movement correction: possible shifts of the antennal lobe from one stimulation experiment to the next one were corrected by aligning frame 20 of each experiment to frame 20 of the median experiment in a given animal. The outline of the antennal lobe and remains of tracheae served as guides for this movement correction procedure. Increased neural activity, indicated as an increase of the intracellular calcium concentration after stimulation with the diagnostic odors, was leading to spatially restricted spots of increased fluorescence in the antennal lobe. In the center of each activity spot, the average deltaF/F was recorded in an area of the size of a small- to medium-sized glomerulus (60 µm × 60 µm). Time traces of deltaF/F were averaged over three successive frames for each activity spot. In these smoothed time traces, the maximum deltaF/F after stimulus onset was determined. The average of the maximum value and the value before and after the maximum were calculated and were defined as the response of the animal to the odor stimulation at the given activity spot.

Analysis of response to headspace stimulation

Request a detailed protocol

Activation patterns evoked by the diagnostic odors helped to establish an individual schematic of 23 putative glomeruli for each animal (Figure 3). Then, responses in these 23 glomeruli were calculated for stimulations with headspaces and dichloromethane. To identify headspace-activated glomeruli, we tested for each glomerulus if its mean response towards the two control stimulations with dichloromethane was clearly different from the response to the headspaces (p<0.01, Friedman test with Dunn’s multiple-comparisons test). We then normalized responses for each glomerulus in a given animal according to its response to headspaces and dichloromethane (lowest response = 0, highest response = 100) to balance for variability between individuals.

Statistical analysis

Request a detailed protocol

Sample sizes and statistical tests used are given in the text and figure legends. Statistical tests were performed with PAST (version 3.26, http://folk.uio.no/ohammer/past/) and GraphPad InStat (version 3.10, GraphPad Software, San Diego, CA, https://www.graphpad.com).

Data availability

Figure 1—source data 1, Figure 2—source data 1 and Figure 3—source data 1 contain the numerical data used to generate the figures. Figure 3—source code 1 contains custom written software for processing calcium imaging data.

References

    1. Hansson BS
    2. Carlsson MA
    3. Kalinovà B
    (2003) Olfactory activation patterns in the antennal lobe of the sphinx moth, Manduca sexta
    Journal of Comparative Physiology. A, Neuroethology, Sensory, Neural, and Behavioral Physiology 189:301–308.
    https://doi.org/10.1007/s00359-003-0403-5
    1. King JR
    2. Christensen TA
    3. Hildebrand JG
    (2000)
    Response characteristics of an identified, sexually dimorphic olfactory glomerulus
    The Journal of Neuroscience 20:2391–2399.
  1. Conference
    1. Medina AL
    (2003)
    Historical and recent flora of the Santa Rita Experimental Range
    USDA Forest Service Proceedings, RMRS-P-30. pp. 141–148.
    1. Reisenman CE
    2. Dacks AM
    3. Hildebrand JG
    (2011) Local interneuron diversity in the primary olfactory center of the moth Manduca sexta
    Journal of Comparative Physiology. A, Neuroethology, Sensory, Neural, and Behavioral Physiology 197:653–665.
    https://doi.org/10.1007/s00359-011-0625-x
    1. Stöckl AL
    2. Kelber A
    (2019) Fuelling on the wing: sensory ecology of hawkmoth foraging
    Journal of Comparative Physiology. A, Neuroethology, Sensory, Neural, and Behavioral Physiology 205:399–413.
    https://doi.org/10.1007/s00359-019-01328-2
    1. Torretta JP
    2. Navarro F
    3. Medan D
    (2009)
    Nocturnal floral visitors of sunflower (Helianthus annuus, Asterales: Asteraceae) in Argentina
    Revista de La Sociedad Entomologica Argentina 68:339–350.
    1. Webster B
    2. Cardé RT
    (2017) Use of habitat odour by host-seeking insects
    Biological Reviews of the Cambridge Philosophical Society 92:1241–1249.
    https://doi.org/10.1111/brv.12281

Decision letter

  1. Marcel Dicke
    Reviewing Editor; Wageningen University, Netherlands
  2. Meredith C Schuman
    Senior Editor; University of Zurich, Switzerland
  3. Sylvia Anton
    Reviewer; INRAE, France

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Unique neural coding of crucial versus irrelevant plant odors in a hawkmoth" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individual involved in the review of your submission has agreed to reveal their identity: Sylvia Anton (Reviewer #3).

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided that this work will not be considered further for publication by eLife.

Your manuscript addresses the olfactory signals that guide the specialist hawkmoth Manduca sexta towards plants that are used for oviposition and for nectar feeding in a natural environment in Southern Arizona, US. The study describes how different complex, ecologically-relevant olfactory signals are detected and represented in the moth brain. Its interest derives from the fact that insects usually live in a complex odorant space, which contains myriad odorants. How insects distinguish useful information from irrelevant information is an important question.

While being well done, the reviewers found the innovation of the study limited, as exemplified by a range of published papers; and they also raised the need to connect behavioural data and neurobiological data. In sum, the reviewers see the potential of this work to contribute to our understanding of how moths use plant volatiles to seek nectar and select host plants, but the agreement was that previous relevant work is not sufficiently recognized and reported, and that additional experimental work is needed to provide a substantial advance for the field.

Reviewer #1 (Recommendations for the authors):

General:

As someone interested in chemical ecology and insect-host naturalistic interactions, I highly appreciate and welcome this type of work. The use of natural systems and state of the art methodologies and well-conceived and conducted experiments that couple signals with neuron activity is highly suited for better understanding how behavioral decisions, such as feeding and choosing an appropriate oviposition site, are governed. Technically and intellectually, the experiments are well conducted and the amount of collected information is impressive. The manuscript, however, somehow lacks in novelty because much of it has been previously reported, in particular, the chemical composition and bioactivity of components in the floral odor of D. wrightii and A. palmieri, and how those two nectar sources differentially activate antennal lobe neurons (Riffell et al. 2008, 2009a, 2009b, 2014). These previous results highly overlap with the findings presented here. How moths could identify a behaviorally relevant odor bouquet (such as the floral scent of D. wrightii) in an environment of irrelevant odors (e.g. creosote bush) has also been previously examined (Riffell et al. 2014), and it would be good to discuss this here.

D. wrightii is used both as a nectar source for adults and hostplant for the larvae, but previous work showed that the presence of flowers increases oviposition both in the field and in laboratory experiments (Reisenman et al. 2010), suggesting that floral odors (owing to the high emission of VOCs) can attract females at a distance for oviposition. In particular, a compound within the D. wrightii flower odor (linalool, which is typical of highly reflective white night-blooming flowers) selectively activate PNs in a female specific glomerulus, providing further support for this idea.

The imaging experiments are impressive, in an insect for which neuronal markers that could facilitate this are not available. However, the conclusions from this experiment only apply to those glomeruli (one third approximately) imaged. This is mentioned throughout the manuscript, but the authors should be careful in their statements (e.g. lines 240-242, 247-252, 452-453, 471-473)

Introduction:

Lines 45-46: In the case of Manduca sexta in the environment described (Southern Arizona, USA), D. wrightii floral odors likely serve to additionally signal moths the presence of an oviposition site (Reisenman et al. 2010).

Lines 51-66: Previous work has characterized the components, amounts, and ratios within the natural flowers (same species as this work) used by M. sexta that are capable of mimicking the behavior towards the real flower. The neural activation patterns evoked by these sources at the antennal lobe level have been specifically examined in M. sexta, and the compounds that evoke strong responses identified. (Riffell et al. 2008, 2009a, 2009b, 2013). The influence of background odors (e.g. background vegetation from non-hosts) on the AL representation of behaviorally relevant blends has been examined as well (Riffell et al. 2014).

Lines 79-81: This sentence needs the following citations: Bronstein et al. (2009), Mechaber et al. (2002).

Line 80-81: The following reference could be added here, which discusses the idea of antagonistic mutualism for the system under study: Adler and Bronstein 2004.

Line 93-94: Because about one third of all glomeruli are imaged, it can't be discarded that further responses/discrimination take place in glomeruli not imaged. This should be mentioned whenever relevant, and statements should be cautious (e.g. lines 452-453: "m sexta hostplant bouquets each activated only a single glomerulus"; but see also 240-242, 247-252, 471-473)

Line 96-97: Learning is an important factor that modulates the representation of odors in the AL (Daly et al. 2004, Riffell et al. 2013).

Lines 103-104: This finding is in agreement with previous findings, strong responses to odors related to nectar sources (Riffell et al. 2009a,b). Although odors from vegetative parts of hosts elicit weak responses, in the case of D. wrightii moths likely use the floral odors to signal oviposition sites. Linalool, which is present in the floral scent of D. wrightii (but not in A. palmieri) strongly activates a female-specific olfactory glomeruli (King et al. 2000). Evidence suggests that these glomeruli are necessary and sufficient to mediate oviposition (Kalberer et al. 2010). Vegetative host-plant odors (tomato leafs) have been also reported to evoke responses from the two female-specific glomeruli (King et al. 2000, Reisenman et al. 2009).

Results and discussion:

In general, the findings regarding the number of GC peaks, the chemical composition and identity need to be discussed in the context of previous findings because there's substantial overlap. Previous work identified some of the same chemicals as most abundant and strong activators (e.g. ethyl sorbate in Agave flower, benzyl alcohol, ocimene, geraniol, linalool in Datura flowers, Figure 2C). The host-plant vegetative odors of intact and larva-damaged natural hosts D. wrightii and D. discolor, and cultivated tomato have been investigated previously and should be cited here (Reisenman et al. 2013).

Figure 1: The GC traces in Figure A is not very informative because the reader does not know what the different peaks are. If the purpose is to show that there are a lot more peaks (more components) in the floral nectar, or in host sympatric hosts, that's also achieved in Figure B. Maybe provide a key to some of the components for each sample?

Lines 157-171: Refer to previous findings about those same compounds from at least two of the sources being strong activators (e.g. aliphatic esters, terpenes and aromatics, Figure 2C). Although many components are present, reduced floral mimics with just 3-6 compounds were shown to be sufficient to evoke behavior identical to that evoked by real flower, at least in the laboratory setting (Riffell et al. 2009a,b).

Lines 166-177: It should be mentioned here that projections neurons in a sexually isomorphic glomerulus in M. sexta are selective and highly sensitive (below 10-6 vol/vol dilution in females) to cis-3-hexenyl-acetate (Reisenman et al. 2005).

Line 240: This statement applies only to the glomeruli imaged here (ca. 1/3 of all AL glomeruli) and so this should be acknowledged.

Lines 344-348: It should also be cited a previous study showing high specificity and sensitivity towards cis-3-hexenyl-acetate in projection neurons from an identified glomerulus (Reisenman et al. 2005). These PNs respond most strongly to that compound, but also respond to another ester, cis-3-hexenyl propionate.

Lines 355-357: Strong responses to Agave floral odors previously reported (Riffell et al. 2009b)

Lines 360-361: A clear distinction should be made between neuron responses and behavioral responses when discussing learning.

Lines 364-366: Indeed there are PNs which are selective and extremely sensitive to this odor compound (below 10-6 vol/vol) in a sexually isomorphic glomerulus (Reisenman et al. 2005).

Lines 399-405: Other reports in M sexta show widespread inhibition in the AL (Lei et al. 2004) and that altering the balance of excitation and inhibition alters blend odor representation impeding tracking (Riffell et al. 2014).

Lines 444-451: Previous findings suggest that these glomeruli are involved in mediating oviposition, at least in part: (1) At least some PNs in these glomeruli (ca. 20%) respond to vegetative odors from tomato (a plant used for only oviposition) (King et al. 2000, Figure 7), and some examples were also reported for the medLFG (Reisenman et al. 2009, Figure 1). (2) Experiments in which the antennal imaginal disk of a female is transplanted in the developing male larvae show that the presence of the induced female glomeruli in these gynandromorph animals is necessary and sufficient for orientation towards host-plants (Kalberer et al. 2010).

It is also possible that the LFGs use floral odors to orient females towards oviposition sites (in the case of datura at least) because: 1) the latLFG (glomerulus # 2 in Figure 3A) is activated by Datura floral odorants (Figure 3B), responds selectively to (+)-linalool (Reisenman et al. 2004, Bisch-Knaden 2018), an odorant which occur in hawkmoth pollinated flowers including D. wrightii, Raguso and Pichersky 1999; Reisenman et al. 2010); 2) females oviposit much more in presence of Datura flowers, including vegetation presenting a mimic floral scent containing (+) but not (-) linalool (Reisenman et al. 2010); responses to the enantiomers being dependent on context and accession (He et al. 2019); 3) the female antenna expresses one female-specific OR which is homologous of a Bombyx mori female-specific OR which detects linalool (Grobe-Wilde et al. 2011, Anderson et al. 2008).

Lines 510-513: It has been shown that M. sexta has reduced oviposition in some larva-damaged hosts and that total emission of VOCs are higher in these plants (Reisenman et al. 2013).

Lines 531-540: The effect of background odors in odor tracking of relevant source has been investigated to some extent in the Datura wrightii/M. sexta system (Riffell et al. 2014), showing that moths can track the source better in backgrounds of non-overlapping odorants.

Reviewer #2 (Recommendations for the authors):

1. This work is based primarily on M. sexta-plant relationships previously reported and does not include behavioral experimental data. It would be very helpful to explain the present results if some behavioral data are available.

2. In the headspace collections, the flowers or branches were used for Agave palmeri and Datura wrightii. Why the authors did not use the flower branches, which should be comparable with the treatments for other plants? From the present data, the flower had a great influence on odorant collection.

3. How about the male moth's responses to these odors? Generally, both male and female moths search for nectar, while only female moths search for oviposition sites. By analyzing how male and female moths respond to these scents, it may be possible to estimate which scents are associated with the nectar searching or oviposition searching.

4. In Figure 2A, the GC-EAD recordings of Datura wrightii was not included, why? I think Datura should be one of the most important plants in the system.

5. Overall, the in vivo calcium imaging experiment is not complete. It is better to link the active glomeruli with some important EAD-active compounds. Are the compounds with high EAD activities the ones that activate strongly the glomeruli?

6. The activation of glomeruli depends on mating status. Is this difference derived from the differential peripheral inputs or the changed modulation on the activity of glomeruli by pre-and postsynaptic modulation?

7. I suggest that the odorants from two hostplants activating the glomeruli in the antenna lobe be identified. It would be very nice if they can also determine the attractiveness of these chemicals to the mated adults.

Reviewer #3 (Recommendations for the authors):

This is an easy to read, very exciting manuscript. I only have a few minor suggestions for improvement.

In Figure 2C at least in the version I have, some numbers are cropped on the right side (those >10).

Line 272: graph depicts

Line 677: replace "fluorescent" either by "fluorescent dye" or "fluorescence"?

Discussion lines 310-315:

When you discuss the antennal detection of volatiles, you talk about discrimination capability. I would maybe not talk about discrimination at this level. You find indeed differential detection, which could provide them with the necessary information to discriminate, but I would maybe formulate this with a little more caution.

In the same paragraph, I am not sure that I can follow your argument that detection and discrimination capability appeared to be better than could be inferred from the chemical analysis.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Unique neural coding of crucial versus irrelevant plant odors in a hawkmoth" for further consideration by eLife. Your revised article has been evaluated by Meredith Schuman (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

The three reviewers are supportive of the revised manuscript and the revisions you made as well as the extensive explanations you provided in the rebuttal. The last suggestions for modification, made by reviewers #1 and #2 involve some explanation to support some statements that will further increase the quality of the manuscript.

Reviewer #1 (Recommendations for the authors):

I thank the authors for taking the time to carefully consider the reviewers comments, making suggested changes, clarifying some of their statements, etc. The addition of sentences clarifying differences and similarities with previous studies, both methodological and in terms of findings, is very helpful and useful for a reader who is not particularly familiar with the specifics of the system under study.

With the changes introduced by the authors, I think that the manuscript now better conveys the novelty and relevance of the findings: while much was previously known about some of the important sources used by hawkmoths for nectar feeding and oviposition (e.g. Datura and Agave flowers, Datura and Prosbocidea foliage), the comparison in terms of chemistry and antennal and antennal lobe responses with those of non-host plant provides a framework for better explaining how moths can distinguish relevant from irrelevant hostplants and nectar sources.

I think that in its present form, this is a very interesting study about how specialist insects find olfactory important resources, distinguishing from those that are irrelevant or non-suitable, in their environmentally complex olfactory environments. The combination of field collections in the moths' natural environmental, and of chemical analysis coupled to recording from peripheral olfactory organs and imaging of neuronal activity in the primary olfactory center, plus comparison between virgin and mated females, is a strength of the manuscript.

One general comment that I have is regarding the comparison between antennal responses to single (GC-EAD) compounds and antennal lobe responses (vegetative/floral blends), and comparison with previous studies which use GC-multiunit recording. It is not surprising to me that the full blend evokes responses in many glomeruli, as the responses at this level might also be due to interglomerular interactions (reciprocal synapses, inhibition, etc.): it is possible that single compounds evoke responses in few glomeruli, while blends evoke wide-spread responses. I think it is a good addition that the authors now describe the difference between methodologies and what can be learned from each of them. There's also a couple of points (in the list below) that I think the authors should revise (comments on lines 461-471, 502-503, 571-573).

Line 20: "Responses to bouquets"

Line 77: the sentence starting with "However.." should be directly after the previous one for better flow of the logic.

Line 263: "…mating status (Table 2), at least among the 23 glomeruli imaged in this study."

Line 264: … activated only one glomerulus out of the 23 imaged in the antennal lobe…."

Regarding Figure 3C: which plant headspace is used here? This needs clarification. I assume is a single species headspace -it wouldn't be appropriate to mix headspaces (the captions says "plant headspaces").

Line 345: between the bouquets of host plant vegetation and surrounding…"

Line 351: "..in typical hawkmoth-pollinated flowers…"

Line 359: "…. Is at least as sensitive to promising floral blend…"

Line 367: "…with the duration a female shows feeding behavior (i.e proboscis contact time with a scented filter paper flower, Pearson correlation…., Bisch-Knaden et al. 2018). In contrast…"

Line 371: "… i.e. a behavior related to oviposition….."

Line 381: "In addition, the antenna might harbor narrowly tuned olfactory receptor neurons strongly responding…"

Line 414: "… the active GC-peaks overlapped between…"

Line 424: "…influence both the composition and the quantity…"

Line 444: "… presynaptic level, i.e. at the level…"

Line 450: "…. not elicit activity in the 23 imaged antennal lobe glomeruli."

Lines 461-471: I think the explanation for the different studies misses the fact that while single components might not evoke broad activity at the AL lobe level, blends/mixtures might do so due to emergent properties of AL circuitry. The current study uses blends as stimuli for imaging of AL activity, which might explain that the authors found broad activation across the array of imaged glomeruli. So the two studies not only used different techniques, each with its own advantages, but seek to answer different questions. Indeed, in their previous publication (Bisch-Knaden et al. 2018), they used monomolecular odorants and for the most part each odorant activates a few glomeruli (at least medium to strong, Figure 2D), including the esters. In line 469 the authors say that ethyl sorbate and benzyl alcohol evoke responses in the AL in the previous study; the responses to ethyl sorbate are small and limited mostly to glomeruli 12 and a few others; similarly, the responses to benzyl alcohol were not very strong and involved about 5 glomeruli. In their previous study, monoterpenes seem to evoke the strongest responses and more widespread (i.e. involving more glomeruli). The way the ms refers to these results gives the reader the impression that the single monomolecular odorants evoke broad responses, comparable to the blend-evoked responses, which I don't think is correct. I suggest that the authors modify this paragraph accordingly.

Line 500: "… to evoke a response in sensilla targeting mostly (but not only) the two female specific glomeruli" (Shields & Hildebrand shows that while most sensilla dye-filled target the LFGs, some target a few other glomeruli).

Lines 502-503: I still think that the authors do not have sufficient arguments for the statement at it is in this sentence. This is because: 1) the authors do not find response to vegetation in the LFGs, but the imaging technique, as the authors state, reveals activity from AL afferent mostly, not AL outputs. Although a cultivated plant, King et al. (2000) and Reisenman et al. (2009) reported conspicuous responses to tomato leaves in one of these glomeruli; 2) it is possible that the LFGs act in concert with other glomeruli to guide oviposition behavior (concerted responses not revealed by imaging of afferents might have important downstream effects); 3) males with induced LFGs fly more towards host plants (because these are the most prominent female specific glomeruli, this suggest that these glomeruli process some odorants which directly or indirectly signal an oviposition site. I thus suggest for this line something like this: "In spite of this, it is still possible that these female-specific glomeruli act in concert with other glomeruli to guide the female-specific behavior of identifying an oviposition site, a hypothesis that need further investigation."

Line 520: "However, host plants activate only one of these glomeruli, …… activated additional glomeruli. While it is possible that host plants activate glomeruli not imaged in this study, the resulting neural representation…"

Line 550: here add Goyret 2010 (J Exp Biol, Look and touch: multimodal sensory control of flower inspection movements in the nocturnal hawkmoth Manduca sexta).

Line 552: "…when searching for oviposition…"

Line 571-573: I still think that the case of D. wrightii is particularly interesting and the fact that the plant has a faint vegetative scent but powerful floral odor has significance. It is entirely possible and supported by previous findings that female might simply use the floral odors for long distance olfactory attraction, as these are fragrant and abundant (Raguso et al. 2003, and evoke strong responses). Once in the vicinity or closer, females might use vegetative odors to decided whether or not oviposit on leaf tissues (in addition to feeding on nectar, as it is known that females mix feeding and oviposition bouts). Therefore, the two processes, oviposition and feeding, guided by weak and strong odors but at different timescales, might entangled with each other in the M. sexta-D. wrightii system. Also, it is commonly observed that M. sexta moths mix oviposition and feeding bouts on this plant, and it is reported that flowers increased oviposition both in the lab and in the field (Reisenman et al. 2010). I suggest the authors modify the sentence starting with "Hence, M. sexta…" to reflect this fact.

Line 583: "… in contrast to the weak but specific activation of single glomeruli (among those imaged) by host plants of M. sexta."

Reviewer #2 (Recommendations for the authors):

The authors responded to the questions I raised, and then I think that this work deserves publication in the journal. However, I still have two suggestions:

(1) Since Datura wrightii is a unique and important plant in this study system, which is both the nectar source and host plant of this moth species, I'm sticking to my opinion that the representative traces of GC-EAD and in vivo calcium imaging recordings of headspace stimulations of the flowers and foliage of Datura wrightii should be added in Figure 2A and Figure 3B although these data were reflected in other figures and source data. For floral and foliage odors of Datura could attract females for oviposition, a comparison of responses of antennae or antennal lobes between floral and foliage of the same plant would be significant.

(2) The EAD activities and calcium imaging activities are the core contents of this study, it is better to analyze their linkage. As the dye used in vivo calcium imaging experiments is Calcium Green-1 AM, the information reflected in the calcium imaging activity may be mainly the input of the olfactory sensory neurons from the antennae. Therefore, there should be positive relationships between EAD activities and calcium imaging activities in theory. I am very curious about the following questions: based on the authors' previous research (Bisch-Knaden et al., 2018), which compound(s) could elicit activities of the only glomerulus (Glomerulus 4) activated by Datura foliage? Were these compounds present in the headspace blend of Datura foliage? If so, how about EAD responses or even behavioral responses to these compounds? The same questions with Glomerulus 15 and 12 for Proboscidea. These could be discussed in the Discussion section.

https://doi.org/10.7554/eLife.77429.sa1

Author response

[Editors’ note: The authors appealed the original decision. What follows is the authors’ response to the first round of review.]

Reviewer #1 (Recommendations for the authors):

General:

1. As someone interested in chemical ecology and insect-host naturalistic interactions, I highly appreciate and welcome this type of work. The use of natural systems and state of the art methodologies and well-conceived and conducted experiments that couple signals with neuron activity is highly suited for better understanding how behavioral decisions, such as feeding and choosing an appropriate oviposition site, are governed. Technically and intellectually, the experiments are well conducted and the amount of collected information is impressive.

We thank the reviewer for this positive evaluation of our study in general.

2. The manuscript, however, somehow lacks in novelty because much of it has been previously reported,

We are surprised about the statement that our study lacks novelty.

A) We collected nocturnal headspace in the field (17 samples from 16 different plant species); 13 of these 17 headspaces (76% new data) have not been published before to the best of our knowledge. The previously published headspaces are floral scents of Agave and Datura (see our reply to point 3), and leaf scents of Datura and Proboscidea (see B).

B) We next tested the antenna of female moths in GC-EAD experiments. Of the species tested here, only Datura foliage and Proboscidea have been studied before by Fraser et al. 2003; a study we mention in the introduction (line 58) and the discussion in detail (line 374-390). In Riffell et al. 2014, a GC-EAD example trace for stimulation of a male antenna with Datura headspace is shown in the supplementary material. However, no detailed analysis is presented except two arrowheads marking EAD-responses towards unspecified oxygenated aromatic compounds. Hence, GC-EAD results from 15 out of 17 tested headspaces are new data (88%).

C) Regarding headspace-evoked activation of identified glomeruli in the female antennal lobe, we are aware of two studies showing results for one single glomerulus in each. King et al. 2000 recorded intracellularly from projection neurons innervating one identified glomerulus using a tomato leaf as stimulus (line 445). Reisenman et al. 2009 recorded from another identified glomerulus using also a tomato leaf. We now added this reference based on the reviewer’s comment (see our reply to point 18). However, headspace-evoked spatial activation patterns in an array of identified glomeruli (n=23) in the female antennal lobe have not been investigated before in M. sexta (100% new data; tomato was not included in our study as it is a crop species not present in the natural habitat under investigation). Furthermore, we studied for the first time the effect of the females’ mating status on detection at the antenna and for spatial coding in the antennal lobe.

3. … in particular, the chemical composition and bioactivity of components in the floral odor of D. wrightii and A. palmieri, and how those two nectar sources differentially activate antennal lobe neurons (Riffell et al. 2008, 2009a, 2009b, 2014). These previous results highly overlap with the findings presented here.

Raguso et al. 2003 and Raguso 2004 first described the chemical composition of floral odors of D. wrightii and A. palmeri. In our original manuscript, we cited Raguso et al. 2003 for the scent of Datura, but unfortunately missed citing Raguso 2004 for the scent of Agave. We now added Raguso 2004, and based on the reviewer’s suggestion, we added as well Riffell et al. 2008 as an additional reference for the scent of these two flowers (revised line 88).

Regarding the chemical composition, we appreciate that our results for Agave and Datura overlap with the former studies and now mention this in the discussion (revised line 323-326). “Our results regarding the number and identity of compounds emitted by Agave and Datura flowers, and the dissimilarity between both floral bouquets confirm earlier studies (Raguso 2004; Raguso et al. 2003; Riffell et al. 2008).”

Regarding the bioactivity of components in the same two flowers that Riffell et al. studied in the male antennal lobe, we do not agree with the reviewer’s opinion as there is only little overlap with our results. Riffell et al. 2009 report that the neurons they recorded from responded only to a small fraction (Datura: 9 compounds, Agave: 6 compounds) of the > 60 compounds present in each of the headspaces. We found in our GC-EAD recordings that more fractions were active at the level of the antenna (Datura: 17 compounds, Agave: 20 compounds), and in our calcium imaging recordings, we found that both floral bouquets activated each of the 23 identified glomeruli in the antennal lobe.

One reason for the discrepancy between Riffell et al. and our study might be that Riffell et al. used male moths while we worked with females. Probably most important, however, are the different recording methods used by the labs. We briefly discussed this topic in a previous paper (Bisch-Knaden et al. 2018), but realize that it might be necessary to discuss it here again and in more detail as the reviewer and many readers might not be aware of the methodological differences, which have significant effects on the results and conclusions. Riffell et al. investigated antennal lobe activation in male moths using a multiunit neural-ensemble probe with 4 recording shanks in a linear row, with a distance of 125 μm from each other and a recording depth of 200 μm, i.e. about half the depth of the antennal lobe. The first shank is inserted in the macroglomerular complex as a reference point and not analyzed further, the other 3 shanks are inserted parallel to the axis of the antennal lobe with each shank impaling up to 3 glomeruli in the lateral region of the lobe, i.e. less than 10 glomeruli are impaled in each animal (Riffell et al. 2009). From experiment to experiment, the targeted glomeruli are adjacent but not always identical due to the impossibility to place the recording probe exactly in the same position and anatomical variability between animals. This recording method therefore allows analyzing the temporal characteristics of odor-evoked activity in the respective region (several glomeruli and some part of the central neuropil) of the male antennal lobe. However, it does not allow conclusions regarding the activation patterns across identified glomeruli; a limitation that is stated by Riffell and colleagues in all of their publications: “Although identification of individual glomeruli is an important prerequisite for assigning functional significance to a given glomerulus, it was beyond the scope of this study.” This is, in contrast, the strength of our calcium imaging study as we could consistently address the same 23 glomeruli in the dorsal part of the antennal lobe by testing a panel of diagnostic odorants in each animal based on our previous work (Bisch-Knaden et al. 2018), followed by stimulation with plant headspaces. We were thus able to assign functional significance to a given glomerulus and analyze the spatial representation of the tested headspaces across the same 23 glomeruli in the female antennal lobe. On the other hand, we did not analyze temporal coding patterns because calcium signals are rather slow, and as calcium influx indicates neural activation, no clear conclusions about inhibitory networks can be drawn. The analysis of temporal characteristics of odor-evoked activation of mixed populations of local interneurons and projection neurons in the antennal lobe, and the study of inhibitory interactions between these neurons, are the strength and advantage of the multiunit neural-ensemble recording method used by the Riffell lab. We therefore added a paragraph in the introduction explaining the different methods (revised line 59-72).

“At the level of the first olfactory processing center in the insect brain, the antennal lobe, previous studies have investigated the temporal coding patterns evoked by floral bouquets. By impaling a restricted region of the antennal lobe with a multiunit probe, simultaneous recordings from a mixed population of local interneurons and projection neurons in this area were performed (Riffell et al. 2009a+b). Using the same approach, inhibitory interactions between these neurons could be studied in addition (Lei et al. 2004). However, this recording technique does not allow assigning functional significance to individual olfactory glomeruli, which are the morphological and functional subunits of the antennal lobe (Gao et al. 2000; Hansson et al. 1992). An analysis of the spatial coding patterns, however, is possible via functional calcium imaging. Although this technique does not inform about temporal coding patterns or inhibitory interactions, it was used in different insect species to provide a detailed insight into the spatial representation of natural odor blends across the glomerular array.”

4 How moths could identify a behaviorally relevant odor bouquet (such as the floral scent of D. wrightii) in an environment of irrelevant odors (e.g. creosote bush) has also been previously examined (Riffell et al. 2014), and it would be good to discuss this here.

We apologize that we missed mentioning this paper in the original version of the manuscript. The findings by Riffell et al. 2014 are indeed interesting; however, they seem contradictory to the fact that creosote bushes (the only background species studied in Riffell et al. 2014) are obviously omnipresent in some Datura/Manduca environments, while Datura still regularly becomes pollinated by the moth in the same habitat (Alarcon et al. 2008). We did not observe creosote bushes in the near vicinity of the Datura plants in our restricted study area, at the time (August 2018) when we collected headspaces. We therefore could not include creosote in our collection of eleven background plant species. Anyhow, we now added the study of Riffell et al. 2014 in the discussion (revised line 463-465). “Interestingly, lab experiments in a wind tunnel suggest that even such a strong scent as the one from Datura flowers can become less attractive to male M. sexta when presented in the olfactory background of a selected non-host plant (Riffell et al. 2014).”

5. D. wrightii is used both as a nectar source for adults and hostplant for the larvae, but previous work showed that the presence of flowers increases oviposition both in the field and in laboratory experiments (Reisenman et al. 2010), suggesting that floral odors (owing to the high emission of VOCs) can attract females at a distance for oviposition.

We agree that this is interesting to mention and added the reference to a sentence in the introduction (revised line 94-98). “Datura plants thus have to interact with an insect that is at the same time an important pollinator and a damaging herbivore (Bronstein et al. 2009), enabling the moth to find an oviposition site by navigating towards the scent of nectar-providing flowers emitted by the same plant (Reisenman et al. 2010).”

However, as Datura foliage alone can attract egg-laying M. sexta, we decided to analyze bouquets from foliage and flowers separately and added an explanation to the method section (revised line 606-609). “Although flowering Datura plants receive more eggs than non-flowering Datura plants (Reisenman et al. 2010), foliage alone attracts egg-laying females in the field (Allmann et al. 2013) and in the lab (Spaethe et al. 2013). We therefore tested flowers and leaves of Datura separately.”

6. In particular, a compound within the D. wrightii flower odor (linalool, which is typical of highly reflective white night-blooming flowers) selectively activate PNs in a female specific glomerulus, providing further support for this idea.

Linalool is a very common volatile organic compound, not only a floral odor. Linalool also acts for example as indirect plant defense against herbivores; therefore, linalool is one of the compounds that can repel ovipositing moths (Kessler and Baldwin 2001). In our study, linalool was not only emitted from Datura flowers but in addition from non-flowering plants at a concentration that was EAD-active (Figure 2C). Finally, several glomeruli besides the female-specific LFGs respond to linalool (Hansson et al. 2003, Bisch-Knaden et al. 2018, Reisenman et al. 2004). In our opinion, discussing a single odorant with many and sometimes opposing potential meanings for M. sexta would distract the reader from the topic of our manuscript.

7. The imaging experiments are impressive, in an insect for which neuronal markers that could facilitate this are not available. However, the conclusions from this experiment only apply to those glomeruli (one third approximately) imaged. This is mentioned throughout the manuscript, but the authors should be careful in their statements (e.g. lines 240-242, 247-252, 452-453, 471-473)

We indeed mention this throughout the manuscript: results line 223-239, Figure 3, Table 2, methods line 694-6. Based on the reviewer’s suggestion we added this information again when we discuss the sparse representation of host plant and background odors.

“In contrast to the wide and strong activation of antennal lobe glomeruli by flower odors, M. sexta’s host plant bouquets each activated only a single glomerulus of the 23 glomeruli under investigation.” (revised line 496-497).

“Interestingly, virgin and mated females differed markedly in their response to the odor of background plants: out of the 23 identified glomeruli, these plants activated only a small number in virgin females (range: 0-9), and even less glomeruli (range: 0-2) in mated females.” (revised line 515-517).

Introduction:

8. Lines 45-46: In the case of Manduca sexta in the environment described (Southern Arizona, USA), D. wrightii floral odors likely serve to additionally signal moths the presence of an oviposition site (Reisenman et al. 2010).

Please see our reply to point 5.

9. Lines 51-66: Previous work has characterized the components, amounts, and ratios within the natural flowers (same species as this work) used by M. sexta that are capable of mimicking the behavior towards the real flower.

This comment is about the floral scents of Agave and Datura, and their reduced mimics that were described as being as attractive to male M. sexta as the whole bouquets (Datura: 9 compounds in Riffell et al. 2009a, and 3 compounds in Riffell et al. 2009b; Agave: 6 compounds (Riffell et al. 2009b)). There was, however, no difference in the attractiveness of the main component of the published floral mimics and the whole mimic when females were tested in the wind tunnel (Bisch-Knaden et al. 2018). We already discussed in Bisch-Knaden et al. 2018 that this discrepancy might be based on the sex of the moths tested (Riffell used male moths), or on methodological differences. We, therefore, would prefer to not cite Riffell et al. 2009a+b in this context, as the discussion of the contradictory results would distract too much and is not relevant for the present study.

10. The neural activation patterns evoked by these sources at the antennal lobe level have been specifically examined in M. sexta, and the compounds that evoke strong responses identified. (Riffell et al. 2008, 2009a, 2009b, 2013).

Please see our reply to point 3.

We added a paragraph in the discussion dealing with this (revised line 453-463):

“In former studies, only a small fraction of the compounds present in the two floral bouquets (Agave: 10%, Datura: 15%) activated neurons in the antennal lobe of male M. sexta (Riffell et al. 2009a+b). However, the recording technique used by Riffell and colleagues targets about 10 glomeruli and part of the adjacent neuropil in the lateral part of the male antennal lobe, while we recorded from 23 identified glomeruli in the dorsal part of the female lobe. The differing results might thus be due to these different approaches or to potential sex-specific coding differences. However, compounds that were identified to activate neurons in the male antennal lobe, like ethyl sorbate (Agave) and benzyl alcohol (Datura), were also EAD-active in our present study (Figure 2C) and evoked responses in the female antennal lobe in a previous imaging study (Bisch-Knaden et al. 2018).”

11. The influence of background odors (e.g. background vegetation from non-hosts) on the AL representation of behaviorally relevant blends has been examined as well (Riffell et al. 2014).

Please see our reply to point 4.

12. Lines 79-81: This sentence needs the following citations: Bronstein et al. (2009), Mechaber et al. (2002).

We now cite Bronstein et al. 2009 as a reference for the mutualism between pollinating/damaging insects and plants. Mechaber et al. 2002 report the effect of age and mating status on the behavior of female M. sexta (we cited this paper in the original manuscript in another context, line 98). The reviewer probably meant Mechaber et al. 2000, which we cited in the original manuscript just one line afterwards (line 82).

13. Line 80-81: The following reference could be added here, which discusses the idea of antagonistic mutualism for the system under study: Adler and Bronstein 2004.

Adler and Bronstein 2004 deal with the question if the amount of floral nectar might affect larval herbivory in Datura. This topic seems to be beyond the scope of our manuscript.

14. Line 93-94: Because about one third of all glomeruli are imaged, it can't be discarded that further responses/discrimination take place in glomeruli not imaged. This should be mentioned whenever relevant, and statements should be cautious (e.g. lines 452-453: "m sexta hostplant bouquets each activated only a single glomerulus"; but see also 240-242, 247-252, 471-473)

Please see our reply to point 7.

15. Line 96-97: Learning is an important factor that modulates the representation of odors in the AL (Daly et al. 2004, Riffell et al. 2013).

We agree that learning is an important factor; however, we tested only naïve animals with no previous experience with plant odors or food sources. To clarify this point already in the introduction, we moved a sentence from the discussion to the introduction (revised line 115-117).

“The moths tested in our study were laboratory-reared on artificial diet, naïve to plant odors, not fed, and tested only once, as we were interested in the insets’ innate neuronal responses.”

16. Lines 103-104: This finding is in agreement with previous findings, strong responses to odors related to nectar sources (Riffell et al. 2009a,b).

The reviewer mentioned this before; please see our reply to points 3 and 10.

17. Although odors from vegetative parts of hosts elicit weak responses, in the case of D. wrightii moths likely use the floral odors to signal oviposition sites.

The reviewer mentioned this before; please see our reply to point 5.

18. Linalool, which is present in the floral scent of D. wrightii (but not in A. palmieri) strongly activates a female-specific olfactory glomeruli (King et al. 2000). Evidence suggests that these glomeruli are necessary and sufficient to mediate oviposition (Kalberer et al. 2010). Vegetative host-plant odors (tomato leafs) have been also reported to evoke responses from the two female-specific glomeruli (King et al. 2000, Reisenman et al. 2009).

We cited King et al. 2000 in the original manuscript as a reference for the assumption that LFGs might be involved in mediating oviposition (line 445), and now added Reisenman et al. 2009 as well (revised line 488). However, Reisenman et al. 2009 show an example tomato-response trace but give no sample size or analyzed data for this result. We therefore previously judged the tomato-result as anecdotal evidence. There is, however, substantial data presented in Shields and Hildebrand 2000, who found that vegetative headspace from tomato (13 sensilla tested), failed to evoke a response in sensilla targeting the two LFGs. In addition, no response was found for leaves of Datura (18 sensilla tested) and Nicotiana (3 sensilla tested). We cited this paper in our original manuscript as a reference for the assumption that LFGs might not be involved in mediating oviposition (line 449).

The gynandromorph males in Kalberer et al. 2010 developed a complete female antennal lobe; we see no evidence that only the LFGs might be responsible for the observed female-like attraction towards host plants. Besides the two LFGs, there are three more female-specific glomeruli (Grosse-Wilde et al. 2011). We are currently investigating the role of female-specific glomeruli and the corresponding receptors in another study.

Results and discussion:

19. In general, the findings regarding the number of GC peaks, the chemical composition and identity need to be discussed in the context of previous findings because there's substantial overlap. Previous work identified some of the same chemicals as most abundant and strong activators (e.g. ethyl sorbate in Agave flower, benzyl alcohol, ocimene, geraniol, linalool in Datura flowers, Figure 2C). The host-plant vegetative odors of intact and larva-damaged natural hosts D. wrightii and D. discolor, and cultivated tomato have been investigated previously and should be cited here (Reisenman et al. 2013).

The reviewer mentioned this before; please see our reply to points 3 and 10.

Reisenman et al. 2013 investigated two Datura species and cultivated tomato. They show that oviposition was reduced in larva-damaged tomato plants but not in larva-damaged Datura plants, although all damaged plant species emitted more volatiles than intact plants. Tomato is a crop species not involved in our study, and in addition, we studied the headspace of natural plant populations with omnipresent herbivory. We therefore prefer not citing this reference here in order not to confuse the reader.

20. Figure 1: The GC traces in Figure A is not very informative because the reader does not know what the different peaks are. If the purpose is to show that there are a lot more peaks (more components) in the floral nectar, or in host sympatric hosts, that's also achieved in Figure B. Maybe provide a key to some of the components for each sample?

In Figure 1A, we show representative GC traces for each headspace to illustrate i) the abundance of compounds, analyzed in Figure 1B, and ii) the difference in chemical composition (GC-peaks at different retention times) and in the amount of emissions (GC-peaks of different heights), displayed in Figure 1C.

However, in the representative GC-EAD traces shown in Figure 2A, we provide a key for those GC-peaks that are EAD-active (shown in Figure 2C) as only the identity of biologically active compounds is interesting in our context. Therefore, we would like to keep the presentation as it is.

21. Lines 157-171: Refer to previous findings about those same compounds from at least two of the sources being strong activators (e.g. aliphatic esters, terpenes and aromatics, Figure 2C).

In this section of the results, we describe our GC-EAD results across all 17 plant headspace samples. Regarding previous results for floral odors of Agave and Datura, see our reply to point 3 and 10.

22. Although many components are present, reduced floral mimics with just 3-6 compounds were shown to be sufficient to evoke behavior identical to that evoked by real flower, at least in the laboratory setting (Riffell et al. 2009a,b).

The reviewer mentioned this before; please see our reply to point 9.

23. Lines 166-177: It should be mentioned here that projections neurons in a sexually isomorphic glomerulus in M. sexta are selective and highly sensitive (below 10-6 vol/vol dilution in females) to cis-3-hexenyl-acetate (Reisenman et al. 2005).

Thanks for this suggestion; we added the reference in the discussion (revised line 398-401).

“Interestingly, some projection neurons in the female antennal lobe targeting an identified glomerulus were reported to specifically respond to low concentrations of (Z)-3-hexenyl acetate (Reisenman et al. 2005). “

24. Line 240: This statement applies only to the glomeruli imaged here (ca. 1/3 of all AL glomeruli) and so this should be acknowledged.

The reviewer mentioned this before; please see our reply to point 7.

25. Lines 344-348: It should also be cited a previous study showing high specificity and sensitivity towards cis-3-hexenyl-acetate in projection neurons from an identified glomerulus (Reisenman et al. 2005). These PNs respond most strongly to that compound, but also respond to another ester, cis-3-hexenyl propionate.

The reviewer mentioned this before; please see our reply to point 23.

26. Lines 355-357: Strong responses to Agave floral odors previously reported (Riffell et al. 2009b)

Response to Agave odors are reported by Riffell et al. 2009b to be only moderate (only 6 out of >60 Agave components evoke a response in the neurons). Please see our reply to point 3 and 10.

27. Lines 360-361: A clear distinction should be made between neuron responses and behavioral responses when discussing learning.

We changed “responses” to “neuronal responses” and moved the sentence to the introduction (revised line 115-117) based on point 15.

28. Lines 364-366: Indeed there are PNs which are selective and extremely sensitive to this odor compound (below 10-6 vol/vol) in a sexually isomorphic glomerulus (Reisenman et al. 2005).

The reviewer mentioned this before; please see our reply to point 23.

29. Lines 399-405: Other reports in M sexta show widespread inhibition in the AL (Lei et al. 2004) and that altering the balance of excitation and inhibition alters blend odor representation impeding tracking (Riffell et al. 2014).

The reviewer mentioned this before; please see our reply to point 3.

30.1. Lines 444-451: Previous findings suggest that these glomeruli are involved in mediating oviposition, at least in part: (1) At least some PNs in these glomeruli (ca. 20%) respond to vegetative odors from tomato (a plant used for only oviposition) (King et al. 2000, Figure 7), and some examples were also reported for the medLFG (Reisenman et al. 2009, Figure 1). (2)

The reviewer mentioned this before; please see our reply to point 18.

30.2. Experiments in which the antennal imaginal disk of a female is transplanted in the developing male larvae show that the presence of the induced female glomeruli in these gynandromorph animals is necessary and sufficient for orientation towards host-plants (Kalberer et al. 2010).

The reviewer mentioned this before; please see our reply to point 18.

31. It is also possible that the LFGs use floral odors to orient females towards oviposition sites (in the case of datura at least) because:

31.1. The latLFG (glomerulus # 2 in Figure 3A) is activated by Datura floral odorants (Figure 3B), responds selectively to (+)-linalool (Reisenman et al. 2004, Bisch-Knaden 2018), an odorant which occur in hawkmoth pollinated flowers including D. wrightii, Raguso and Pichersky 1999; Reisenman et al. 2010).

All 23 glomeruli tested in our study respond to Datura and Agave floral headspace (Figure 3, Table 2), and several glomeruli respond to linalool (Hansson et al. 2003, Bisch-Knaden et al. 2018, Reisenman et al. 20004). The finding in our previous imaging study (Bisch-Knaden et al. 2018) that glomerulus #2 responded stronger to (+)-linalool than to (-)-linalool was indeed a hint that glomerulus #2 might be the lateral LFG based on results from Reisenman et al. 2004. This enantiomer-selective response of the lateral LFG is definitely fascinating but we see little relevance to discuss this in our present work: i) Datura flowers emit a 50:50 mixture of linalool enantiomers (Reisenman et al. 2010), ii) Datura foliage emits no linalool (Reisenman et al. 2010 and our manuscript Figure 2C), iii) the enantiomeric ratio of linalool in the headspace of the host plant Proboscidea is unknown, and iv) linalool was an EAD-active compound emitted by a non-flowering host plant for sympatric hawkmoths and two background plants, one with flowers and one without flowers (Figure 2C), showing that linalool is not flower-specific. However, we feel that discussing this in detail is beyond the scope of this manuscript. Please see also our reply to point 6.

31.2. Females oviposit much more in presence of Datura flowers, including vegetation presenting a mimic floral scent containing (+) but not (-) linalool (Reisenman et al. 2010); responses to the enantiomers being dependent on context and accession (He et al. 2019);

Please see our comment to 5 and 31(1).

31.3. The female antenna expresses one female-specific OR which is homologous of a Bombyx mori female-specific OR which detects linalool (Grobe-Wilde et al. 2011, Anderson et al. 2008).

The expression of two female-specific ORs on the antenna of M. sexta that are homologous to a female-specific, linalool-detecting OR of B. mori (Grosse-Wilde et al. 2011, Anderson et al. 2009) is very interesting and implies that at least one of the two homologous M. sexta ORs might also detect linalool. We think this is a very plausible hypothesis; however, we see little relevance to discuss this single odorant (linalool) in our work dealing with the detection and coding of crucial versus irrelevant plant bouquets (see our arguments in response to points 6,18, 19, and 31(1). We briefly discuss only the three single odorants that were found to be extremely effective in activating the antenna although they were present in low concentrations (a-copaene, (Z)-3-hexenyl acetate, b-ocimene, Figure 2D), and the aliphatic esters present in Agave flower as they activated the antenna to a high degree (Figure 2D, E).

32. Lines 510-513: It has been shown that M. sexta has reduced oviposition in some larva-damaged hosts and that total emission of VOCs are higher in these plants (Reisenman et al. 2013).

The reviewer mentioned this before; please see our reply to point 19.

33. Lines 531-540: The effect of background odors in odor tracking of relevant source has been investigated to some extent in the Datura wrightii/M. sexta system (Riffell et al. 2014), showing that moths can track the source better in backgrounds of non-overlapping odorants.

The reviewer mentioned this before; please see our reply to point 4.

Reviewer #2 (Recommendations for the authors):

1. This work is based primarily on M. sexta-plant relationships previously reported and does not include behavioral experimental data. It would be very helpful to explain the present results if some behavioral data are available.

The reviewer is right; our study is based on the well-documented ecological meaning of plant species that are nectar sources and host plants for M. sexta in its habitat in Arizona. This previously reported knowledge enabled us to decide which of the plants in the field are relevant for M. sexta and which are irrelevant. However, we did not aim at challenging previous results or showing again, which plants are relevant for M. sexta and which are not relevant by performing behavioral experiments. (see also our reply to the reviewer’s comment above.)

2. In the headspace collections, the flowers or branches were used for Agave palmeri and Datura wrightii. Why the authors did not use the flower branches, which should be comparable with the treatments for other plants? From the present data, the flower had a great influence on odorant collection.

We thank the reviewer for raising this question and now explain our choice in more detail in the method section (revised line 599-609) “We sampled from flowering plants or flowering branches if the respective plant was blooming during the experimental nights. Otherwise, non-flowering branches were sampled (Table 1). Agave is a succulent plant with a basal rosette of sharp-edged leaves, each with a length of c. 1m and long spines at the tip. These leaves did not fit in our collection bags (see below). We therefore only collected headspace from Agave flowers, which appear in umbels at the end of a long bloom stalk without leaves (about 5 to 6 m above the basal rosette of leaves). In the case of Datura, we collected headspace separately from flowers and from foliage because the flower is a valuable nectar source and the leaves are an oviposition substrate for M. sexta. Although flowering Datura plants receive more eggs than non-flowering Datura plants (Reisenman et al. 2010), foliage alone attracts egg-laying females in the field (Allmann et al., 2013) and in the lab (Spaethe et al. 2013). We therefore tested flowers and leaves of Datura separately.”

3. How about the male moth's responses to these odors? Generally, both male and female moths search for nectar, while only female moths search for oviposition sites. By analyzing how male and female moths respond to these scents, it may be possible to estimate which scents are associated with the nectar searching or oviposition searching.

It would indeed be interesting to compare results from male and female moths. This is part of future projects in our lab; however, this comparison would go beyond the scope of our present study.

4. In Figure 2A, the GC-EAD recordings of Datura wrightii was not included, why? I think Datura should be one of the most important plants in the system.

We present Agave as an example GC-EAD trace for nectar sources because Agave had the highest number of EAD-active compounds of all headspaces tested (Figure 2B, C). Results for Datura flower and all other headspaces are displayed in the heatmap in Figure 2C, and the EAD-responses of all individual antenna towards all headspaces tested can be found in the source data file S2.

5. Overall, the in vivo calcium imaging experiment is not complete. It is better to link the active glomeruli with some important EAD-active compounds. Are the compounds with high EAD activities the ones that activate strongly the glomeruli?

Our approach was to study activation patterns across the antennal lobe using a puff of plant bouquet. The patterns we observed seemed to be already modulated by presynaptic inhibition via local interneurons (line 397-405). It would indeed be interesting to compare the antennal lobe activity to all individual compounds of a blend with the activity to the whole blend. However, this would need a very tedious concentration control, as all compounds appear at compound-specific concentrations in the blend. Although interesting, this experiment is beyond the scope of the manuscript. However, we discuss an example odorant, which “elicited a strong antennal response, was present in 11 out of 17 plant samples tested, and when tested on its own activates several glomeruli (Bisch-Knaden et al., 2018). However, two of the background plants although releasing this odor, and accordingly evoking a strong antennal response, did not elicit any activity in the antennal lobe” (line 409-413). Repeating the calcium imaging experiments with all individual GC-identified compounds would be an interesting follow-up study but is beyond the scope of the present manuscript. We, therefore, believe that regarding our interest in how full blends of relevant versus irrelevant plants are coded in the brains of virgin and mated females, the calcium imaging experiment is complete.

6. The activation of glomeruli depends on mating status. Is this difference derived from the differential peripheral inputs or the changed modulation on the activity of glomeruli by pre-and postsynaptic modulation?

We did not find an effect of the females’ mating status at the periphery (see original manuscript line 178-180) and can only speculate (and did so) about possible mechanisms for the observed modulation at the level of the antennal lobe (line 478-482).

7. I suggest that the odorants from two hostplants activating the glomeruli in the antenna lobe be identified. It would be very nice if they can also determine the attractiveness of these chemicals to the mated adults.

We agree that identifying those individual components of host plants that might be most responsible for the attraction of mated females is interesting. However, it might as well be that no individual compound can be found because a blend of odorants at specific ratios might be necessary to attract a mated female to a host plant. This topic is part of a project already started in our lab and will be published in future. Our present study, however, focusses on the analysis of spatial activation patterns evoked by the whole, complex plant headspace.

Reviewer #3 (Recommendations for the authors):

This is an easy to read, very exciting manuscript. I only have a few minor suggestions for improvement.

We are grateful for this very positive and encouraging review.

In Figure 2C at least in the version I have, some numbers are cropped on the right side (those >10).

We apologize for having accidentally cut the figure at its right margin before pasting it into the text. The missing numbers are 11 (for the occurrence of (Z)-3-hexenyl acetate) and 10 (b-ocimene).

Line 272: graph depicts

We corrected this error.

Line 677: replace "fluorescent" either by "fluorescent dye" or "fluorescence"?

We replaced "fluorescent" by "fluorescent dye".

Discussion lines 310-315:

When you discuss the antennal detection of volatiles, you talk about discrimination capability. I would maybe not talk about discrimination at this level. You find indeed differential detection, which could provide them with the necessary information to discriminate, but I would maybe formulate this with a little more caution.

In the same paragraph, I am not sure that I can follow your argument that detection and discrimination capability appeared to be better than could be inferred from the chemical analysis.

We rephrased the paragraph in order to clarify our argument and be more careful with our formulation.

“When we tested the antenna of female M. sexta with plant headspaces using GC-EAD, we found that the moths in most of the cases detect at least some compounds, even in wind-pollinated background vegetation like grass or careless weed, plants that, based on the chemical analysis, had a weak smell consisting of a small number of components. Female moths, both virgin and mated, therefore seem to be equipped with the sensory capability to not only distinguish between strong and complex scents emitted by nectar sources but also between the bouquets of host plants and surrounding background plants.” (revised line 335-341)

[Editors’ note: what follows is the authors’ response to the second round of review.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

The three reviewers are supportive of the revised manuscript and the revisions you made as well as the extensive explanations you provided in the rebuttal. The last suggestions for modification, made by reviewers #1 and #2 involve some explanation to support some statements that will further increase the quality of the manuscript.

Reviewer #1 (Recommendations for the authors):

I thank the authors for taking the time to carefully consider the reviewers comments, making suggested changes, clarifying some of their statements, etc. The addition of sentences clarifying differences and similarities with previous studies, both methodological and in terms of findings, is very helpful and useful for a reader who is not particularly familiar with the specifics of the system under study.

With the changes introduced by the authors, I think that the manuscript now better conveys the novelty and relevance of the findings: while much was previously known about some of the important sources used by hawkmoths for nectar feeding and oviposition (e.g. Datura and Agave flowers, Datura and Prosbocidea foliage), the comparison in terms of chemistry and antennal and antennal lobe responses with those of non-host plant provides a framework for better explaining how moths can distinguish relevant from irrelevant hostplants and nectar sources.

I think that in its present form, this is a very interesting study about how specialist insects find olfactory important resources, distinguishing from those that are irrelevant or non-suitable, in their environmentally complex olfactory environments. The combination of field collections in the moths' natural environmental, and of chemical analysis coupled to recording from peripheral olfactory organs and imaging of neuronal activity in the primary olfactory center, plus comparison between virgin and mated females, is a strength of the manuscript.

One general comment that I have is regarding the comparison between antennal responses to single (GC-EAD) compounds and antennal lobe responses (vegetative/floral blends), and comparison with previous studies which use GC-multiunit recording. It is not surprising to me that the full blend evokes responses in many glomeruli, as the responses at this level might also be due to interglomerular interactions (reciprocal synapses, inhibition, etc.): it is possible that single compounds evoke responses in few glomeruli, while blends evoke wide-spread responses. I think it is a good addition that the authors now describe the difference between methodologies and what can be learned from each of them. There's also a couple of points (in the list below) that I think the authors should revise (comments on lines 461-471, 502-503, 571-573).

Line 20: "Responses to bouquets"

This seems to be a misunderstanding. We refer first to the chemistry of the bouquets, and in the next sentence deal with the responses towards these bouquets (lines 19-23): “Bouquets of larval host plants and most background plants, in contrast, were subtle, thus potentially complicating host identification. However, despite being subtle, antennal responses and brain activation patterns evoked by the smell of larval host plants were clearly different from those evoked by other plants.”

Line 77: the sentence starting with "However…" should be directly after the previous one for better flow of the logic.

Done.

Line 263: "…mating status (Table 2), at least among the 23 glomeruli imaged in this study."

Done.

Line 264: … activated only one glomerulus out of the 23 imaged in the antennal lobe…."

Done.

Regarding Figure 3C: which plant headspace is used here? This needs clarification. I assume is a single species headspace -it wouldn't be appropriate to mix headspaces (the captions says "plant headspaces").

We did not mix headspaces but identified for this analysis the maximum headspace-evoked response for each glomerulus, no matter which headspace did evoke the maximum response. However, Datura flower evoked the maximum response in 69% of the cases, and Agave flower in 17%.

We added a sentence (underlined) to clarify this.

“Graph depicts for each glomerulus (color code as in A) the average maximum responses (bars) and one standard deviation (whiskers) of 10 virgin and 10 mated females after stimulation with plant headspaces. In 69% of 460 cases (20 maximum values in 23 glomeruli), Datura flower was the headspace eliciting the maximum response, and in 17% it was Agave flower.”

Line 345: between the bouquets of host plant vegetation and surrounding…"

We rephrased the sentence for clarification (line 343-346).

“Female moths, both virgin and mated, therefore seem to be equipped with the sensory capability to distinguish not only strong and complex scents emitted by nectar sources but also bouquets of host plants and surrounding background vegetation.”

Line 351: "..in typical hawkmoth-pollinated flowers…"

Done.

Line 359: "…. Is at least as sensitive to promising floral blend…"

Done.

Line 367: "…with the duration a female shows feeding behavior (i.e proboscis contact time with a scented filter paper flower, Pearson correlation…., Bisch-Knaden et al. 2018). In contrast…"

We rephrased the sentence (line 367-371).

“EAD responses evoked by these 31 shared odors belonging to seven chemical classes are indeed positively correlated with the duration a female moth shows feeding behavior when encountering the same odors (EAD amplitude versus duration of proboscis contacts with a scented filter paper; Pearson correlation coefficient r=0.41, p=0.023).”

Line 371: "… i.e. a behavior related to oviposition….."

Done.

Line 381: "In addition, the antenna might harbor narrowly tuned olfactory receptor neurons strongly responding…"

Done.

Line 414: "… the active GC-peaks overlapped between…"

We rephrased the sentence for clarification (line 410-412).

“In the case of Proboscidea headspace, none of its EAD-active compounds found in our experiments was identified in the former study, and vice versa.”

Line 424: "…influence both the composition and the quantity…"

Done.

Line 444: "… presynaptic level, i.e. at the level…"

Done.

Line 450: "…. not elicit activity in the 23 imaged antennal lobe glomeruli."

Done.

Lines 461-471: I think the explanation for the different studies misses the fact that while single components might not evoke broad activity at the AL lobe level, blends/mixtures might do so due to emergent properties of AL circuitry. The current study uses blends as stimuli for imaging of AL activity, which might explain that the authors found broad activation across the array of imaged glomeruli. So the two studies not only used different techniques, each with its own advantages, but seek to answer different questions. Indeed, in their previous publication (Bisch-Knaden et al. 2018), they used monomolecular odorants and for the most part each odorant activates a few glomeruli (at least medium to strong, Figure 2D), including the esters. In line 469 the authors say that ethyl sorbate and benzyl alcohol evoke responses in the AL in the previous study; the responses to ethyl sorbate are small and limited mostly to glomeruli 12 and a few others; similarly, the responses to benzyl alcohol were not very strong and involved about 5 glomeruli. In their previous study, monoterpenes seem to evoke the strongest responses and more widespread (i.e. involving more glomeruli). The way the ms refers to these results gives the reader the impression that the single monomolecular odorants evoke broad responses, comparable to the blend-evoked responses, which I don't think is correct. I suggest that the authors modify this paragraph accordingly.

We added a sentence to mention this further difference between our study and Riffell’s studies (stimulation with a puff of headspace versus a GC-separated stimulus). Line 476-479.

“In addition, we used a puff of the floral headspace as stimulus, i.e. the antennal lobe was activated by the full floral blend, whereas in previous studies a GC-coupled stimulus was applied, i.e. the antennal lobe was activated by temporally separated single compounds present in the floral blend.”

We did not want to raise the impression that single monomolecular odorants in our previous study evoked broad responses comparable to the blend-evoked responses in the present study.

We changed the sentence to clarify this.

“However, compounds that were identified to activate neurons in the male antennal lobe, like ethyl sorbate (Agave) and benzyl alcohol (Datura), were also EAD-active in our present study (Figure 2C) and evoked responses in some glomeruli of the female antennal lobe in a previous imaging study (Bisch-Knaden et al., 2018).”

Line 500: "… to evoke a response in sensilla targeting mostly (but not only) the two female specific glomeruli" (Shields & Hildebrand shows that while most sensilla dye-filled target the LFGs, some target a few other glomeruli).

Shields and Hildebrand cut 5-10 neighboring trichoid sensilla and bathed the respective annulus for 2-3 days in a dye solution. It is not clear if the few not-LFG glomeruli that were stained are really targeted by trichoid sensilla or by other sensillum types that had been accidentally damaged and stained in the same dye pool. Back-fillings from individual trichoid sensilla can answer this question.

We changed the sentence:

“These odors failed to evoke a response in sensilla targeting mainly the two female-specific glomeruli {Shields, 2000}.”

Lines 502-503: I still think that the authors do not have sufficient arguments for the statement at it is in this sentence. This is because: 1) the authors do not find response to vegetation in the LFGs, but the imaging technique, as the authors state, reveals activity from AL afferent mostly, not AL outputs. Although a cultivated plant, King et al. (2000) and Reisenman et al. (2009) reported conspicuous responses to tomato leaves in one of these glomeruli; 2) it is possible that the LFGs act in concert with other glomeruli to guide oviposition behavior (concerted responses not revealed by imaging of afferents might have important downstream effects); 3) males with induced LFGs fly more towards host plants (because these are the most prominent female specific glomeruli, this suggest that these glomeruli process some odorants which directly or indirectly signal an oviposition site. I thus suggest for this line something like this: "In spite of this, it is still possible that these female-specific glomeruli act in concert with other glomeruli to guide the female-specific behavior of identifying an oviposition site, a hypothesis that need further investigation."

We rephrased the paragraph (line 506-519)

“Two enlarged, female-specific glomeruli that are located at the entrance of the antennal nerve into the female antennal lobe — at the same position as the sex pheromone-processing macroglomerular complex in males {Matsumoto, 1981;Rossler, 1998} — seem predisposed to be involved in oviposition choice. This hypothesis is supported by the fact that output neurons targeting both glomeruli respond to headspace of tomato leaves, another host plant for M. sexta {King, 2000;Reisenman, 2009}. On the other hand, the two host plant bouquets tested in our imaging experiments did not activate these glomeruli (glomeruli 1 and 2, Table 2), confirming results of a study using vegetative headspace from the hosts Datura, Nicotiana, and tomato. These scents failed to evoke a response in sensilla targeting mainly the two female-specific glomeruli {Shields, 2000}. Therefore, the question if these glomeruli might be involved in identifying an oviposition site is still open.”

Line 520: "However, host plants activate only one of these glomeruli, …… activated additional glomeruli. While it is possible that host plants activate glomeruli not imaged in this study, the resulting neural representation…"

We rephrased the sentence (line 537-540).

“Even if non-host plants as well as host plants would activate more glomeruli in areas of the antennal lobe that were inaccessible in our imaging study, the resulting neural representation of non-host plants in the antennal lobe of mated females would remain different from the pattern evoked by host plants.”

Line 550: here add Goyret 2010 (J Exp Biol, Look and touch: multimodal sensory control of flower inspection movements in the nocturnal hawkmoth Manduca sexta).

In this paragraph, we discuss the females’ choice of oviposition sites. A citation about cues used during flower inspection movements is not fitting here in our opinion.

Line 552: "…when searching for oviposition…"

Done.

Line 571-573: I still think that the case of D. wrightii is particularly interesting and the fact that the plant has a faint vegetative scent but powerful floral odor has significance. It is entirely possible and supported by previous findings that female might simply use the floral odors for long distance olfactory attraction, as these are fragrant and abundant (Raguso et al. 2003, and evoke strong responses). Once in the vicinity or closer, females might use vegetative odors to decided whether or not oviposit on leaf tissues (in addition to feeding on nectar, as it is known that females mix feeding and oviposition bouts). Therefore, the two processes, oviposition and feeding, guided by weak and strong odors but at different timescales, might entangled with each other in the M. sexta-D. wrightii system. Also, it is commonly observed that M. sexta moths mix oviposition and feeding bouts on this plant, and it is reported that flowers increased oviposition both in the lab and in the field (Reisenman et al. 2010). I suggest the authors modify the sentence starting with "Hence, M. sexta…" to reflect this fact.

We fully agree with the reviewer’s statements about the long-distance attraction using flower odors, and entangled feeding and oviposition behaviors in the case of Datura. However, non-flowering plants receive eggs, and we were therefore interested in a comparison between the odors of Datura leaves and Proboscidea. To clarify this, we added a sentence at the beginning of the paragraph (line 559-564).

M. sexta females intersperse feeding and oviposition bouts when visiting a flowering Datura {Raguso, 2003}, and lay more eggs on flowering than on non-flowering plants {Reisenman, 2010}. However, Datura foliage alone attracts egg-laying females in the field (personal observations) and in the lab {Spaethe, 2013;Nataraj, 2021}. We therefore tested leaves of Datura separately and compared their headspace with that of Proboscidea, the only other host plant in our study area.”

Line 583: "… in contrast to the weak but specific activation of single glomeruli (among those imaged) by host plants of M. sexta."

Done.

Reviewer #2 (Recommendations for the authors):

The authors responded to the questions I raised, and then I think that this work deserves publication in the journal. However, I still have two suggestions:

(1) Since Datura wrightii is a unique and important plant in this study system, which is both the nectar source and host plant of this moth species, I'm sticking to my opinion that the representative traces of GC-EAD and in vivo calcium imaging recordings of headspace stimulations of the flowers and foliage of Datura wrightii should be added in Figure 2A and Figure 3B although these data were reflected in other figures and source data. For floral and foliage odors of Datura could attract females for oviposition, a comparison of responses of antennae or antennal lobes between floral and foliage of the same plant would be significant.

Based on the reviewer’s suggestion, we included representative imaging results for Datura flower and Datura foliage, both for a virgin and a mated female (Figure 3B), to make a comparison of antennal lobe responses towards the flower and the leaves of Datura possible.

(2) The EAD activities and calcium imaging activities are the core contents of this study, it is better to analyze their linkage. As the dye used in vivo calcium imaging experiments is Calcium Green-1 AM, the information reflected in the calcium imaging activity may be mainly the input of the olfactory sensory neurons from the antennae. Therefore, there should be positive relationships between EAD activities and calcium imaging activities in theory. I am very curious about the following questions: based on the authors' previous research (Bisch-Knaden et al., 2018), which compound(s) could elicit activities of the only glomerulus (Glomerulus 4) activated by Datura foliage? Were these compounds present in the headspace blend of Datura foliage? If so, how about EAD responses or even behavioral responses to these compounds? The same questions with Glomerulus 15 and 12 for Proboscidea. These could be discussed in the Discussion section.

We agree that the link between peripheral and central responses is interesting and already discussed this in our original manuscript. Based on the reviewer’s suggestion, we now use Datura foliage and its two main active components to illustrate the possible presence of non-linear processing in the moth’s lobe (underlined in the paragraph below, line 425-465), and removed the example we had given in the original manuscript. The observed mixture interactions already at the input level of the antennal lobe preclude simple correlations between EAD and imaging results.

“Bath application of a fluorescent calcium-sensor allows monitoring of odor-induced neural activity in the brain. Each neuron type in the treated brain region might take up the marker molecules. However, as each glomerulus in the antennal lobe receives input from 4000-5000 olfactory sensory neurons {Oland, 1988}, and is targeted by only four to five projection, i.e. output neurons {Homberg, 1988}, odor-evoked activation patterns in calcium imaging experiments can be assumed to reflect mainly the activity of input neurons. Additionally, about 360 local interneurons per antennal lobe {Homberg, 1988} with inhibitory and/or excitatory functions {Reisenman, 2011} might synapse back onto the sensory neurons, thus modulating their activity and accordingly the observed calcium signal. Although most of these interneurons arborize in many, if not all glomeruli, some interneurons have a more restricted innervation pattern and connect only a few glomeruli {Christensen, 1993}. This type of interneuron seems predisposed to play a role in the coding of complex odor blends released by plants. Interestingly, patchy interneurons are present mainly in female M. sexta {Matsumoto, 1981}. In the vinegar fly Drosophila melanogaster, patchy interneurons are responsible for non-linear processing of binary odor mixtures {Mohamed, 2019}. For some glomeruli in D. melanogaster, this modulation occurred already at the presynaptic level, i.e. at the level we monitored in our calcium imaging experiments. To estimate if non-linear interactions might occur in the antennal lobe of M. sexta, we compared headspace-evoked activation patterns with activation patterns evoked by EAD-active, single compounds that were present in the respective headspace. From the bouquet of Datura foliage, for example, (Z)-3-hexenyl acetate elicited the strongest antennal response (Figure 2C) and activates mainly four glomeruli (glomeruli 6, 13, 16, and 12) when tested on its own {Bisch-Knaden et al. 2018}. After stimulation with the complex headspace, however, none of these glomeruli was responding (Figure 3E, Table 2). The second best antennal activator in the headspace of Datura foliage, geraniol, activates mainly three glomeruli (glomeruli 6, 4, and 5, {Bisch-Knaden, 2018}). Of these, glomerulus 4 was the only one responding towards stimulation with the headspace (Figure 3E, Table 2). We thus conclude that there are indications of local inhibition, as we otherwise would have expected to observe more activated glomeruli after stimulation with the complex headspace. A similar inhibition of glomeruli in mixtures of odors was reported in a calcium imaging study in honeybees, where the inhibitory effect was stronger in ternary than in binary mixtures {Joerges, 1997}. As the plant bouquets tested in our study contained up to 20 EAD-active components, and as local interneurons in M. sexta, like in most insects, are mainly inhibitory {Christensen, 1993}, the observed inhibitory mixture interactions after stimulation with complex blends seem plausible.”

In the case of Proboscidea, β-ocimene was the best antennal activator. When tested on its own, β-ocimene activates mainly glomeruli 6, 12, and 17 {Bisch-Knaden et al. 2018}. However, headspace of Proboscidea activated only glomerulus 15 in virgin females (in our previous study we investigated only virgin females), which again indicates mixture interactions already at the input level of the antennal lobe.

https://doi.org/10.7554/eLife.77429.sa2

Article and author information

Author details

  1. Sonja Bisch-Knaden

    Max-Planck-Institute for Chemical Ecology, Department of Evolutionary Neuroethology, Jena, Germany
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Supervision, Validation, Visualization, Writing - original draft, Writing - review and editing
    For correspondence
    sbisch-knaden@ice.mpg.de
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3424-8376
  2. Michelle A Rafter

    CSIRO Health and Biosecurity, Dutton Park Queensland, Australia
    Contribution
    Formal analysis, Funding acquisition, Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5180-0062
  3. Markus Knaden

    Max-Planck-Institute for Chemical Ecology, Department of Evolutionary Neuroethology, Jena, Germany
    Contribution
    Conceptualization, Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6710-1071
  4. Bill S Hansson

    Max-Planck-Institute for Chemical Ecology, Department of Evolutionary Neuroethology, Jena, Germany
    Contribution
    Conceptualization, Funding acquisition, Supervision, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4811-1223

Funding

No external funding was received for this work.

Acknowledgements

We thank Brett C Blum and Mark E Heitlinger for kindly hosting us at the Santa Rita Experimental Range, Sascha Bucks for breeding M. sexta, Kerstin Weniger for help with chemical analyses, Mohammed Khallaf Ali for introducing us to XCMS, and Daniel Veit for building equipment for mobile headspace collection. This work was funded by Max-Planck Society (all authors) and CSIRO Health and Biosecurity Acorn Grant (MAR).

Senior Editor

  1. Meredith C Schuman, University of Zurich, Switzerland

Reviewing Editor

  1. Marcel Dicke, Wageningen University, Netherlands

Reviewer

  1. Sylvia Anton, INRAE, France

Publication history

  1. Received: January 28, 2022
  2. Preprint posted: February 10, 2022 (view preprint)
  3. Accepted: May 9, 2022
  4. Version of Record published: May 27, 2022 (version 1)

Copyright

© 2022, Bisch-Knaden et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 396
    Page views
  • 117
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Sonja Bisch-Knaden
  2. Michelle A Rafter
  3. Markus Knaden
  4. Bill S Hansson
(2022)
Unique neural coding of crucial versus irrelevant plant odors in a hawkmoth
eLife 11:e77429.
https://doi.org/10.7554/eLife.77429
  1. Further reading

Further reading

    1. Ecology
    Tom WN Walker et al.
    Research Article Updated

    Climate warming is releasing carbon from soils around the world, constituting a positive climate feedback. Warming is also causing species to expand their ranges into new ecosystems. Yet, in most ecosystems, whether range expanding species will amplify or buffer expected soil carbon loss is unknown. Here, we used two whole-community transplant experiments and a follow-up glasshouse experiment to determine whether the establishment of herbaceous lowland plants in alpine ecosystems influences soil carbon content under warming. We found that warming (transplantation to low elevation) led to a negligible decrease in alpine soil carbon content, but its effects became significant and 52% ± 31% (mean ± 95% confidence intervals) larger after lowland plants were introduced at low density into the ecosystem. We present evidence that decreases in soil carbon content likely occurred via lowland plants increasing rates of root exudation, soil microbial respiration, and CO2 release under warming. Our findings suggest that warming-induced range expansions of herbaceous plants have the potential to alter climate feedbacks from this system, and that plant range expansions among herbaceous communities may be an overlooked mediator of warming effects on carbon dynamics.

    1. Ecology
    2. Epidemiology and Global Health
    Feifei Zhang et al.
    Research Article

    Background: The variation in the pathogen type as well as the spatial heterogeneity of predictors make the generality of any associations with pathogen discovery debatable. Our previous work confirmed that the association of a group of predictors differed across different types of RNA viruses, yet there have been no previous comparisons of the specific predictors for RNA virus discovery in different regions. The aim of the current study was to close the gap by investigating whether predictors of discovery rates within three regions-the United States, China, and Africa-differ from one another and from those at the global level.

    Methods: Based on a comprehensive list of human-infective RNA viruses, we collated published data on first discovery of each species in each region. We used a Poisson boosted regression tree (BRT) model to examine the relationship between virus discovery and 33 predictors representing climate, socio-economics, land use, and biodiversity across each region separately. The discovery probability in three regions in 2010-2019 was mapped using the fitted models and historical predictors.

    Results: The numbers of human-infective virus species discovered in the United States, China, and Africa up to 2019 were 95, 80 and 107 respectively, with China lagging behind the other two regions. In each region, discoveries were clustered in hotspots. BRT modelling suggested that in all three regions RNA virus discovery was better predicted by land use and socio-economic variables than climatic variables and biodiversity, though the relative importance of these predictors varied by region. Map of virus discovery probability in 2010-2019 indicated several new hotspots outside historical high-risk areas. Most new virus species since 2010 in each region (6/6 in the United States, 19/19 in China, 12/19 in Africa) were discovered in high-risk areas as predicted by our model.

    Conclusions: The drivers of spatiotemporal variation in virus discovery rates vary in different regions of the world. Within regions virus discovery is driven mainly by land-use and socio-economic variables; climate and biodiversity variables are consistently less important predictors than at a global scale. Potential new discovery hotspots in 2010-2019 are identified. Results from the study could guide active surveillance for new human-infective viruses in local high-risk areas.

    Funding: FFZ is funded by the Darwin Trust of Edinburgh (https://darwintrust.bio.ed.ac.uk/). MEJW has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No. 874735 (VEO) (https://www.veo-europe.eu/).