1. Neuroscience
  2. Plant Biology
Download icon

Hawkmoths evaluate scenting flowers with the tip of their proboscis

Research Advance
  • Cited 11
  • Views 2,187
  • Annotations
Cite this article as: eLife 2016;5:e15039 doi: 10.7554/eLife.15039

Abstract

Pollination by insects is essential to many ecosystems. Previously, we have shown that floral scent is important to mediate pollen transfer between plants (Kessler et al., 2015). Yet, the mechanisms by which pollinators evaluate volatiles of single flowers remained unclear. Here, Nicotiana attenuata plants, in which floral volatiles have been genetically silenced and its hawkmoth pollinator, Manduca sexta, were used in semi-natural tent and wind-tunnel assays to explore the function of floral scent. We found that floral scent functions to increase the fitness of individual flowers not only by increasing detectability but also by enhancing the pollinator's foraging efforts. Combining proboscis choice tests with neurophysiological, anatomical and molecular analyses we show that this effect is governed by newly discovered olfactory neurons on the tip of the moth's proboscis. With the tip of their tongue, pollinators assess the advertisement of individual flowers, an ability essential for maintaining this important ecosystem service.

https://doi.org/10.7554/eLife.15039.001

Introduction

Floral scent has been associated with insect pollination since the 18th century (Sprengler, 1793); however, the complex functions of floral volatiles have been only recently investigated in more detail, due to the availability of new molecular and analytical techniques (Raguso, 2008). Floral scent not only attracts pollinators (Klahre et al., 2011), but also manipulates them through chemical mimicry (Stökl et al., 2010) and repels herbivores (Junker and Blüthgen, 2010), altogether increasing plant fitness (Kessler et al., 2008). However, research studying the function of floral scent has been divided along two themes with little cross-fertilization: 1) studies examining the fitness effects of floral scent without the causal behavioral responses of pollinators (Kessler et al., 2008), or 2) studies examining the sensory physiology of pollinators, neglecting the ecological consequences for the plant (Raguso, 2008). Here, we meld these approaches and show that floral scent increases the fitness of individual flowers not only by increasing their detectability (Raguso and Willis, 2002), but also by enhancing the pollinator’s foraging motivation, and demonstrate that this is mediated by olfactory receptors on the tip of the moth’s proboscis which detect floral scent.

The hawkmoth Manduca sexta (Figure 1A, Video 1) is a major pollinator of the wild tobacco Nicotiana attenuata in the Great Basin Desert (USA) (Kessler et al., 2008; 2010; 2015). N. attenuata emits a relatively simple floral scent dominated by a single compound: benzyl acetone (BA) (Euler and Baldwin, 1996). In spite of this simplicity, producing BA might come at considerable metabolic but especially ecological cost, as BA might attract not only nectar thieves and florivores, but also female hawkmoths in search of oviposition sites (Baldwin et al., 1997; Kessler et al., 2010). Plants might therefore reduce the amount of floral volatiles released as much as possible without losing their pollination services. Field experiments using plants in which the emissions of BA had been silenced by RNAi of the biosynthetic gene NaChal1 (CHAL) have shown that BA is required to maximize pollination success (Kessler et al., 2008; 2015). It was suggested that lacking this scent made plants in nature 'invisible' to hawkmoth pollinators. However, the precise mechanisms by which odors of single flowers influence pollinator behavior and thereby plant fitness have rarely been examined in the direct interaction between plant and pollinator (Klahre et al., 2011; Riffell et al., 2008). Hence, how floral scent emitted by individual flowers functions in this mutualistic interaction remained unknown, particularly as it is unclear how pollinators detect single volatile compounds within complex natural environments (Hansson and Stensmyr, 2011; Riffell et al., 2014).

Figure 1 with 1 supplement see all
Even though M. sexta visited the same number of benzyl acetone (BA)-emitting (EV) and non-emitting flowers (CHAL), BA-emitting flowers received superior pollination services, increasing seed production.

(A) M. sexta feeding from N. attenuata flowers. (B) Number of EV and CHAL flowers visited per moth on each foraging flight when 10 randomly placed plants per line were presented in a two-choice, free-flight tent assay (Wilcoxon signed rank test). (C) Seeds matured per antherectomized flower after visitations by a moth experimentally loaded with pollen (Wilcoxon rank sum test). Extreme values are shown as numbers.

https://doi.org/10.7554/eLife.15039.002
Video 1
M. sexta foraging on EV flowers emitting BA in a free-flight tent.
https://doi.org/10.7554/eLife.15039.004

Results and discussion

We investigated the function of floral scent in the context of individual flower-moth interactions, by offering individual male moths the choice between BA-emitting flowers (i.e. empty-vector transformed flowers (EV)) and non-emitting flowers (i.e. CHAL) in a free flight tent (24 m × 8 m × 4 m, 10 CHAL and 10 EV plants, spaced 50 cm apart). The flight tracking revealed that moths chose to visit the same number of emitting and non-emitting flowers in a random sequence (Figure 1B, Figure 1—figure supplement 1A, probability of changing between EV or CHAL flowers during consecutive visits: 0.47). In a second bioassay conducted in a wind tunnel (2.4 m × 0.9 m × 0.9 m, moonlight [0.5 lux of sunlight spectrum]), we presented plants with either emitting or non-emitting flowers to individual moths and analyzed their flight patterns, approaches and flower contacts. In none of these analyses, we found any difference between plants with emitting and non-emitting flowers (Figure 1—figure supplement 1B,C). These results suggest that visual cues and general vegetative plant odors provided sufficient information for the moths to locate flowers, consistent with previous work using artificial flowers (Raguso and Willis, 2002) and clearly showing that non-scenting flowers are not 'invisible' to moths.

If non-scenting flowers are found by moths, why is plant fitness reduced? Does BA emission change the pollination probability? To test this, we loaded the moth’s proboscis with a standardized number of pollen grains using a fine brush. When such pollen-enhanced moths were allowed to forage freely on antherectomized N. attenuata flowers, seeds produced per flower of EV and CHAL plants differed significantly (Figure 1C). Scentless flowers matured very few seeds, reflecting the inferior pollination services provided by the moths despite a similar number of visits. This result highlights the importance of BA emission for the fitness of individual flowers and confirms the results of previous studies, which investigated the effects of BA emission on plant fitness at a population level (Kessler et al., 2008; 2015). But if the flowers were equally detectable by the moths, what behavioral mechanism was responsible for the plant fitness consequences?

To analyze the effect of BA emission on moth behavior in greater detail, we quantified the time invested by a moth at individual flowers in a wind tunnel assay. The moths spent significantly more time at emitting than at non-emitting flowers (Figure 2A) particularly while trying to insert their proboscis, so even before tasting the floral nectar (Figure 2—figure supplement 1A). However, having successfully inserted their proboscis, the time of nectar uptake was similar between them (Figure 2—figure supplement 1B). This suggests that BA emission increased the motivation of moths to forage when individual flowers were evaluated at a close range, possibly because BA emissions, are closely linked to the physiological state and thereby also to the potential nectar amount of a flower (Bhattacharya and Baldwin, 2012; Yon et al., 2015; Kessler et al., 2015). By increasing the probing time in BA-emitting flowers, moths increased their success rate at their first as well as at consecutive flower visits and, therefore, collected more nectar per flower visit in tent (Figure 2B, Videos 1 and 2, Figure 2—figure supplement 1C) and wind tunnel assays (Figure 2—figure supplement 1D). These results agree with a study using different Petunia lines which found that although flower scent aided navigation, increased nectaring was the most consistent effect of floral scent (Klahre et al., 2011).

Figure 2 with 1 supplement see all
Moths spent more time and removed more nectar from BA-emitting (EV) flowers than from scentless (CHAL) flowers.

(A) Time spent by moths at single flowers (Wilcoxon rank sum test) in a wind tunnel assay. (B) Nectar remaining in flowers after moths attempted to feed in a two choice tent assay (Wilcoxon signed rank test).

https://doi.org/10.7554/eLife.15039.005
Video 2
M. sexta attempting to forage on Chal flowers not emitting BA in a free-flight tent.
https://doi.org/10.7554/eLife.15039.007

The large fitness consequences of floral volatiles for both moth and plant beg the question: how do moths evaluate the headspace of individual flowers? The wide spread of the antennae and their distance from the flower resulting from the moth’s long proboscis which is fully extended during nectaring suggests that the olfactory spatial resolution of the antennae might be too low to resolve individual flowers in an inflorescence or even between neighboring plants (Willis et al., 2013). Hence, we inferred that the moth’s proboscis might play a role in flower perception (Goyret and Raguso, 2006). Using reverse transcription PCR, we qualified the accumulation of transcripts of olfactory genes in the proboscis of M. sexta. Similar to the mosquito Aedes aegypti, but contrasting with predictions for nectar feeding insects (Jung et al., 2015), we found that the olfactory co-receptor Orco was expressed in the tip region of the proboscis along with the ionotropic co-receptor, IR25a (Figure 3A). Notably, Orco was only expressed in the first centimeter of the proboscis whereas the ionotropic co-receptor, IR8a, was only found in upper sections. This heterogeneous distribution of olfactory genes is consistent with the idea that the moth proboscis plays a more complex role in chemoreception than previously thought (Reiter et al., 2015). Screening the proboscis tip by scanning electron microscopy (Figure 3B, Figure 3—figure supplement 1A,B) we found a sensillum type that was not previously described for M. sexta (Reiter et al., 2015). This sensillum resembled the known sensillum styloconicum, but instead of a single tip pore, had a multiporous cone (Figure 3B4). Similar sensillum types have been described in other lepidopteran species (Faucheux, 2013), but their function remained unknown, although the presence of odorant-binding proteins suggested a role in olfaction (Nagnan-Le Meillour et al., 2000). We used an antibody raised against Orco (Nolte et al., unpublished), and found a single Orco- positive cell only in the first multiporous sensilla styloconica (mSt) at the tip of the proboscis (Figure 3C, Figure 3—figure supplement 1).

Figure 3 with 1 supplement see all
The M. sexta proboscis harbors sensilla, which house sensory neurons expressing olfactory genes.

(A) Reverse transcription PCR using either the proboscis shaft (S), the first centimeter of the proboscis tip (T) or a water control (C) with primers for the three major olfactory co-receptors (IR8a, IR25a, Orco). The transcripts of the ribosomal gene RL131 and water were included as positive and negative control. (B1) Scanning electron microscopy images of the M. sexta proboscis tip show three types of potential chemosensory sensilla: sensilla styloconica (B2) with a uniporous (uSt (B3)) and multiporous (mSt (B4)) cone and uniporous sensilla basiconica (uBa (B5, Figure 3—figure supplement 1B)) as well as aporouse sensilla chaetica (aCh, Figure 3—figure supplement 1A). Asterisks mark tip pore, arrowheads indicate side pores. Neuronal labeling using anti-bodies against horseradish peroxidase (C1) and against Orco(C2) indicate three neurons close to the first mSt sensillum (arrows), of which one expresses the olfactory co-receptor Orco (C3, Figure 3—figure supplement 1).

https://doi.org/10.7554/eLife.15039.008

Given the presence of potential olfactory sensilla on the proboscis of M. sexta, we wondered whether neurons housed in these sensilla play a role in the detection of BA and could thus help explain the pollination differences of EV and CHAL flowers. We performed single sensillum recordings and tested the response of neurons present in all sensillum types occurring at the tip of the proboscis to an air puff of BA at an ecologically relevant concentration (0.1 mM Kessler and Baldwin, 2007). Only neurons in the Orco-positive sensillum reacted to this compound (Figure 4B, Figure 4—figure supplement 1A). In a further test with 41 other ecologically relevant odorants, the first mSt was found to be more sensitive to BA and the structurally related benzylacetate (Figure 4C, Figure 4C—source data 1). Though these results show that neurons in the proboscis tip of M. sexta can detect volatile BA, it remained unclear whether the moth would also respond behaviorally to this compound based only on the input from neurons of the proboscis sensilla.

Figure 4 with 1 supplement see all
Olfactory sensory neurons housed in proboscis sensilla respond to BA and are sufficient for flower evaluation.

(A) Single sensillum recordings from the first multiporous sensilla styloconica. Upper trace depicts a characteristic response to the water control; lower trace shows a response to BA from the neuron in the same sensillum. Red bar indicates time of stimulus. (B) Boxplot shows ∆ spikes per second recorded from the first mSt when stimulating with water control or BA (0.1 mM) for 0.5 s. Neurons responded with a significantly higher spike rate to BA than to the water control (Wilcoxon signed-rank test). (C) Response profile of the first mSt to 42 different odorants. Black bars indicate S.E.M. Names and spike rate of each odorant can be found in Figure 4C—source data 1. (D) Behavioral assay to test the response to either humidified air with BA (0.1 mM) or humidified air only. Exhaust excludes antennal olfactory input. (E) Moth inserted their proboscis significantly longer into the arm in which BA was present (Wilcoxon signed-rank test). (F) Moths chose equally often between Y-tube arms containing BA-scented air or solvent control at the first approach (Exact binominal test).

https://doi.org/10.7554/eLife.15039.010

To disentangle the proboscis input from other chemosensory organs, we devised a behavioral experiment in which the corolla tube of a flower was replaced by a Y-maze choice assay (Figure 4D). Each arm of the Y-maze was either connected to a source of humidified air or humidified air scented with BA (0.1 mM). By drawing air directly behind the entrance of the Y-maze, the experimental set-up excluded antenna-based olfaction. Hence, as soon as a moth entered the flower aperture during free hovering flight, only the proboscis experienced the air stream containing either a solvent control, or BA (Video 3). During their first and subsequent insertions, the moths chose both Y-tube arms with equal frequency (Figure 4F, Figure 4—figure supplement 1B), but inserted their proboscis for a significantly longer time into the arm containing the BA-scented air (Figure 4E), demonstrating that the moth was able to detect BA with only the proboscis. Moths seem to use the olfactory input from the proboscis not for orientation on the corolla, but rather to assess the specific quality of an individual flower, consistent with the notion that the close-range orientation of the proboscis on the flower can be informed by mechanical and visual cues (Goyret and Raguso, 2006; Sponberg et al., 2015).

Video 3
Y-maze of the proboscis choice assay, BA-scented air is provided at the right arm and the solvent control in the left arm.

Odors were removed by applying a vacuum at the entrance of the Y-maze.

https://doi.org/10.7554/eLife.15039.013

Our findings show that hawkmoths are well adapted to visit and detect volatiles of single flowers. Floral volatiles, such as BA, not only function as navigational cues (Haverkamp et al., 2016), but also inform pollinators about the identity and the physiological state of individual flowers (Bhattacharya and Baldwin, 2012; Yon et al., 2015). Only BA-emitting flowers encourage the moth to visit a flower long enough to lead to successful pollination. Our results show that floral scent is an essential chemical feature for hawkmoths to gain nectar from, and pollinate, a single flower. Interestingly, many flowers require that their pollinators acquire particular handling skills on their first visits, before the insects are able to use the flower efficiently (Laverty, 1994). This energy investment by the pollinator not only helps ensure outcrossing for the plant, but provides the insect with a more exclusive nectar source (Heinrich, 1979). In a recent study, inexperienced M. sexta were found to sometimes expend more energy on handling flowers than they gained from the nectar; if additional experience increases foraging efficiency, this would compel the moth to visit additional flowers of the same species (Haverkamp et al., 2016). Notably, the ability to smell BA with the tip of the proboscis may not only increase the motivation of M. sexta to invest energy into BA emitting flowers, but also strengthen the moth’s learning of these flowers, as the nectar reward becomes associated with the presence of BA. Such a BA-conditioned learning rate might help to ensure a positive energy balance for the moth while at the same time ensuring cross-pollination for the plant (Heinrich and Raven, 1972). However, the question to what extent the interaction of moths and plants relies on the moths’ learning ability requires additional attention in future studies.

Although both – metabolically costly and risky in terms of herbivory – thousands of plant species actively emit floral scent (Wright and Schiestl, 2009). These emissions might be a consequence of the physiological requirement for scent compounds by certain pollinator guilds when collecting nectar, even when visual cues would be sufficient to attract pollinators to a plant. The ongoing evolutionary interaction between plants and pollinators relies heavily on floral scent (Parachnowitsch et al., 2012; Schiestl and Johnson, 2013), and may explain the absence of scent-free plants in native N. attenuata populations (Kessler et al., 2015). To pollinators with the appropriate sensory system, floral scent provides a wealth of information, highlighting the importance of chemical communication in this mutualism, on which many of our crops rely (Radera et al., 2015).

Materials and methods

Plants

We used two transgenic Nicotiana attenuata Torr. (Solanaceae) lines derived from Agrobacterium tumefaciens (strain LBA 4404) transformation of wild type N. attenuata plants which were collected in a native population at the DI Ranch (Santa Clara, UT, USA) in 1988 and subsequently inbred for 22 generations (Krügel et al., 2002). Both lines have been described earlier, empty vector control plants (EV) transformed with pSOL3NC (line number A-04-266-3) (Bubner et al., 2006), as well and plants silenced by RNAi in the production of floral scent, CHAL line (N. attenuata chalcone synthase; pRESC5CHAL, line number A-07-283-5) (Kessler et al., 2008). Seeds were sterilized and germinated on Petri dishes with Gamborg’s B5 media as described in Krügel et al. (2002). Petri dishes with 30 seeds were maintained under LD (16 hr light and 8 hr dark) conditions in a growth chamber (Percival, Perry, Iowa, USA) for 10 days, and seedlings were transferred to small pots (TEKU JP 3050 104 pots, Pöppelmann, Germany) with Klasmann plug soil (Klasmann-Deilmann, Germany) in the glasshouse. After 10 days, plants were transferred to 9 cm × 9 cm pots for wind tunnel experiments or to 1 l pots for tent experiments. The glasshouse growth conditions are described in Krügel et al. (2002). For use in the wind tunnel, plants were transferred to a York Chamber (Johnston Controls, USA) with an inverted day/night cycle; daylight time was from 17–9, maintained at 25°C and night time temperatures were 22°C, with a humidity of 60–80%. Plants used in the tent experiment were cultivated in the Max Planck Institute for Chemical Ecology (MPICOE), Jena, Germany, main glasshouse. After attaining the rosette stage of growth, plants were transferred to a second glasshouse, maintained at the same conditions, located in Isserstedt, Germany, where the plants were cultivated until used for the tent experiments.

Insects

Moths used in the wind tunnel experiments were obtained from a colony maintained at the in Jena, Germany. Animals were reared as previously described (Koenig et al., 2015). Eggs were collected from female M. sexta moths, which could freely oviposit on N. attenuata plants. For the tent experiments, eggs from native M. sexta populations at the Utah field station were collected and shipped to Germany. After hatching, caterpillars were maintained on artificial diet (wind tunnel) or on Nicotiana tabacum plants (tent) at an ambient temperature of 27°C, 70% relative humidity and a light regime of 16:8 (light: dark). Fifth instar caterpillars were transferred into individual wood chambers for pupation. One week before hatching, pupae were sexed and male pupae were transferred to a flight cage with 15.5 hr daylight and 7.5 hr dim light (0.5 lux). Temperature and relative humidity were set to 25°C and 70% during day-light phase and to 20°C and 60% during the dim light phase. A transition phase of 30 min was used between phases. Animals were used for experiments 72–76 hr after hatching.

Wind tunnel

No-choice assays were performed in a Plexiglas wind tunnel (220 cm × 90 cm × 90 cm). Charcoal-filtered air was pushed through the tunnel at a speed of 0.37 m/s. Air temperature and relative humidity was adjusted to 25°C and 70%. Plants were transferred to the wind tunnel chamber at least 1 hr before the experiment; to avoid contamination, plants were cultivated in a separated compartment with an additional charcoal air filter. Directly before each trial, a single plant was placed at the front of the wind tunnel with the flower positioned 70 cm above the tunnel floor, 20 cm from the tunnel front and 45 cm from each tunnel side.

Moths were placed individually in mesh-cages (15 cm × Ø13 cm) 1 hr before the experiment and transferred to a pre-exposure chamber set to the same light and climate conditions as the wind tunnel. For each trial, a single M. sexta moth was placed on a platform 35 cm above the tunnel floor, 20 cm from the tunnel end and 45 cm from each tunnel side. After placement, every animal was given 5 min to innate wing- fanning. Animals which did not start wing-fanning (63% ) within this time frame were considered as non-responders and excluded from subsequent analyses. After take-off, the behavior of each moth was recorded for 4 min using a custom-made 3D video tracking system. The tracking system consisted of four cameras (Logitech C615, USA, infrared filter removed) recording at 30 Hz and a resolution of 800 × 600 pixels (each pixel 0.3 cm²). Using a background subtraction algorithm implemented in C, the 3D position of the moth was calculated at a rate of 10 Hz. Based on these tracking data, we analyzed the flight pattern of the moth during the last two seconds before encountering the flower in the wind tunnel using costume-written Matlab scripts (Mathworks, USA). All recorded flight tracks were cross-checked with video data and only complete recordings were used for further analyses. In order to avoid learning effects, we only considered the first flower approached by each moth.

Moth behavior at the flower was recorded at a rate of 30 Hz by a fifth camera (Logitech C615, infrared filter removed), which had been placed into the wind tunnel at a distance of about 30 cm from the flower. Recordings were automatically started by a custom-written movement detection algorithm. Flower probing times and total contact time were measured by manually analyzing the individual video files. Similar to the flight analyses, we only considered the first flower contact in the statistical analyses.

Tent

To emulate a natural environment, we conducted pollination experiments in a large tent (height, 4 m; width, 8 m; length, 24 m) with an enclosed roof to protect from rain and lateral mesh for natural airflow (Kessler et al., 2015). Plants were moved from the glasshouse to the tent which was located directly adjacent to the glasshouse. Experiments in the tent were conducted between August 27 and September 8 2014. Ten plants of each of the two lines EV and CHAL were aligned in rows at the central section of the tent. EV and CHAL plants were positioned directly next to each other, even touching inflorescences. The position of the EV and CHAL plants were changed after each moth, to minimize potential position effects. In order to use only freshly opened flowers, all open flowers were removed each morning, before experiments were conducted. Six to eight male M. sexta moths were released sequentially per night. A new moth was released only when the previous had stopped flying and every moth was only used once. Single flower visitations were observed, the genotype and time at a single flower was noted, and after the visitation, the approached flowers were removed to measure the remaining nectar. Each moth had up to 10 flower encounters and for each moth the mean nectar gain across all flower encounters was calculated.

Nectar and pollen analysis

Directly after each experimental wind tunnel trial, both the moth and the plant were removed. The remaining nectar in the flower was measured by carefully removing the flower base and removing the nectar with a pre-weighed capillary. The nectar amount was determined by reweighing the capillary and subtracting the two weights. In the tent trials, nectar volume was quantified directly using a BLAUDBRAND graduated capillary with a volume of 25 µL (Brand, Germany) by gently removing the corolla (Kessler et al., 2007).

The pollen load on the moth proboscis was determined by rinsing the proboscis three times in 1 mL of 1% Tween solution. 10 µL of 0.5% safranin (Sigma Aldrich, Germany) were added to each sample to stain the pollen outer layer. The samples were vortexed and centrifuged for 2 min at 10000 rpm. Thereafter, the supernatant was discarded without disturbing the pollen pellet and 100 µL distilled water was added to each sample. The samples were then vortexed and 10 µL were pipetted into a four-field Neubauer-counting chamber to determine the pollen number in each sample. Every sample was counted twice independently, and the mean value was used for statistical analysis.

Cross pollination experiment

To measure pollination rates in EV and CHAL plants by M. sexta in the wind tunnel, fully developed flowers were emasculated in the previous corresponding daylight morning cycle to avoid self-pollination (Kessler et al., 2008). For this, 3–4 flowers per plant of each line (EV and CHAL) were used in the wind tunnel, one plant and one moth at each time. Fresh pollen was collected in the corresponding morning from plants not being used for pollination. EV pollen which had been collected the previous night was rubbed on the hawkmoth proboscis using a fine brush prior to its release in the wind tunnel, in order to measure the pollen delivery to experimental flowers. If moths did not take flight voluntarily within 3 min after being placed in the wind tunnel they were excluded from the study. Moths, which took flight (73% ), were allowed to do so for four minutes in each wind tunnel trial. The numbers of matured capsules, as well as the seeds produced from each capsule were counted after ripening. Capsules were collected shortly before opening, approximately 14 days after the experiment, dried in a desiccator, and once opened, the seeds were counted in petri dishes. After each trial the pollen from each M. sexta was collected by washing the proboscis to ensure that similar amounts of pollen had been placed on the proboscis. For pollen counts, the same procedure as for the pollen retrieval was used. On average, we found that 548.75 (n= 41, SEM= 88.85) pollen grains had been placed on a single proboscis. No difference was found between moths tested with EV (n= 21, mean= 569, SEM= 92.3) or CHAL (n=20, mean= 500, SEM= 159.3) plants (Student’s t-test, p= 0.71).

Scanning electron microscopy

M. sexta proboscises were cut 1 cm from the tip and fixed in 4% glutaraldehyde at 4°C overnight. Proboscises were then dehydrated in an ascending ethanol series (70%, 80%, 90%, 96%, 3x 100% ethanol, 10 min each), critical point dried (BAL-TEC CPD 030, Bal-Tec Union Ltd., Liechtenstein), mounted on aluminium stubs with conductive carbon cement (Agar Scientific, UK) and sputter coated with gold on a BAL-TEC SCD005 (Bal-Tec, Liechtenstein). Specimens were examined in a LEO 1530 Gemini scanning electron microscope (Zeiss, Germany) set at 8 kV and 11 to 15 mm working distance.

Immunohistochemistry and confocal laser scanning microscopy

The tip region of 20 M. sexta proboscises were carefully dissected into three small parts, cutting behind the first, before the fourth and behind the fifth sensillum styloconica. Directly after dissection, the proboscis parts were fixed in 4% paraformaldehyde (ROTH, Germany) in 1 M NaHCO3 (Sigma Aldrich, pH 9.5) overnight at 4°C. Subsequently, the samples were washed six times for 30 min in 1× phosphate-buffered saline containing 0.1% Trition X (PBS-T) (Sigma Aldrich, USA) and thereafter blocked for 3 hr in normal goat serum (NGS). The primary anti-body against Orco (kindly provided by Prof. Jürgen Krieger, University of Halle-Wittenberg, Germany) was applied at a 1:500 dilution in 2% NGS- PBS-T and incubated for 5 days at 4°C. Detection of the Orco antibody was performed by incubating in a goat-anti rabbit antibody linked to Alexa 488 (Invitrogen, USA) at a dilution of 1:200 in 2% NGS-PBS-T for 3 days at 4°C. In addition, we added an goat anti-horseradish peroxidase antibodies conjugated to Cy3 (Jackson Immuno Research, USA) at a dilution of 1:50 in 2% NGS-PBS-T to visualize neuronal tissue. For visualization, the samples were mounted in 50% glycerol on a microscope slide and scanned using confocal laser scanning microscopy (LSM 880, Zeiss, Germany). Alexa 488 was exited using the 488 nm line of the microscopes Argon laser, while a Helium Neon 543 laser was used to activate Cy3. Signals were detected by a spectral detector (quasar: 490–553 nm and 555–681 nm). All pictures were taken using a 20× air objective (N.A. 0.8). Scanning resolution was set to 1024 × 1024 pixel.

Total RNA isolation

Proboscises of ten male M. sexta were dissected and were cut 1 cm from the tip. Each tissue sample (tips and rest) was directly transferred to Tri-reagent (Sigma-Aldrich, USA). The samples were then homogenized with two 3 mm steel beads (Qiagen, Germany) using a TissueLyser (Qiagen, Germany) for 5 min at 50 Hz. Samples were stored at -20°C. Finally, RNA isolation was performed using TRI- Reagent (Sigma-Aldrich, USA) according to the manufacturer's instructions.

cDNA synthesis

RNA samples were treated with TurboDNAse (Ambion, USA) according to the manufacturer's instructions. DNAse was removed using Tri-reagent following the instructions of the producer. RNA was dissolved in 25 µL RNA storage solution (Ambion, USA). For cDNA synthesis 1 µg total RNA per sample was used as template for the Super Script III kit (Invitrogen, Canada).

Reverse transcription-PCR

For RT-PCR dNTPS (Thermo Fisher Scientific, Lithuania), cDNA, gene-specific primers and the Avantage 2 Polymerase mix (Clontech, Canada) were used following the manufacturer’s instructions. Primers were designed according to Koenig et al. (2015): (RL31: GGA GAG AGG AAA GGC AAA TC and CGG AAG GGG ACA TTT CTG AC; MsexIR8a: CAA CCC CGA CGC GTA TCC GTA TCC and TTA CGG CCT ATA TTC ATT TTT AGG AAA AAC GCT TAT ATA TG; MsexIR25a: GGA GTC CGT ATA GCT ATC AGA ATA ATC GAG and TCA AAA TTT AGG TTT CAA ATT AGA TAA ACC TAA ATT TCT GGA TC; MsexORCo: ATG ATG GCC AAA GTG AAA ACA CAG G and CTA TTT CAG CTG CAC CAA CAC CAT G). Reaction was done in a thermocycler (GeneAmp PCR System 9700, PE Applied Biosystems, USA) with 95°C for 1 min, followed by 35 cycles of 95°C for 30 s, 60°C (for MsexIR25a: 62°C) for 30 s and 68°C for 90 s. The final step was incubation at 68°C for 3 min. The samples were loaded on a 1.5% agarose gel.

Electrophysiology

For electrophysiological recordings, moths were placed into a 15 mL reaction tube, from which the tip had been cut; in such a way that only the proboscis would extend from the tube. Eachanimal was then mounted on a microscope slide, and the proboscis was fixed with dental wax. Next we unrolled the first centimetre of the proboscis and fixed this part upside down on a small wax pedestal, so that most of the sensilla were approachable for electrophysiological recordings. Subsequently, the preparation was positioned under a microscope (BX51W1, Olympus, Japan) and a tungsten reference electrode was inserted into the proboscis shaft. The recording electrode was then inserted into the target styloconic sensillum via a motorized, piezo-translator-equipped micromanipulator (DC-3K/PM-10, Märzhauser, Germany). A constant air stream of humidified air was applied to the preparation. For stimulus delivery either 0.1 mg BA diluted in distilled water or distilled water only was loaded onto a filter paper, inserted in a glass pipette and puffed onto the proboscis using a Syntech stimulus controller (CS- 55, Syntech, The Netherlands). For the odor screen individual compounds were diluted in hexane (10–2 v/v) and 10 µL were loaded on to a filter paper and puffed as described before. A single puff lasted for 0.5 s. The recorded signal was then amplified (UN-06, Syntech, The Netherlands), digitally converted (IDAC-4, Syntech, The Netherlands), and recorded at a rate of 2400 Hz using AutoSpike v3.2. (Syntech, The Netherlands). Traces were exported as ASCII files and manually analyzed using R. Spikes were counted 2 s before the stimulus onset and 2 s thereafter. The number of spikes before the stimulus was then subtracted from the spikes counted after the stimulus onset. The resulting number of ∆ spikes was then divided by the number of seconds analyzed. In all experiments, three day old male moths were used.

Proboscis choice

Olfactory preference of the proboscis was tested in a custom-built Y-maze (5 cm × 3 cm × 0.5 cm). Previous studies had found a 0.1 mM suspension of BA in nectar to be ecologically relevant in the interaction between M. sexta and N. attenuata (Kessler and Baldwin, 2007). Here, we tested 10 µL of 0.1 mM BA suspension in distilled water against the same amount of distilled water only. Both stimuli were pipetted onto a small filter paper discs and placed into 50 mL glass bottles. Bottles were connected to the Y-tube arms via Teflon tubing (Ø 6 mm). Charcoal-filtered air was pushed into the bottles so that the air flow at each Y-tube arm reached 0.1 L/ min. To prevent the moths’ antenna from contacting BA headspace and assure a homogenous flow, air was removed from the opening of the Y-tube at a rate of 0.2 L/ min. The movement of the moth proboscis was recorded via a video camera (Logitech C615, infrared filter removed) at 30 Hz. Videos were captured using the software package Media recorder (Noldus, The Netherlands) and subsequently viewed and manually analyzed using EthoVision (Noldus, The Netherlands). For tests, the Y-maze set-up was placed into the wind tunnel described above and moths were allowed to forage freely for 4 min. In order to attract the moths to the Y-maze, we attached the corolla of a freshly cut Nicotiana alata flower, which does not release BA (Raguso et al., 2003), onto the Y-maze opening.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
    Spatiotemporal coding of individual chemicals by the gustatory system
    1. S Reiter
    2. C Campillo Rodriguez
    3. K Sun
    4. M Stopfer
    (2015)
    The Journal of Neuroscience : The Official Journal of the Society for Neuroscience 35:12309–12321.
    https://doi.org/10.1523/JNEUROSCI.3802-14.2015
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36

Decision letter

  1. Marcel Dicke
    Reviewing Editor; Wageningen University, Netherlands

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Hawkmoths evaluate scenting flowers with the tips of their tongues" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by Marcel Dicke as the Reviewing Editor and Detlef Weigel as the Senior Editor.

The following individual involved in review of your submission has agreed to reveal their identity: Rob Raguso (peer reviewer).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

This manuscript describes an interesting study of the perception of benzyl acetone by the proboscis of naive Manduca sexta female moths. The description of the multi-pored sensillum that can respond to the floral volatile benzyl acetone is new and electrophysiological as well as behavioural assays show that the sensillum responds to benzyl acetone. Yet, the reviewers feel that some more information is needed on this new sensillum to better understand its role in the perception of the flowers. The main issues to be dealt with are the following: (a) the title where the word proboscis would be the correct word to be used, (b) the specificity of the sensillum by testing more than only the response to benzyl acetone, but also other volatiles emitted by Nicotiana attenuata. After all, N. attenuata nectar produces more volatiles than only benzyl acetone (Kessler et al. 2007 Plant J.), (c) is the sensillum also involved in sugar perception: does the sensillum also house a sugar responding dendrite? and (d) the role of benzyl acetone perception by the probiscis of naïve moths versus moths that have gained experience that do not require scented nectar; this has consequences for the discussion (Discussion section, last paragraph) on the herbivore-pollinator trade-off. In addition, the reviewers have made many valuable additional comments that will be useful to further improve the manuscript.

Reviewer #1:

This manuscript builds upon previous studies dissecting the role of floral chemistry in mediating interactions between Nicotiana attenuata plants and their insect visitors, including Manduca sexta as a co-pollinator/herbivore. This study addresses a neglected dimension of flower visitation by M. sexta, the role of the proboscis during the final stages of flower visitation, where a moth physically negotiates a nectar tube or spur and discovers nectar.

The authors ask an intriguing question: could the extended proboscis detect volatiles at the scale of an individual flower, and if so, could that assist in their handling of those flowers, and might such handling have fitness consequences for the plant?

What they find is fascinating. Indeed, the proboscis has a (surprise!) multi-pored sensillum that can respond to a floral volatile (benzyl acetone = BA). They record from it in response to BA, show that ORCO is expressed within the sensillum (as one would expect in an antenna) and show with a Y-tube assay that the presence of BA extends moth probing time in the correct arm of the "nectary". I looked for such sensilla using SEM 20 years ago, but apparently not carefully enough, as it looks superficially just like the single pored styloconic sensillum type known to respond to sugar solutions in this and other hawkmoths. Kelber, A. (2003). Sugar preferences and feeding strategies in the hawkmoth Macroglossum stellatarum. Journal of Comparative Physiology A, 189(Haverkamp et al., 2016), 661-666.

The authors use a series of flight cage, wind tunnel and y-tube assays to test for the effects of scent on different spatial scales, exploiting a line of silenced N. attenuata plants (CHAL) that do not emit floral BA, and comparing them with the empty vector (EV) control plants.

The primary result, consistent across experiments, is that scent perception by the extended proboscis is associated with extended probing time in a scented flower, and that this in turn is associated with better acquisition of nectar and better pollen transfer. One of several caveats (see below, details) of this study is that the assays were performed with flower naïve moths, so we don't yet know if these behaviors persist, or whether an experienced moth has acquired the motor skills in handling flowers that obviate the need for them to be scented. Certainly, M. sexta can learn to visit many, many kinds of natural and artificial flowers that lack scented nectar or even strong corolla scent, if they are immersed in a scent cloud. I have several comments below that address areas where the authors should clarify their wording or interpretation of their results.

1) Introduction: The first sentence is weak; ecosystem function really is about nutrient and energy flow and not about pollination success. It is just window dressing and can be cut.

2) Results and Discussion: "possibly because BA emissions, unlike visual cues, are a good predictor of nectar reward". This sentence requires clearer explanation. It refers specifically to the N. attenuata system but cites a global study by Junker and Bluthgen. Presumably both BA and visual cues are imprecise indicators of nectar if the flowers have been emptied recently? A cue emanating from the nectar itself, either nectar-specific scent (nicotine?) or relative humidity, would be the most accurate index of nectar presence, as discussed in several other papers (e.g. van Arx et al. PNAS).

3) Results and Discussion, third paragraph: precision of wording. It is generally regarded as negative for foraging animals to increase handling time in an optimal foraging sense, because it reduces profitability of a resource by increasing the denominator (reward over HT). What you mean here, I think, is that scented flowers increase persistence of probing beyond mechanosensory stimulation, and this leads to greater success in flower discovery by untrained moths. I would recommend calling this variable "probing time" instead of handling time to avoid confounding these ideas, which are very different. In fact, beyond what is shown in Figure 2, one would like to see a pair of acquisition curves for moths learning to use EV vs. those learning to use CHAL. Have the authors done this?

4) Concerning the multiporous cone sensillum found near tip of proboscis; does it also house a sugar responding dendrite? Have you tried recording from it when both BA and sucrose are present? Similarly, is there any way to cauderize or KO the tongue tip? Zinc sulfate dip treatments, as have been done with crickets responding to cuticular HCs? The problem is that the same or similar sensilla probably respond to sugar.

5) Y-maze for proboscis, blind test of whether it can orient to source of scent. Figure 4 nicely shows that moths spend more time with proboscis in scented arm (vs. humidified), but do not ask if they do same for humidified air over dry air. Also, the authors should show a similar panel for first choice, even if there is no difference. That would provide a nice parallel to the wind tunnel experiment showing no differences in flower choice but significant increase in probing time. (Oh, I see that this is shown Figure 4—figure supplement 1B – it might be more effective if moved to Figure 4).

6) Figure 4 caption, part B is written as if aqueous solutions with and without BA are being presented in the Y tube arm, but the description in the text is of presenting humidified air with vs. without BA. Please clarify here, as is more clear in the Methods.

7) Results and Discussion: the following lines are misleading: "Floral volatiles, such as BA, not only function as navigational cues (Raguso and Willis, 2002), but also inform pollinators about the state of individual flowers". That is not quite accurate.

The array assays in this paper and the one cited (Raguso and Willis 2002) clearly show that M. sexta moths do not choose individual flowers, in flight, by whether they are scented or not (p. 691, section entitled Lack of Discrimination between Natural and Paper Flowers), if those flowers are embedded in a scent cloud. That finding is consistent across nearly all published experiments with M. sexta. Raguso and Willis 2002 was not a wind tunnel test, but a set of bioassays performed in small flight cages in still air. Thus, scent was not interpreted as a navigational aid, but rather as a sign stimulus, synergizing visual display to elicit proboscis extension and flower visitation.

Instead, what the authors' data show is that BA encourages extended probing time once a moth has decided to probe a flower (chance, in their behavioral assays). They do not have an experimental result here that indicates remote (without probing) evaluation of flowers with vs. without nectar.

Once M. sexta and other hawkmoths commit to feeding in a patch of flowers, they extend the proboscis and keep it extended until they leave that patch. That observation led me and others to suspect that the extended proboscis could function as a third antenna, so to speak, providing spatial information to the brain about odor plume sources. To my knowledge, that possibility has not been examined, and it seemed unlikely to me given the moths' hierarchical choices for probing at bright objects once stimulated by scent (Raguso and Willis 2005, Goyret et al. 2007).

8) I would add that the last paragraph of the Results and Discussion is not accurate either, given the large body of work showing how floral scent impacts foraging behavior in flower naïve M. sexta that have not yet learned to associate scent with nectar, and many papers about innate preferences for scent (see Schiestl 2015 New Phytologist for a recent review). The present experiments were not done with experienced moths, so it is possible that scent-aided probing is primarily a benefit to flower-naïve moths, rather than being generally essential for them to choose and pollinate flowers.

9) Methods, subsection “Insects”: female moths.

10) Methods, subsection “Tent”: spatial resolution of scent cloud: if the EV and CHAL plants are touching, it is likely that the entire array is embedded within a scent cloud that precludes moths distinguishing between odor sources on the fly. If they repeated these experiments with plants separated by a few meters, they might get a different result. This problem is common to all array experiments performed with hawkmoths over the last decade (using Petunia, Mimulus, Hemerocallis).

11) Methods, subsection “Tent”: please clarify whether batches of moths were released individually or in batches. The latter would violate assumptions of independence, or would render each released batch as a replicate, as in Drosophila bioassays.

12) Methods, subsection “Cross pollination experiment”: Please address the extent to which adding pollen to tongue tips impacted moth performance. They don't like being handled and having their probosces extended, but it is possible that doing this during the photophase, when they are quiescent, would not impair foraging behavior during scotophase.

13) Methods, subsection “Scanning electron microscopy”: provide replicates please, for males and females.

14) Methods, subsection “Proboscis choice”: wouldn't it be more accurate to have used sucrose solutions with BA instead of BA in distilled water? The colligative properties of the sucrose (salting out) likely would change emission rates of BA from that solution.

15) Figure 1—figure supplement 1A: the diagram does not show an array (i.e. a checkerboard), but rather, two lanes of plants, color coded to indicate EV vs. CHAL. Is that true, or is the diagram a simplification? Please clarify, because the former design is more effective than the latter in terms of avoiding side of cage biases, etc.

16) Panel C does not make sense. I don't understand what you are showing on the left side (Y axis wind tunnel width, x axis wind tunnel length) – please explain more clearly what you are testing here and what you found vs. expectations. You are not experimentally varying the dimensions of the wind tunnel! The right panel simply shows a non-significant trend, with 3:2 ratio of sample sizes, and is not strongly supported.

Reviewer #2:

This study makes an interesting contribution to our understanding of an important problem. It shows that certain sensilla on the proboscis of Manduca are olfactory. A very clever behavioral paradigm is used to show that these sensilla detect a major component, benzyl acetone (BA), of the scent of wild Nicotiana attenuata. The study also shows that flowers engineered not to emit BA receive inferior pollination services, and that less nectar is removed from them. The simplest interpretation of these results taken together is that pollination and nectar removal depend on the sensilla, although this is not demonstrated directly by ablation or masking experiments.

An extensive study of the specificity of the sensilla is beyond the scope of this study, but the story would be more compelling if it were shown that the sensilla showed at least some odorant-specificity.

https://doi.org/10.7554/eLife.15039.015

Author response

This manuscript describes an interesting study of the perception of benzyl acetone by the proboscis of naive Manduca sexta female moths. The description of the multi-pored sensillum that can respond to the floral volatile benzyl acetone is new and electrophysiological as well as behavioural assays show that the sensillum responds to benzyl acetone. Yet, the reviewers feel that some more information is needed on this new sensillum to better understand its role in the perception of the flowers. The main issues to be dealt with are the following: (a) the title where the word proboscis would be the correct word to be used, (b) the specificity of the sensillum by testing more than only the response to benzyl acetone, but also other volatiles emitted by Nicotiana attenuata. After all, N. attenuata nectar produces more volatiles than only benzyl acetone (Kessler et al. 2007 Plant J.), (c) is the sensillum also involved in sugar perception: does the sensillum also house a sugar responding dendrite? and (d) the role of benzyl acetone perception by the probiscis of naïve moths versus moths that have gained experience that do not require scented nectar; this has consequences for the discussion (Discussion section, last paragraph) on the herbivore-pollinator trade-off. In addition, the reviewers have made many valuable additional comments that will be useful to further improve the manuscript.

For the title of the manuscript we would now suggest: “Hawkmoths evaluate scenting flowers with the tip of their proboscis”.

We followed the reviewers’ advice and have now tested the electrophysiological response of the multiporouse sensillum styloconica to 41 additional, ecological relevant odors to further characterize the specificity of the sensillum. We have included these results as an additional panel in Figure 4C, as a table in Figure 4C—source data 1 and in the fifth paragraph of the Results and Discussion section.

The question whether this sensillum also houses a sugar responsive neuron, is certainly of interest for the foraging ecology of M. sexta. However, we feel that it is beyond the scope of our study as we are mainly focusing here on the influence of floral volatiles detected by the moth proboscis. Furthermore, our behavioral data indicates that BA influences the decision of the moth before the moth contacts the nectar, but not necessarily during feeding (Figure 2—figure supplement 1A, B). We have now aimed to make this clearer to the reader by rephrasing the text in the third paragraph of the Results and Discussion section.

Similarly, we feel that the influence of BA on the flower learning of M. sexta is an interesting and important point, which would deserve a more thorough investigation than we can achieve in this study. Nonetheless, we would hypothesize that experience might further increase the difference between emitting and non-emitting flowers as moth that forage on emitting flowers have a higher change of receiving a nectar reward and might therefore continue to forage on emitting flowers, and not at non-emitting once. Moths are also more likely to learn a flower which provides both visual as well as olfactory information (Riffell and Alarcón, 2013), which might again increase the preference of the moth for scented flowers. This hypothesis is consistent with the results of our tent experiments in which we allowed the moth to encounter up to 10 flowers consecutively and analysed the mean nectar gain of each moth (we have now tried to clarify this in the Methods section, subsection “Tent”). We found that after these consecutive flower encounters, the difference in the nectar uptake between the emitting and the non-emitting flowers was even greater than in our wind tunnel assays, in which the moth encountered only a single flower. Moreover, we have analysed the percentage of foraging success for consecutive visits on EV and CHAL plants in the free flight tent and included this as Figure 2—figure supplement 1C. These results indicate the high success rate of M. sexta on emitting flowers is indeed maintained also during consecutive flower visits whereas the foraging success on non-emitting flowers remains relatively low. Hence we would expect that the difference between emitting and non-emitting flowers would rather be maintained or even increased (as one could assume that moths become more efficient in handling emitting flowers, after they learned to associate a nectar reward with the presence of BA). We have added a short discussion on this point in the seventh paragraph of the Results and Discussion section.

Finally, we would like to point out that our study only was conducted with male moths (we tried make this clearer in the first paragraph of the Results and Discussion section and in the Methods subsection “Tent”) and we have therefore removed the discussion on the plant-herbivore trade-off to avoid any further confusion; the interaction between oviposition and nectar foraging has been and will be an important question for future research, and you were correct in pointing out that by including it here, the waters are only muddied.

Reviewer #1:

This manuscript builds upon previous studies dissecting the role of floral chemistry in mediating interactions between Nicotiana attenuata plants and their insect visitors, including Manduca sexta as a co-pollinator/herbivore. This study addresses a neglected dimension of flower visitation by M. sexta, the role of the proboscis during the final stages of flower visitation, where a moth physically negotiates a nectar tube or spur and discovers nectar.

[…]

The primary result, consistent across experiments, is that scent perception by the extended proboscis is associated with extended probing time in a scented flower, and that this in turn is associated with better acquisition of nectar and better pollen transfer. One of several caveats (see below, details) of this study is that the assays were performed with flower naïve moths, so we don't yet know if these behaviors persist, or whether an experienced moth has acquired the motor skills in handling flowers that obviate the need for them to be scented. Certainly, M. sexta can learn to visit many, many kinds of natural and artificial flowers that lack scented nectar or even strong corolla scent, if they are immersed in a scent cloud. I have several comments below that address areas where the authors should clarify their wording or interpretation of their results.

We fully agree that flower learning is an important dimension in the interaction which deserves more attention than we can provide in this study. However, we would also like to point out that in our flight tent experiment, the moths were allowed to visit up to 10 different flowers in a row and a mean value of the remaining nectar was then calculated over these 10 flowers (Figure 2B). Therefore, if the observed effect would have only been present during the first flower encounter one would have expected that the effect would be diminished after consecutive flower visits; however, we found the opposite. The effect was even stronger after these consecutive visits than what was observed in the wind tunnel where the moth had only excess to a single flower (Figure 2—figure supplement 1D). Furthermore, we have now calculated the success rate of the moth in the tent assay over consecutive flower visits and found that both the high success rate on emitting flowers as well as the relatively low success rate on non-emitting flowers are maintained over consecutive flower encounters (Figure 2—figure supplement 1C). Thus learning did not appear to change the overall success rate of the moth. Nonetheless, we would hypothesize that the moth might still be able to learn how to handle flowers more effectively and we would expect that in the case of the scented flower this effect might be even stronger than with non-scented flower, since the moths were more likely to obtain a reward from the scented flowers and reward quantity is thought to influence the strength of the motor learning (Wright et al., 2009). Moreover, the fact that the moths can also make use of visual and olfactory information, the scent might further reinforce and increase their learning rate (Riffell and Alarcón, 2013). We have added a short discussion on this in the seventh paragraph of the Results and Discussion section.

1) Introduction: The first sentence is weak; ecosystem function really is about nutrient and energy flow and not about pollination success. It is just window dressing and can be cut.

We agree with the reviewer on this point and have modified the sentence to “Floral scent has been associated with insect pollination since the 18th century (Sprengler, 1793); however, only recently have the complex functions of floral volatiles been investigated in detail, due to the availability of new molecular and analytical techniques (Raguso, 2008).”

2) Results and Discussion: "possibly because BA emissions, unlike visual cues, are a good predictor of nectar reward". This sentence requires clearer explanation. It refers specifically to the N. attenuata system but cites a global study by Junker and Bluthgen. Presumably both BA and visual cues are imprecise indicators of nectar if the flowers have been emptied recently? A cue emanating from the nectar itself, either nectar-specific scent (nicotine?) or relative humidity, would be the most accurate index of nectar presence, as discussed in several other papers (e.g. van Arx et al. PNAS).

The reviewer is right to point out that BA emissions from the corolla are not directly linked to the nectar production and we have rephrased the sentence in the third paragraph of the Results and Discussion section to be more precise.

While BA is not a direct predictor of floral nectar it remains a good indicator of the physiological state and the metabolic activity of N. attenuata flowers (Bhattacharya and Baldwin, 2012, Yon et al., 2015; Kessler et al., 2015) and might thus help to inform moths seeking more rewarding flowers. We furthermore agree with the reviewer that other cues such as humidity and potentially nicotine might be more informative to M. sexta in selecting flowers with standing nectar, and the detection of these cues might be interesting questions for further studies.

3) Results and Discussion, third paragraph: precision of wording. It is generally regarded as negative for foraging animals to increase handling time in an optimal foraging sense, because it reduces profitability of a resource by increasing the denominator (reward over HT). What you mean here, I think, is that scented flowers increase persistence of probing beyond mechanosensory stimulation, and this leads to greater success in flower discovery by untrained moths. I would recommend calling this variable "probing time" instead of handling time to avoid confounding these ideas, which are very different. In fact, beyond what is shown in Figure 2, one would like to see a pair of acquisition curves for moths learning to use EV vs. those learning to use CHAL. Have the authors done this?

We would like to thank the reviewer for this comment. We have exchanged “handling time” with “probing time”. Learning curves have not been done, since the focus of this study was on the ecological characterization of a newly discovered sensillum working at a single flower basis rather than the differential learning capacity for scented and non-scented flowers. Nevertheless, we agree that learning and/ or performance curves of M. sexta foraging on scented and unscented flowers would be an important further step to understand the interaction between M. sexta and N. attenuata. We have now included the success rate of M. sexta after consecutive flower visits (Figure 2—figure supplement 1C).

4) Concerning the multiporous cone sensillum found near tip of proboscis; does it also house a sugar responding dendrite? Have you tried recording from it when both BA and sucrose are present? Similarly, is there any way to cauderize or KO the tongue tip? Zinc sulfate dip treatments, as have been done with crickets responding to cuticular HCs? The problem is that the same or similar sensilla probably respond to sugar.

We agree with the reviewer that cauterizing or ablating the sensilla or even the proboscis tip is problematic as these sensilla are likely to also house additional mechanosensory or gustatory neurons which might also influence the flower probing behavior of M. sexta. A double test with sugar has not been done, given that the experimental procedure to measure volatile and soluble cues are different, therefore both potential responses are difficult to measure simultaneously. Moreover, the objective of the study was more focused on olfaction than on taste. As our Y-maze assay shows, the presence of volatiles alone is sufficient to increase the moths probing time. Whether or not gustatory cues like sugar add to this effect, is interesting, but beyond the scope of this study.

5) Y-maze for proboscis, blind test of whether it can orient to source of scent. Figure 4 nicely shows that moths spend more time with proboscis in scented arm (vs. humidified), but do not ask if they do same for humidified air over dry air. Also, the authors should show a similar panel for first choice, even if there is no difference. That would provide a nice parallel to the wind tunnel experiment showing no differences in flower choice but significant increase in probing time. (Oh, I see that this is shown Figure 4—figure supplement 1B – it might be more effective if moved to Figure 4).

We would like to thank the reviewer for raising these points and agree that it might be more effective to move Figure 4—figure supplement 1B to Figure 4F, which we have now done.

We have always analyed the response towards 10 µL of a 0.1 mM BA suspension in distilled water on filter paper again 10 µL water on filter paper. Hence we have always tested the response of the proboscis in a slightly humidified environment. Given that BA is mostly present in a humid environment, both in the nectar and on the corolla (von Arx et al., 2012), we believe that this is the most likely situation in which the proboscis would encounter BA. As water was present on both arms of the Y-maze (see comment on correct figure legend below), humidity is less likely to explain the observed results of increased probing time in the presence of BA.

6) Figure 4 caption, part B is written as if aqueous solutions with and without BA are being presented in the Y tube arm, but the description in the text is of presenting humidified air with vs. without BA. Please clarify here, as is more clear in the Methods.

Legend in Figure 4B has been modified.

7) Results and Discussion: the following lines are misleading: "Floral volatiles, such as BA, not only function as navigational cues (Raguso and Willis, 2002), but also inform pollinators about the state of individual flowers". That is not quite accurate.

The array assays in this paper and the one cited (Raguso and Willis 2002) clearly show that M. sexta moths do not choose individual flowers, in flight, by whether they are scented or not (p. 691, section entitled Lack of Discrimination between Natural and Paper Flowers), if those flowers are embedded in a scent cloud. That finding is consistent across nearly all published experiments with M. sexta. Raguso and Willis 2002 was not a wind tunnel test, but a set of bioassays performed in small flight cages in still air. Thus, scent was not interpreted as a navigational aid, but rather as a sign stimulus, synergizing visual display to elicit proboscis extension and flower visitation.

Instead, what the authors' data show is that BA encourages extended probing time once a moth has decided to probe a flower (chance, in their behavioral assays). They do not have an experimental result here that indicates remote (without probing) evaluation of flowers with vs. without nectar.

Once M. sexta and other hawkmoths commit to feeding in a patch of flowers, they extend the proboscis and keep it extended until they leave that patch. That observation led me and others to suspect that the extended proboscis could function as a third antenna, so to speak, providing spatial information to the brain about odor plume sources. To my knowledge, that possibility has not been examined, and it seemed unlikely to me given the moths' hierarchical choices for probing at bright objects once stimulated by scent (Raguso and Willis 2005, Goyret et al. 2007).

This is a valid point. We have recently investigated the flight of M. sexta towards different Nicotiana flowers in the absence of meaningful visual cues in more detail, and found that in such a situation the hawkmoths are also able to navigate towards an odor source emitting the scent of N. attenuata (Haverkamp et al., 2016). We added this citation to the main text and rephrased the second part of the sentence to “…,but also inform pollinators about the identity and the physiological state of individual flowers.” and added Bhattacharya and Baldwin (2012) and (Yon et al., 2015) as an additional references in line 203.

In the sentence we point at the general assumption that flower scent, such as BA, work over distance as navigational cues (Haverkamp et al. 2016). The reviewer is of cause right to point out that flower volatiles are not directly indicating whether nectar is present in the flower or not. Nevertheless, the volatile emissions correlate with the physiological activity of the flower (Bhattacharya and Baldwin, 2012), which might still be an important cue for the moth as freshly opened, young scenting flowers are more likely to contain nectar than older non-scenting flowers.

The hypothesis of the reviewer, that the proboscis might be used as a ‘third antenna’ to obtain more spatial information on the odor source is certainly an interesting one, however just like the reviewer, we also do not think that this is very likely. In our experiments we have not observed any indication, that the proboscis would enable an enhanced flower location; on the contrary in the Y-maze, we observed no difference in the first choice of the proboscis, which indicates that the proboscis is not able to navigate by chemical cues directly (Figure 4F).

8) I would add that the last paragraph of the Results and Discussion is not accurate either, given the large body of work showing how floral scent impacts foraging behavior in flower naïve M. sexta that have not yet learned to associate scent with nectar, and many papers about innate preferences for scent (see Schiestl 2015 New Phytologist for a recent review). The present experiments were not done with experienced moths, so it is possible that scent-aided probing is primarily a benefit to flower-naïve moths, rather than being generally essential for them to choose and pollinate flowers.

We agree with the reviewer that the sentence has been phrased too strongly and we have changed this sentence and following paragraph accordingly in the seventh paragraph of the Results and Discussion section. As discussed above, the data are not consistent with the hypothesis that learning alters the outcome of the plant-pollinator interaction observed here. However, we expect that the difference between scented and non-scented flowers would remain similar or even increase further due to the ability of the moth to associate the floral scent with the nectar reward. In the tent assay the moths were allowed to probe 10 consecutive flowers and were therefore already after the first visit, not fully naïve. However, despite their experience, the observed effect was not only maintained but was even increased in comparison to the single-experience experiments in the wind tunnel (Figure 2B and Figure 2—figure supplement 1D). Additionally, our new results also indicated that the moth maintained their high success rate on scented flowers as well as their relatively flow success rate on non-scented flowers also over several flower encounters (Figure 2—figure supplement 1C).

9) Methods, subsection “Insects”: female moths.

Changed.

10) Methods, subsection “Tent”: spatial resolution of scent cloud: if the EV and CHAL plants are touching, it is likely that the entire array is embedded within a scent cloud that precludes moths distinguishing between odor sources on the fly. If they repeated these experiments with plants separated by a few meters, they might get a different result. This problem is common to all array experiments performed with hawkmoths over the last decade (using Petunia, Mimulus, Hemerocallis).

We are thankful to the reviewer for raising this issue. It would indeed be possible that non-emitting flowers are “hiding” in the scent cloud of the emitting plants, which could (Figure 2B) and we had briefly discussed this in the fourth paragraph of the Results and Discussion section. We have now extended the relevant sentence.

Nonetheless, our results from the tent are in line with our findings in the wind tunnel where plants were introduced individually and non-scented flowers could thus not have profited from the scented flowers (Figure 1—figure supplement 1B), indicating that visual cues and general plant odorants are sufficient for a moth to detect a non-scented flower.

11) Methods, subsection “Tent”: please clarify whether batches of moths were released individually or in batches. The latter would violate assumptions of independence, or would render each released batch as a replicate, as in Drosophila bioassays.

The reviewer raises an important point and we have tried to clarify this in our Methods section (subsection “Tent”). Moths were released sequentially and visually tracked in the tent until they stop flying; therefore, individual moths did not repeat the assay and at no time there was more than one active moth in the tent.

12) Methods, subsection “Cross pollination experiment”: Please address the extent to which adding pollen to tongue tips impacted moth performance. They don't like being handled and having their probosces extended, but it is possible that doing this during the photophase, when they are quiescent, would not impair foraging behavior during scotophase.

We agree with the reviewer that hawkmoths are sensitive to handling and we have thus only included animals which took flight voluntarily within 3 min (73%) after the pollen was applied. This information was indeed missing from the Methods section and we have now added this in the subsection “Cross pollination experiment”. However, we would further argue that moths approaching emitting and non- emitting flowers were handled in the same way, it is therefore conceivable that our treatment might have decreased the overall rates of pollinations, but not have influenced the observed dramatic differences between emitting and non-emitting flowers.

13) Methods, subsection “Scanning electron microscopy”: provide replicates please, for males and females.

The difference between the male and female hawkmoth proboscis is a very interesting aspect to explore as we also mention in the Discussion. Nonetheless, we feel that it is out of the scope of our current paper, which focuses exclusively on male moth and nectar foraging. We now state in the manuscript more clearly that we worked only with male moths.

14) Methods, subsection “Proboscis choice”: wouldn't it be more accurate to have used sucrose solutions with BA instead of BA in distilled water? The colligative properties of the sucrose (salting out) likely would change emission rates of BA from that solution.

The reviewer raises a very interesting point, especially as many hawkmoth pollinated flowers vary strongly in their sugar concentration and composition (Contreras et al., 2013; Kaczorowski et al., 2005). This might then also influence the perception of secondary nectar metabolites. However, in N. attenuata BA is mainly emitted at the corolla limb surface (Euler and Baldwin 1996; Kessler and Baldwin, 2007). We would therefore assume that the BA emissions which influence the behavior of M. sexta in our study are mostly independent of the nectar sugar concentration and composition.

However, we followed the reviewer’s advice and tested the neuronal response to a 0.1 mM BA solution solved in a 0.5 M sucrose solution in 4 animals, as we did not find any significant difference to the BA dissolved in water (median= 37.25 ∆spikes/s, p= 0.12), we prefer to exclude these results from the manuscript.

15) Figure 1—figure supplement 1A: the diagram does not show an array (i.e. a checkerboard), but rather, two lanes of plants, color coded to indicate EV vs. CHAL. Is that true, or is the diagram a simplification? Please clarify, because the former design is more effective than the latter in terms of avoiding side of cage biases, etc.

The plants were arranged as in the diagram for feasible tracking and removal of visited flowers, and we have now aimed to make this clearer in the Materials and methods (subsection “Tent”) and in the figure legend of Figure 1—figure supplement 1. The reviewer is right that the design might be more susceptible to side bias and we have altered the sides of EV and CHAL plants between different moths and experimental days to reduce this effect. This point has indeed been missing in the Methods section and we have now added this to the aforementioned subsection and to the legend of Figure 1—figure supplement 1. Moreover, moth visited EV and CHAL plants (i.e. plants on the left and right side) in the same proportion and in a random sequence (probability of visiting a plant on the other side: 0.47, Figure 1—figure supplement 1A).

16) Panel C does not make sense. I don't understand what you are showing on the left side (Y axis wind tunnel width, x axis wind tunnel length) – please explain more clearly what you are testing here and what you found vs. expectations. You are not experimentally varying the dimensions of the wind tunnel! The right panel simply shows a non-significant trend, with 3:2 ratio of sample sizes, and is not strongly supported.

We are thankful to the reviewer for this comment and we have updated the figure legend. The left side of panel C represents a top view projection of the 3 dimensional flight tracking from two example flights in the frontal section of the wind tunnel. We agree that the sample size of the experiment is uneven and that the replicate number is rather low. However, the results are still of interest even if further careful experiments might be needed to further explore the mechanisms by which plant odors aid the navigation of M. sexta towards a plant or flower in the presents of visual cues.

Reviewer #2:

This study makes an interesting contribution to our understanding of an important problem. It shows that certain sensilla on the proboscis of Manduca are olfactory. A very clever behavioral paradigm is used to show that these sensilla detect a major component, benzyl acetone (BA), of the scent of wild Nicotiana attenuata. The study also shows that flowers engineered not to emit BA receive inferior pollination services, and that less nectar is removed from them. The simplest interpretation of these results taken together is that pollination and nectar removal depend on the sensilla, although this is not demonstrated directly by ablation or masking experiments.

An extensive study of the specificity of the sensilla is beyond the scope of this study, but the story would be more compelling if it were shown that the sensilla showed at least some odorant-specificity.

We agree that ablation or masking experiment might provide even further support for our findings; however, the tested sensilla also potentially contain mechano- and other chemoreceptors which might be important for flower probing. Results from such experiments might therefore seriously confound different sensory modalities and be difficult to interpret, as also mentioned by the first reviewer.

We agree that testing the response of the first multiporouse sensilla might enhance the manuscript and we have therefore followed the reviewer’s advice and tested the neuronal response to 41 additional odors (Figure 4C and Figure 4C–source data 1) and presented these results in the fifth paragraph of the Results and Discussion section. Interestingly, we found that while the senisillum responded most strongly to BA and the structurally related benzyl acetate, it could also detect some other potentially relevant compounds such as nicotine although at a much lower level.

https://doi.org/10.7554/eLife.15039.016

Article and author information

Author details

  1. Alexander Haverkamp

    Department of Evolutionary Neuroethology, Max-Planck Institute for Chemical Ecology, Jena, Germany
    Contribution
    AH, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Contributed equally with
    Felipe Yon
    Competing interests
    No competing interests declared.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3512-9659
  2. Felipe Yon

    Department of Molecular Ecology, Max-Planck Institute for Chemical Ecology, Jena, Germany
    Contribution
    FY, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Contributed equally with
    Alexander Haverkamp
    Competing interests
    No competing interests declared.
  3. Ian W Keesey

    Department of Evolutionary Neuroethology, Max-Planck Institute for Chemical Ecology, Jena, Germany
    Contribution
    IWK, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    No competing interests declared.
  4. Christine Mißbach

    Department of Evolutionary Neuroethology, Max-Planck Institute for Chemical Ecology, Jena, Germany
    Contribution
    CM, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    No competing interests declared.
  5. Christopher Koenig

    Department of Evolutionary Neuroethology, Max-Planck Institute for Chemical Ecology, Jena, Germany
    Contribution
    CK, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    No competing interests declared.
  6. Bill S Hansson

    Department of Evolutionary Neuroethology, Max-Planck Institute for Chemical Ecology, Jena, Germany
    Contribution
    BSH, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    BSH: Vice President of the Max Planck Society, one of the three founding funders of eLife, and a member of eLife's Board of Directors
  7. Ian T Baldwin

    Department of Molecular Ecology, Max-Planck Institute for Chemical Ecology, Jena, Germany
    Contribution
    ITB, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    ITB: Senior editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5371-2974
  8. Markus Knaden

    Department of Evolutionary Neuroethology, Max-Planck Institute for Chemical Ecology, Jena, Germany
    Contribution
    MK, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    Contributed equally with
    Danny Kessler
    Competing interests
    No competing interests declared.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6710-1071
  9. Danny Kessler

    Department of Molecular Ecology, Max-Planck Institute for Chemical Ecology, Jena, Germany
    Contribution
    DK, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Contributed equally with
    Markus Knaden
    For correspondence
    dkessler@ice.mpg.de
    Competing interests
    No competing interests declared.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0410-116X

Funding

European Research Council (advanced grant no. 293926 to Ian T. Baldwin)

  • Ian T Baldwin
  • Danny Kessler

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Sandor Nietsche, the EMZ Jena, and Lydia Gruber for their help with the SEM images, Jürgen Krieger for providing the Orco anti-body, Tamara Spingler for supporting the Y-tube assays and Daniel Veit for engineering the set-ups for behavior experiments. Funding was provided by the European Research Council advanced grant no. 293926 to ITB and the Max-Planck-Society.

Reviewing Editor

  1. Marcel Dicke, Wageningen University, Netherlands

Publication history

  1. Received: February 9, 2016
  2. Accepted: May 3, 2016
  3. Accepted Manuscript published: May 5, 2016 (version 1)
  4. Version of Record published: May 26, 2016 (version 2)

Copyright

© 2016, Haverkamp 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

  • 2,187
    Page views
  • 562
    Downloads
  • 11
    Citations

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

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)

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

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

Further reading

    1. Computational and Systems Biology
    2. Neuroscience
    Daniel B Dorman et al.
    Research Article Updated
    1. Evolutionary Biology
    2. Neuroscience
    Benjamin J De Corte et al.
    Research Article Updated