Drosophila suzukii rearing

Drosophila suzukii were reared in insect cages (W47.5 × D47.5 × H93.0 cm, BugDorm – 4 M4590, MegaView Science Co., Ltd., Taichung, Taiwan) on a cornmeal diet (DSCD(a) containing dry deactivated yeast), supplemented with dry baker’s yeast (RUF Lebensmittelwerk KG, Quakenbrück) and an additional 5% sugar solution provided on cotton, under a 16:8 h L:D photoperiod [20]. For all experiments, 4–8-day-old adults were used. The D. suzukii colony was established from field-infested cherries, blueberries, and grapes collected in South Tyrol, Italy, and was refreshed annually with field-collected individuals.

Apis mellifera rearing

Apis mellifera subsp. carnica (Pollmann) were kept in the experimental apiary of the Free University of Bolzano, located in Altenburg (46°23′12.6″N 11°13′57.5″E, South Tyrol, Italy). Five colonies of similar strength were created by shook-swarm in June. The 1.5 kg swarms of honey bees were placed in regular 10-frame Dadant hives for nomadic beekeeping, with six frames of organic wax foundation (Il Pungiglione Soc. Coop.). The swarms were provided with newly introduced sister queens and sugar syrup (Apiinvert®, Südzucker). The hives were arranged in a single row. After seven days, the new colonies were treated with 50 mL of 3.5% (w/v) oxalic acid dihydrate sucrose solution, trickled in between the frames to control the parasitic mite Varroa destructor [30].

Yeast materials and cultivation

Three yeasts were isolated at the Laimburg Research Centre in South Tyrol, Italy, from D. suzukii-infested grapes in South Tyrol, Italy, during a preceding study by Bellutti et al. (2018) [20]: Saccharomycopsis vini strain LB-NB-1.33 (accession number: KP298011, abbreviation: Sv 1.33) and Hanseniaspora uvarum strain LB-NB-1.21 (accession number: KP298009, abbreviation: Hu 1.21), strain LB-NB-2.2 (accession number: MK567898, abbreviation: Hu 2.2), and strain LB-NB-3.4 (accession number: MK567905, abbreviation: Hu 3.4). Saccharomyces cerevisiae (strain S288c, abbreviation: Sc S288c) is a conventional laboratory strain. For long-term storage, purified isolates were cultivated in chloramphenicol yeast glucose broth (5 g/L yeast extract, 20 g/L glucose, and 0.1 g/L chloramphenicol) and stored in 20% glycerol at -80°C. Yeasts were cultured in 220 mL of potato dextrose broth (PDB) (24 g/L Difco™ Potato Dextrose Broth) at 25°C, 120 rpm for 30 h in a 300-mL Erlenmeyer flask closed with cotton and aluminum foil [28].

Plant materials and cultivation

Leaves from strawberry plants (Fragaria x ananassa, cultivar Elsanta) and grape vines (Vitis vinifera, cultivar Vernatsch -also known as Trollinger or Schiava, clone: Edelvernatsch Lb 43, Rootstock: SO4) were used in behavioural experiments. Plants were grown between May and July in the greenhouse under controlled conditions (22 ± 2°C, 75 ± 5% relative humidity, without artificial light) and treated once a week for 20 min with vaporised sulphur against powdery mildew using a sulphur burner (Nivola B.V. 220V, Holland). No sulphur treatments were performed during the assays. Leaves were selected to be similar in size and colouration for both treatment and control samples. Purchased blueberries (Vaccinium corymbosum) from organic production and various seasonal cultivars were used as substrates for oviposition experiments.

Four-choice arena assay

To evaluate the preference of D. suzukii when multiple yeasts are available, the yeasts S. vini, strain 1.33 and H. uvarum, strains 1.21, 2.2 and 3.4 were presented in a competitive arrangement in a four-choice set-up (S1A Fig) [31]. For this purpose, five round glass dishes (diameter: 115 mm, height: 64 mm) were closed with a thin mesh. Each dish contained four traps, made of 4-mL glass vials that were closed with cut pipette tips that allowed the flies to get into the vials but prevented them from leaving. Each vial contained 1 mL of one of the four liquid yeast cultures. Groups of 20 females were placed into each dish and the number of females trapped was scored after 24 h.

The H. uvarum strain 2.2 was selected among the three H. uvarum strains for the rest of the study as there was no preference among them. Additionally, H. uvarum strain 2.2 was previously shown to be highly attractive to and to elicit electrophysiological response in D. suzukii [29].

Flight attraction in wind tunnel

To evaluate odour-mediated attraction of D. suzukii in response to yeast headspace volatiles, wind tunnel experiments were conducted using a slightly modified protocol from Reherman et al. (2022) (S1B Fig) [26]. For each yeast tested, 25 mL of a yeast culture grown in PDB for 24 h was poured into a wash bottle. The stimulus was delivered in charcoal-filtered air (0.3 L/min) that was blown through the wash bottle containing the odour source. The scented airstream was vertically injected at the upwind end into the wind tunnel onto an 18 cm high, 38 mm diameter horizontal platform of aluminium, from which it diffused downwind as an odour plume. Four-to-six days old adult females were starved for 24 h prior to testing. They were transferred individually to a 30-mL glass tube and released at the down-wind end of the tunnel and exposed for 3 min to the main air stream (0.3 m/s) carrying a plume of yeast odours. Mated individual females were tested to see if they responded to the volatiles released by H. uvarum (n = 32), S. vini (n = 32), S. cerevisiae (n = 42), or PDB as a control (n = 48). The number of flies performing an upwind flight towards the scented air was counted.

Field trapping

Trapping trials were performed in July 2020 at a forest edge in Laimburg (46°22’43.2“N 11°17’06.1”E, South Tyrol, Italy). The forest was a deciduous forest containing cherry and elderberry as host plants. The neighbouring crops included cherries, which were harvested shortly before the experiment. The traps consisted of 4-mL transparent glass vials filled with 2 mL PDB or culture of S. vini, H. uvarum or S. cerevisiae, respectively. To reduce the surface tension, 2 µL of Tween 20 per mL were added. The traps were fixed with a wire at a height of 1.5 m on branches, placed 3 m from each other and randomised. After 24 h, the traps were replaced to avoid non-experimental and undesired growth of microorganisms, then newly randomised. The number of trapped D. suzukii was counted. Each treatment was replicated 24 times.

Mortality and oviposition assays

Spinosad (Laser, Corteva AgriscienceTM, Milan, Italy) was selected as a suitable insecticide [26]. Four treatments were created as follow: (1) water, (2) water with spinosad, (3) S. vini + spinosad and, (4) H. uvarum + spinosad. For the samples containing Spinosad, 5.43 mg spinosad per L liquid was added. In each trial, single grape or strawberry plants were considered as replicates. Ten leaves per plant were treated with 10 drops of 10 mL using a multichannel pipette in the greenhouse. Two experiments were performed: one on the first day, and one 7 days following the treatment. Five leaves belonging to one plant were removed and placed in an insect cage with their stems inserted into an Erlenmeyer flask filled with water. The opening around the stems was closed with cotton. A 5% (w/v) sugar solution supplied on cotton in a small Petri dish (diameter 6 cm) served as a water and energy source for the flies. Each cage also contained a Petri dish with water agar (diameter 9 cm, 15 g/L agar) on which four well washed blueberries were placed. Groups of 20 males and 20 females D. suzukii were then inserted. The blueberries and the agar substrates were removed and replaced by a new set 24 h after the start of the experiment. After 48 h, egg-laying was quantified from the number of eggs laid on the agar and berries and adult mortality was assessed. Five replicates were performed for each treatment.

Attraction assay with Apis mellifera

The attractiveness of yeast cultures was assessed on honey bee workers. Experiments were performed near the hives, in September, a season with no major flower blossoming [32]. At least 30 min before each experiment, bees were trained on a plastic board placed 2 meters in front of the hives by providing sugar syrup and honey. A 5% (w/v) sugar water syrup was made fresh and added to the cultures of S. vini, H. uvarum and PDB. Three experimental designs were created: (i) the syrup and yeasts were presented on the same plate, (ii) the syrup and yeasts were presented on separate plates, (iii) the syrup and yeasts were presented in separate plates and dried at 35°C prior to the presentation. The dried conditions represented the yeast cultures after application on leaves. The PDB and sugar water syrup were tested with a similar sugar content to the yeast cultures [28]. The plates were randomly rotated every 5 min. The number of bees reaching each plate over a period of 5 min was assessed. Experiments lasted 80 min and were video recorded using an HD Camera (COAU Action, 4k 20mp) and VLC Media Player 3.0.11 (VideoLAN software).

Yeast headspace characterisation by SPME-GC-TOF-MS

Solid phase microextraction followed by gas chromatograph-time of flight-mass spectrometry (SPME-GC-TOF-MS, QP2010 SE Shimadzu) was used to characterise volatile organic compounds (VOCs) from yeast headspaces, following the methodology of Alves et al. [33]. For each sample (yeast cultures and PDB, six replicates per sample), 5 mL were transferred to a 20-mL vial and 1 g of NaCl (99.0%) was added. The vial was capped with a polytetrafluoroethylene (PTFE) septum and an aluminium screw cap, and the metabolic quenching was achieved by freezing the samples at -80°C. Strict control of the quenching procedure, which arrests the cellular metabolism and enzymatic reactions of yeast, was applied to reduce data variability.

Sample analysis was randomised, and pooled and blank samples were injected after every ten samples for quality control and normalisation. The thawed samples were held in the autosampler at 15°C, incubated at 40°C for 15 min, and extracted with a divinylbenzene-carboxen-polydimethylsiloxane (CAR/DVB/PDMS) fiber 50/30 μm for 45 min. Desorption was performed at 250°C for 2 min. The GC instrument was operated in spitless mode using helium as the carrier gas with 1 mL/min, and the separation was achieved on a ZB-WAX column (30 m x 0.25 mm ID, 0.25 µm thickness, Phenomenex). A gradient temperature programme was used: 35°C for 5 min, then increased linearly from 35°C to 250°C at a rate of 3°C/min, then held at 250°C for 5 min. The transfer line and ion source temperatures were set at 250°C, with the ion source voltage at 70 eV. Mass spectrometric data were acquired in full scan mode over an m/z range of 40–510.

MS-DIAL was used for deconvolution, peak picking and alignment [34]. A total of 564 features were extracted based on peak height. After blank subtraction and combination of fragments of the same peak, 217 compounds were identified. To characterise each compound, the mass spectra were compared with mass spectral libraries, including an in-house library of standards, the NIST 2017 database and retention indices. The experimental linear retention index (LRI) of each compound was calculated using a series of n-alkanes (C8-C20) under the same experimental conditions. In total, 156 compounds were annotated, each presenting similarity matches > 800 with the libraries and LRI-matches.

Yeast headspace volatile detection in EAG

Electroantennography (EAG) experiments were conducted to identify which chemicals are detected by adult D. suzukii. In each experiment, a fly was immobilised in a truncated plastic pipette tip, with half of its head protruding from the narrow end. Antennal activity was measured by placing a recording electrode over the tip of the antenna and an indifferent electrode near the base of the antenna (through the eye). Ag–AgCl glass electrodes were filled with Beadle–Ephrussi Ringer solution [35].

Signals were passed through a high-input impedance amplifier (2-channel USB acquisition controller, IDAC-2; Syntech) and recorded using GC-EAD 2014 software (v1.2-5, Syntech). An air pulse (stimulus) lasting 3 s was delivered through a cartridge into a carbon-filtered and humidified air stream directed at the fly preparation.

The cartridges consisted of a glass Pasteur pipette (ThermoFisher Scientific) with a filter paper (15 mm diameter, Whatman grade 1) placed within the larger end and closed with a 1-mL pipette tip. A 30 µL aliquot of a 1000-fold diluted chemical in paraffin oil (10^-3 v/v) was deposited onto the filter paper just before the experiment. A cartridge containing only paraffin oil served as a solvent control. A positive control, using 2-heptanone, was used to correct for eventual antennal fatigue and to standardise the responses across replicates. This compound is known to elicit antennal responses in D. suzukii [36]. Each cartridge was used for a maximum of three stimulations to prevent significant changes in chemical concentration. All chemicals used in the experiments were purchased with the highest purity available (S2 Appendix).

Yeast headspace volatile detection in GC-EAD

Coupled gas chromatography-electroantennography detection (GC-EAD) was used to elute each pure chemical onto the antenna and identify any antennally active contaminants present in the standard solutions tested. In this setup, the electroantennography detector (described above) was coupled with a GC (7820A, Agilent Technologies) equipped with a flame ionisation detector (FID). The fly was prepared as described above.

Standards were diluted in dichloromethane and organised into four mixtures, each containing several compounds at a 10^-3 v/v dilution. A 3 µL aliquot of each mixture was injected into the GC column (HP-5MS Agilent 19,091 J-413 column, 0.25 μm coating 30 m length and 0.32 mm diameter) through a cool-on-column (COC) injector. Helium, at a flow rate of 2.5 mL/min, was used as the carrier gas. The oven method was programmed as follows: inject at 50°C and hold for 1.8 min, then 7.3°C/min to 250°C and hold for 3 min. The injector temperature was set at 250°C and the detector temperature was set at 350°C. The column effluent was mixed with a nitrogen make-up and split at a 1:1 ratio. One portion flowed to the FID, while the other portion was directed through a transfer line at 170°C into a charcoal- filtered and humidified airstream over the mounted fly. Signals were amplified via an EAG amplifier (as described above). GC-FID and EAD signals were simultaneously recorded using GC-EAD 2014 software (v 1.2-5, Syntech).

Data analysis

Statistical analyses were performed using R 4.4.1. (R Core Team, 2024). A significance level of P = 0.05 was used for all comparisons. A generalised linear mixed model (GLMM, package `lme4’) fitted with a Poisson error distribution was applied to evaluate the distribution of flies accross the four traps ine the arena, and the number of trapped D. suzukii flies per vial in field trapping experiments.

The upwind flight attraction of yeast odours in the wind tunnel was analysed using a GLMM fitted with a binomial error distribution. The mortality rate of D. suzukii and the number of eggs laid (oviposition) per cage were evaluated using a generalised linear model fitted with a gamma distribution. To handle zero values and to allow the use of a gamma distribution, datasets were transformed using x + 1. Model selection was based on Akaike information criterion (AIC) values, and residuals were examined to verify the distribution of the errors. Treatment and fly sex (where applicable) were included as fixed effects. Post hoc comparisons were performed using Tukey’s contrast pairwise test (package ‘multcomp ’[37]).

The number of honey bees on feeding plates was analysed using linear mixed models (package `lme4’ [38]). Bee counts were considered as the response variable, while the tested solution, the position on the board, and time were considered as predictors. The rounds (i.e., the intervals during which the tested solutions were rotated on the board) were accounted as random effects. Model fitting was evaluated through residuals analysis (package `DHARMa’ [39]). Post-hoc comparisons between solutions were conducted using pairwise comparisons (`multcompview’ package [40]). Type II ANOVA tables (Anova function, `car’ package [41]) were generated to summarise the significance of fixed effects in the mixed-effects model.

The headspace composition was analysed as follows: peak height results were normalised based on the systematic error removal using the random forest (SERRF) method to remove systematic errors [42]. To identify significantly impacted lipid clusters among yeasts, a chemical similarity enrichment analysis (ChemRICH) was performed on the 156 annotated compounds identified, using the ChemRICH platform. This platform clusters significantly impacted metabolites based on chemical similarity and ontology mapping to highlight biologically relevant patterns. Next, a Kolmogorov–Smirnov test was used for subsequent statistical analysis [43].

The EAG and EAD signals were integrated using GC-EAD 2014 v1.2-5 (Syntech). Responses were measured as the maximum voltage deflection following the start of the stimulus. In EAG experiments, the antennal response to the solvent control was subtracted from the response to each standard, and the resulting values were normalised to the response elicited by the positive control (2-heptanone), which was set to 100%. The distribution of response amplitudes was assessed using a Shapiro-Wilk normality test. Subsequently, responses to the standards were compared to those elicited by the solvent control using two-sided paired t-tests (`stats’ package [44]). Standards that elicited a response significantly greater than paraffin oil were considered antennally active. In GC-EAD experiments, a standard was considered antennally active if at least three flies displayed a non-zero voltage deflection. For both EAG and GC-EAD experiments, only flies that responded to the positive control were included in the analysis (n = 6–10).

Saccharomycopsis vini and Hanseniaspora uvarum are attractive in long and short-range behavioural experiments

In the first behavioural assay, S. vini and three strains of H. uvarum were simultaneously presented to D. suzukii for 24 h in a 4-choice arena to evaluate their preference (Fig 3A, S1 Fig). Baits containing S. vini were significantly more attractive than those containing H. uvarum 1.21 (GLMM Poisson distribution, MCM: X²₃= 25.72, z = 3.72, P = 0.001), H. uvarum 2.2 (z = 3.85, P < 0.001) and H. uvarum 3.4 (z = 3.20, P = 0.008). The three H. uvarum strains were not different from each other (P > 0.05). Notably, 2% of the flies tested did not choose any yeast bait.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Short- range and long range attraction assays.

(A) Mean (± SD) number of female Drosophila suzukii trapped in four simultaneously presented baits within 24 h. Baits consisted of yeast culture of Hanseniaspora uvarum and Saccharomycopsis vini in potato dextrose broth (PDB). Three strains of H. uvarum: Hu 1.21, Hu 2.2 and Hu 3.4 were tested. Bars with different letters are significantly different (GLMM Poisson distribution and multiple comparison of means, P < 0.05). B) Percentage of females flying upwind towards headspace volatiles of H. uvarum (Hu 2.2), S. vini, Saccharomyces cerevisiae and PDB in a wind tunnel. Bars with different letters are significantly different (GLM binomial distribution and multiple comparison of means, P < 0.05). C) Mean (± SD) number of females (plain bars) and males (striped bars) caught in traps placed on trees at a forest edge. Traps were baited with 2 ml culture of H. uvarum (Hu 2.2), S. vini, S. cerevisiae or PDB. Bars with different letters are significantly different (GLMM, Poisson distribution, P < 0.05).

https://doi.org/10.1371/journal.pone.0323653.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Mortality and oviposition assays with attract-and-kill formulation.

(A–D) Mean (± SD) mortality rate (%) of female (‘plain’ bars) and male (‘stripe’ bars) Drosophila suzukii exposed to treated strawberry (red bars, panels A and C) or grapevine (green bars, panel B and D) leaves one day after treatment (panels A and B) and seven days after treatment (panels C and D). (E–H) Mean number (±SD) of eggs laid during a 48 h exposure to treated strawberry (red bars, panels E and G) or grapevine leaves (green bars, panels F and H) one day after treatment (panels E and F) and seven days after treatment (panels G and H). Leaves were treated with formulations of spinosad with water, Hanseniaspora uvarum, or Saccharomycopsis vini. A water treatment served as a control. Bars with different letters are significantly different (GLM and Tukey´s multiple comparison, P < 0.05).

https://doi.org/10.1371/journal.pone.0323653.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Attraction assay with Apis mellifera.

(A) Total number of Apis mellifera foragers feeding on four plates simultaneously presented, containing 5% sugar syrup, potato dextrose broth (PDB) growth medium, and yeast cultures of Hanseniaspora uvarum and Saccharomycopsis vini over a period of 80 minutes in three separate experiments where plates contained: 1) liquid yeast cultures + sugar syrup 5%; 2) liquid yeast cultures; 3) dry yeast cultures. The straight lines report the linear trend calculated as y = ax + b. Asterisks indicate statistical significance (** P < 0.01, *** P < 0.001) of the time on the number of feeding bees. B) Mean (± SD) number of bees feeding from the four plates over 5 min. Different letters report a statistical difference between treatments (LMM, and pairwise multiple comparison, P < 0.05). Foragers from five colonies of similar strength were tested. Sixteen observations of 5 min each were performed 30 min after training with sugar syrup.

https://doi.org/10.1371/journal.pone.0323653.g003

The second behavioural assay, evaluated the flight attraction of D. suzukii females to yeast odours in a wind tunnel. During a 3-min test period, flies were exposed to each yeast, placed upwind (Fig 1B, S1 Fig). Flies displayed flight behaviour and were similarly attracted to S. vini and H. uvarum (GLM binomial distribution: X²₃ = 36.85, z = 0.50, P = 0.62). Both yeasts were significantly more attractive than S. cerevisiae (z = ‐2.30, P = 0.02). The control (PDB) elicited significant less upwind flight than any of the three yeasts (P < 0.05).

In a third experiment, traps containing small volumes (2 ml) of either S. vini, H. uvarum or S. cerevisiae cultures, were placed on a forest edge to evaluate their attractiveness to wild D. suzukii populations (Fig 3C). Traps baited with H. uvarum and S. vini captured significantly more flies than those baited with S. cerevisiae (GLM, Poisson distribution, F_2,141 = 193.24, S. vini: z = 4.959, P < 0.001; H. uvarum: z = 5.106, P < 0.001). There was no difference in the number of flies captured in the baits with S. vini and H. uvarum (z = 0.293, P = 0.952). No flies were found in traps containing PDB. Notably, traps baited with H. uvarum captured more males than females (F_1,140 = 4.800, P = 0.028) whereas no sex bias was observed in traps baited with S. vini.

Attract-and-kill formulations with S. vini and H. uvarum effectively reduced oviposition and increased mortality

On strawberry leaves, one day after treatment the mortality was significantly impacted (F_3,36 = 53.653, P < 0.001, Fig 2A). Mortality in the water control was significantly lower compared to spinosad formulations with H. uvarum (z = 4.763, P < 0.001) and S. vini (z = 4.714, P < 0.001). Mortality was also significantly lower in spinosad with water compared to formulations containing S. vini (z = 4.675, P < 0.001) and H. uvarum (z = 4.731, P < 0.001). Oviposition was not significantly affected after exposure to the three formulations or water (F_3,16 = 2.862, P = 0.069, Fig 2E).

Seven days after the application, a significant effect on the mortality was observed (F_3,36 = 8.507, P < 0.001, Fig 2C). Mortality in the water control remained significantly lower compared to spinosad formulated with H. uvarum (z = 3.077, P = 0.009) and S. vini (z = 3.412, P = 0.003). Oviposition was also significantly affected (F_3,16 = 8.390, P = 0.001, Fig 2F) and was significantly higher in the water control compared to spinosad formulations with S. vini (z = 2.880, P = 0.017) and H. uvarum (z = 3.079, P = 0.009).

On grapevine leaves, one day after treatment and after 48 h exposure mortality was significantly affected (F_3,36 = 27.209, P < 0.001, Fig 2B). Mortality in the control was significantly lower compared to spinosad with H. uvarum (z = 5.293, P < 0.001) and S. vini (z = 5,426, P < 0.001). Furthermore, the mortality with spinosad in water was significantly lower compared to formulations containing S. vini (z = 4.872, P < 0.001) and H. uvarum (z = 4.705, P < 0.001). Oviposition was also significantly affected (F_3,16 = 6.105, P = 0.006, Fig 2G). Oviposition in the control was significantly higher compared to spinosad with S. vini (z = 2.873, P = 0.019) and H. uvarum (z = 3.296, P = 0.005).

Seven days after the application, the mortality was also significantly affected (F_3,36 = 54.181, P < 0.001, Fig 2D). Mortality in the control was significantly lower compared to the one following exposure to spinosad formulated with H. uvarum (z = 6.008, P < 0.001), S. vini (z = 5.996, P < 0.001) and water (z = 3.824, P < 0.001). Mortality following exposure to spinosad with water was also significantly lower compared to spinosad with H. uvarum (z = 5.097, P < 0.001) and S. vini (z = 5.064, P < 0.001), which were not significantly different from each other (z = 0.099, P = 1). Oviposition was also significantly affected (F_3,16 = 12.018, P < 0.001, Fig 2H). Oviposition after exposure to spinosad with S. vini was significantly lower compared to spinosad with water (z = 4.489, P < 0.001), with water control (z = 4.979, P < 0.001) and H. uvarum (z = 4.216, P < 0.001).

Yeast cultures did not stimulate foraging of Apis mellifera

The attractiveness of yeast VOCs and sugar content was assessed using trained honey bees to determine whether the application of yeast cultures would interfere with their foraging behaviour (Fig 3). Across all three experimental conditions and throughout the 80-minute trial, a 5% sugar syrup consistently proved more attractive than yeast cultures or PDB: (1) when syrup was presented alongside yeast cultures (LMM, χ²_₃,₆₃ = 17.98, P < 0.01), (2) when syrup and liquid yeast cultures were presented separately (LMM, χ²_₃,₆₃ = 149.17, P < 0.001), and (3) when syrup and dried yeast cultures were presented separately (LMM, χ²_₃,₆₃ = 140.87, P < 0.001). Time had a significant effect in two conditions: (1) when syrup was presented with the cultures (LMM, χ²_1,63= 170.81, P < 0.01), showing an overall increase of visitation (bee counts) and (3) when syrup and dried yeast cultures were presented separately (LMM, χ²_1,63= 32.82, P < 0.001), where visitation decreased over the course of the trial.

Yeasts have distinct and overlapping headspace compositions in SPME

We characterised 156 VOCs by SPME-GC-TOF-MS and performed a ChemRICH analysis to statistically compare the presence and abundance of each compound between S. vini, H. uvarum and S. cerevisiae (S2 Table). From this analysis we extracted classes of chemicals that significantly differed between the yeasts (S3 Table). The headspace compositions of S. vini and H. uvarum were significantly different in alcohols and polyols, benzene and substituted derivatives, branched unsaturated hydrocarbons, carboxylic acid esters, fatty acid esters, fatty alcohols, ketones and monoterpenoids (FDR-adjusted Kolmogorov–Smirnov-test, P < 0.01). Comparing H. uvarum and S. cerevisiae, significant differences were found in alcohols and polyols, benzene and substituted derivatives, carboxylic acid esters, fatty acid esters, ketones, medium-chain fatty acids, monoterpenoids and sesquiterpenoids (FDR-adjusted Kolmogorov–Smirnov-test, P < 0.01). Saccharomycopsis vini and S. cerevisiae headspace compositions also differed significantly in alcohols and polyols, benzene and substituted derivatives, branched unsaturated hydrocarbons, carboxylic acid esters, fatty acid esters, fatty alcohols, ketones, medium-chain fatty acids, monoterpenoids and sesquiterpenoids (FDR-adjusted Kolmogorov–Smirnov test, P < 0.01).

Within these classes, 40 compounds differed significantly between the yeasts (Table 1). Specifically, S. vini produced significantly higher amounts of 14 monoterpenoids and 4-methyl-1-pentanol. H. uvarum produced the largest amount of 2-phenethyl acetate, isobutyl acetate, 2-methylbutyl acetate, 2-nonanol, acetoin and (6E)-nerolidol. Both H. uvarum and S. vini produced significantly higher amounts of ethyl acetate, ethyl propanoate and 2-acetylthiazole compared to S. cerevisiae. Conversely, S. cerevisiae produced the largest amounts of 3-methyl-1-butanol, 2-phenylethanol, 2-methylthiolan-3-one, 1-heptanol, ethyl octanoate, hexanoic acid, octanoic acid, nonanoic acid and decanoic acid.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Mean amounts of headspace compounds in Hanseniaspora uvarum, Saccharomycopsis vini and Saccharomyces cerevisiae, antennal response amplitudes in EAG, and percentage of Drosophila suzukii responding in GC-EAD.

https://doi.org/10.1371/journal.pone.0323653.t001

Yeast headspace volatiles detection in EAG and GC-EAD

In EAG experiments, we tested whether D. suzukii could detect the 40 compounds released in significantly different quantities by the three yeast species (Table 1, S4 Table). Significant antennal responses were observed for 14 compounds (paired t-test, P < 0.05): 3-methyl-1-butanol, 4-methyl-1-pentanol, γ-terpinene, ethyl acetate, ethyl propanoate, isobutyl acetate, 2-methylbutyl acetate, ethyl butanoate, ethyl octanoate, acetoin, 2-methylthiolan-3-one, nonanoic acid, allocimene, and (Z)-geraniol.

Then, we also assessed whether these 40 compounds were antennally active in GC-EAD (Table 1, S5 Fig). Antennal responses (deflection < 0 mV) were observed in 100% of the tested flies for 11 compounds: 3-methyl-1-butanol, 4-methyl-1-pentanol, 3-methyl-1-pentanol, toluene, ethyl acetate, ethyl propanoate, isobutyl acetate, 2-methylbutyl acetate, ethyl butanoate, acetoin and 2-methylthiolan-3-one. In addition, responses were recorded in 4 out of 10 flies for linalool and β-myrcene.