Drosophila melanogaster Dalby-HL strain (Ruebenbauer et al., 2008) and Drosophila suzukii originated from infested fruits in Italy and maintained at the Swedish University of Agricultural Science (SLU), Alnarp (Rehermann et al., 2021) were reared on Bloomington standard cornmeal diet. Three to five days old flies were used for experiments. Colonies of two tephritid species, B. dorsalis and C. capitata were started from pupae provided by the International Atomic Energy Agency (IAEA, Vienna, Austria). Larvae were reared on a carrot-based diet (Biasazin et al., 2018). Adult flies were kept in bugdorm cages (L32.5 × W32.5 × H32.5 cm, Mega View Science Co., Ltd., Taiwan) with mesh sizes (96 um × 26/680 um) opening and fed with a mixture of 3:1 sugar: yeast and were provided with water-soaked cotton balls on a 9 cm plastic petri dish. Females were separated upon emergence and virgin females (4–5 days old) were used for experiments. Both Drosophilidae and Tephritidae flies were reared at 25°C, 50–65% RH and 12:12 h light: dark photoperiod.

Baker’s yeast (S. cerevisiae) (Kronans Jäst, Jästbolaget AB) was used for this study. One gram of the yeast cells was mixed with 3 ml of deionized water and 10 μl was streaked on agar plates and left for 48 h. When colonies were observed, a single colony was transferred to a 50 ml broth of minimal media (MM) containing 10% glucose. Minimal media was selected as it has no odorants that may interfere with the yeast production (Becher et al., 2018). Equal amounts 25 μl of the yeast-broth was then transferred to 4 ml MM and aged for 1, 8, and 15 days under controlled climatic conditions, on an oscillating incubator at 260 rpm and 25°C until needed for behavior, electrophysiology, and mass spectrometry. This procedure was followed throughout the experiments.

A 6-choice olfactory setup modified from Biasazin et al. (2019) was used for the behavioral experiments. The setup was made up of a cubic glass arena (45 cm × 45 cm × 45 cm) covered with a glass plate with six circular 50 mm holes, fitted with inverted conical glass funnels of 45 mm top and 4 mm bottom holes through which flies could pass, the glass funnel was surrounded by a fly collection cylinder of 50 mm diameter × 40 mm high, and on top of which was a similarly sized odor placing cylinder, a glass plate with a 2 mm hole separates the upper and lower cylinders. On the very top, covering the odor placing cylinder was a glass plate cover with 6 holes (5 mm) receiving clean odor (Supplementary Figure 3). A 0.5 l/min air flow was received through the holes of the top glass cover. The airflow was regulated using a pump (Elite 802, Hagen Ltd, UK and airflow meters (Model ExK Kitola Instrument Oy, Muurame, Finland). The airstream was filtered and humidified through two wash bottles containing activated charcoal and distilled water. Teflon tubing was used throughout. The top light source (daylight lamp, Photo studio CFL 45 W, 5000 K) was diffracted using an opaque white plexiglass panel. The whole setup was covered with white fabric to avoid visual disturbance by external movements. Treatments (1 ml of differentially aged yeasts and MM as a control or 100 μl of synthetic blends and paraffin oil as a control) were transferred to a 5 ml cup (Nolato Cerbo AB, Sweden) and placed in each of the odor holding cylinders. Treatment cups and glass chambers were rotated between experiments to avoid effects of cross contamination and position. Thirty flies were released in the arena and the number of flies that chose either of the treatments in the six chambers were recorded after 30 min. Fifteen replicates were made for each species. All glass surfaces of the setup were thoroughly cleansed with 96% ethanol and water after a single run, following drying overnight for next experiments.

Flies were immobilized in a 200 μl micropipette tip with the antennae exposed. Capillaries filled with Beadle-Ephrussi ringer solutions (7.5 g Nacl, 0.35 g Kcl, 0.29 g Cacl₂ dissolved in 1 L of distilled water) were used to create conduction between the electrodes and the insect antenna. The glass capillary attached to the reference electrode was inserted into the head of the fly, and the glass capillary on the recording electrode was connected to the tip of the antennae. The recording electrode was connected to a pre-amplifier probe and then to a high impedance GC amplifier interface box (IDAC-2; Syntech, Kirchgarten, Germany). The temperature program of the GC oven was set to start at 40°C and held for 1 min, increasing by 10°C min^–1 to 250°C and held for 1 min. The inlet was in splitless mode with temperature of 250°C. The headspace sample was carried by H₂ through a polar DB-WAX capillary column (60 m × 0.25 mm i.d., 0.25 μm film thickness, USD608325H Agilent Technologies Inc.). The effluent from the GC was split 1:1 between the flame ionization detector (FID) and the EAD. The column capillary for the EAD passed through a Gerstel olfactory detection port-2 transfer line tracking the GC oven temperature into a glass tube (30 cm × 8 mm), where it was mixed with charcoal filtered and humidified air at a flow rate of 1.5 l min^–1 and passed over the fly’s antenna. EAD peaks were analyzed using syntech data acquisition system software (GcEad 2012 v1.2.4, syntech, Germany).

Odor sampling was performed using SPME (solid phase microextraction) fibers coated with DVB/CAR/PDMS, Merck KGaA, Darmstadt, Germany). Sampling was done for 10 min using 1 ml of yeast-broth in a 20 ml vial (Precision Thread Headspace-Vials 75.5×, Genetec, Västra Frölunda, Sweden) fitted with a PTFE-lined septa. Before sampling the SPME fibers were conditioned for 10 min. After sampling the SPME fibers were injected to either the inlet (250°C) of a Gas Chromatography coupled Mass Spectrometry (GC-MS, Agilent 7890B GC and 5977A MS, Agilent Technologies Inc., Palo Alto, CA, USA) or a Gas Chromatography coupled with Electroantennography (GC-EAD). Both the GC-MS and the GC-EAD used a polar DB-Wax capillary column (60 m × 0.25 mm i.d., 0.25 um film thickness, USD608325H Agilent Technologies Inc.), with helium as carrier gas in the GC-MS and hydrogen in the GC-EAD. The temperature programs of the GC-MS were as per the GC-EAD system described above. Identification and quantification of compounds were conducted using chemstation software (Agilent technologies). Compounds were initially identified by mass spectra searching of the libraries; Nist20, Alnarp 11 and Wiley’s 275. The identification was supported by comparing the retention indices generated by running standard mixtures of alkane (C6-C20) under the same conditions as the samples and comparing the identity of the compounds Kovats retention indices obtained from electrophysiology and the GC-MS with published sources. Antenna active compounds were further confirmed by injecting synthetic compounds. Quantification was performed using 10 ng/μl solutions of three esters (pentyl acetate, ethyl hexanoate and ethyl octanoate) for compounds that elute early (before 8 min), in the middle (until 12 min), and late (after 13 min), respectively.

Based on the antennal responses of drosophilids and tephritids two major blends (base blends) were formulated (Table 1). One of the major blends contained ethyl acetate, 3-methylbutan-1-ol and ethanol which were compounds commonly shared across Tephritidae antennal responses, whereas the second major blend was formulated based on compounds commonly shared across Drosophilidae antennal responses and it was composed of acetoin, acetic acid, and ethyl acetate. Specific blends for each species of fruit flies were also formulated by adding uniquely detected compounds from the 1- to 8-days old cultures to the major blends (Base + 1-day or Base + 8-day). For B. dorsalis, Base + 1-day was composed of the base blend for tephritidae combined with ethyl nonanoate and 3-methylbutyl acetate, whereas Base + 8-day was composed of the tephritidae base blend combined with 2-phenylethanol (Table 1). For C. capitata Base + 1-day refers to the tephritid base blend combined with ethyl hexanoate, whereas Base + 8-day refers to the tephritidae blend combined with butane 2,3-dione. For D. melanogaster and D. suzukii Base + 1-day refers to the drosophilid base blend combined with ethyl octanoate and ethyl hexanoate, whereas Base + 8-day refers to the drosophilid base blend combined with ethyl 2-hydroxypropanoate for D. melanogaster and the drosophilid base blend combined with ethanol and butane 2,3-dione for D. suzukii. These blends were compared with an active culture (1-day old) with 8-days or 1-day old unique compounds added to it (Table 1). All compounds except for ethanol and 3-methylbutan-1-ol were formulated in 1:1 ratio. Ethanol and 3-methylbutan-1-ol were formulated in an 11:3 ratio. The compounds were diluted 10,000 (10^–4) folds for behavioral assays. The compounds used had >98% purity and were purchased from Merck (Darmstadt, Germany).

The number of fruit flies caught by different treatments in the 6-choice assay were analyzed using generalized linear models. The data was fitted with a poisson distribution and tested for over and underdispersion using AER (Kleiber and Zeileis, 2008), if the model was underdispersed a gaussian distribution was used instead. The data was never found to be overdispersed and hence no negative binomial was used. For each of the four species, 30 flies were released per experiment and replicated 15 times. The effects of the different treatments were tested using estimated marginal means using package emmeans (Lenth, 2022).

Electrophysiological data were collected and responses in each run were annotated using the software GC-EAD 2012 (Ver. 1.2.4. Syntech Inc. Kirchzarten, Germany). To be able to compare response profiles across runs and between species, responses were normalized as per (Biasazin et al., 2019), by expressing individual responses as a fraction of the average response to all compounds.

The amount in ng/μl of volatiles in the yeast was determined by calculating the ratio of peak areas in the samples with known injected amounts of (10 ng) of pentyl acetate, ethyl hexanoate and ethyl octanoate, which were compounds also present in the yeast volatilome. The mean difference between the amount in ng/μl of GC-EAD active compounds (n = 16) were analyzed using one-way anova followed by a TukeyHSD post-hoc comparison.

The average peak areas of the volatiles emanated from 16 injections of the three developmental ages (1-, 8-, and 15-days) of the yeast were visualized in a heatmap with a dendrogram separating the variables based on Jaccard dissimilarities indices using the package vegan (Oksanen et al., 2022), the same dissimilarity indices were used for non-metric multidimensional scaling (NMDS) to further analyze the volatilome. All analysis, visualizations and statistics were performed using (R Core Team, 2020).

Gas Chromatography -MS profiles revealed both qualitative as well as quantitative differences in the chemical profiles of differentially aged yeast culture. Volatiles of the active (1-day old) yeast were out grouped from the older cultures (8- and 15-days) both in heatmap 3 visualization using Euclidean distance matrix (Figure 1) and non-metric multidimensional analysis (NMDS) using Jaccard distance matrix (Supplementary Figure 2). The quantitative differences were more discernible than the qualitative differences (Figure 1 and Supplementary Figure 4). That is, 86% of the compounds present in the headspace of active yeast culture (1-day old) volatiles, were also found in older yeast cultures (8- and 15-days). Out of the 48 compounds in the 8-days yeast culture 56% were present in 1-day and 60% in 15-days yeast culture. Yet the absolute amounts and ratios differed with the age of the yeast.

As the yeast cultures became older, more, and higher amounts of acids and alcohols were observed in the headspace (Figure 1 and Supplementary Table 1). For example, in the antennally active compounds, the amounts of acetic acid, ethanol, 2-phenylethanol and 3-methylbutan-1-ol (Figure 2), were lower in 1-day compared to 8-days old media (p < 0.001). The amount of 1-butanol-3-methyl acetate was significantly lower in 15-days old yeast volatiles (P < 0.001). Subsequent TukeyHSD analysis revealed no significant differences between the 8 and 15-days old yeast volatiles in the amounts of acetic acid, ethanol, 2-phenylethanol and 3-methylbutan-1-ol. There was also no significant difference in the amounts of 1-butanol-3-methyl acetate between 1 and 8-days old yeast volatiles (p > 0.05). The amount of ethyl acetate and propan-1-ol in the three developmental ages of the yeast were not significantly different (p > 0.05) in both cases. Ethyl hexanoate and ethyl octanoate were not detected in volatilome of the 15-days old yeast culture and ethyl 2-hydroxypropanoate was not detected in volatilome of the 1-day old yeast culture. Comparison between the two ages revealed a significant difference in the amounts of ethyl hexanoate (p < 0.001) and ethyl 2-hydroxypropanoate (p = 0.03), but not in the amounts of ethyl octanoate (p = 0.19). Since some compounds, such as ethyl nonanoate and two unidentified compounds, elicited antennal response only from active (1-day old) yeast culture (Figures 2, 3), and compounds such as butan-1-ol and butane 2,3-dione elicited antennal response only from the 8-days old yeast culture (Figures 2, 3), no statistical comparison was performed for these compounds.

In a 6-choice olfactory setup, D. melanogaster and B. dorsalis fruit flies manifested distinctive preferences for differentially aged yeast culture. Although all developmental ages were significantly attractive compared to uninoculated minimal media (p < 0.001), distinctive preferences were observed for the different ages of the cultures. While D. melanogaster was more attracted to 1-day old yeast cultures, B. dorsalis preferred 8- and 15-day yeast cultures (Figure 4). There was no significant difference in preference of flies between 8- and 15-day old cultures (p = 0.923 for B. dorsalis and p = 0.955 for D. melanogaster). Like that of D. melanogaster, D. suzukii flies were significantly more attracted to active (1-day old) yeast culture compared to older (8- and 15-day) yeast cultures (p < 0.0001). There was no significant difference between older yeast cultures (8- and 15-day old) in D. suzukii (p = 0.999, Figure 4). However, C. capitata differed in that it preferred the 15-day old culture less than the 8-day old culture (p = 0.027). In addition, 1- and 8-day old cultures were equally attractive (p = 0.225).

Since only 3 compounds (ethyl acetate, ethanol and 3-methylbutan-ol) from the 15-day old yeast cultures induced antennal responses that were already present in the 8-day culture, they were not included in the graphs. Electrophysiological responses of fruit flies to the 1- and 8-days old yeast volatiles were analyzed. In total, 18 compounds elicited antennal response in either of the four species of fruit flies tested, out of which 14 compounds were identified (Figure 3). Mass spectra is provided for the 4 unknown compounds (Supplementary Figure 5). Ethyl acetate, which is found in active and old yeast culture, is the only compound that elicited antennal response in all fruit flies.

Bactrocera dorsalis responded to 10 of the 18 compounds, 9 from 1- to 8 from 8-days old cultures. Whereas ethyl nonanoate and 3-methylbutyl acetate elicited visible responses in B. dorsalis antennae only from 1-day old cultures, 2-phenylethanol elicited detectable responses only from 8-days old cultures. That is, B. dorsalis antennae detected 7 compounds shared between the two development ages of the yeast (Figure 3), although ratios of these compounds differed significantly between 1- and 8-days old cultures.

Ceratitis capitata responded to only 5 of the 18 compounds, with 4 responses to volatiles from 1-day old cultures as well as 8-days old cultures. While ethyl hexanoate elicited responses in C. capitata only from 1-day old cultures, butane 2,3-dione elicited responses only from 8-days old cultures, leaving three compounds shared between the two yeast cultures (Figure 3).

Drosophila melanogaster responded to 9 out of the 18 compounds. Eight were detected in the headspace of 1-day old and 4 in 8-days old headspace. Ethyl octanoate, ethyl hexanoate and two of the unknown compounds (unknown-01 and unknown-02) elicited response in D. melanogaster only from 1-day old yeast, whereas ethyl 2-hydroxypropanoate elicited response only from 8-days old yeast. Four of the 9 compounds were commonly detected among the two substrates (Figure 3).

Drosophila suzukii responded to 8 out of the 18 compounds, 3 compounds were detected from 1- to 8 from 8-days old yeast. Two of the unknowns (unknown-3 and unknown-4) and 3-hydroxybutan-2-one elicited response in D. suzukii only from 8-days old yeast. Four compounds were commonly detected among the two substrates (Figure 3).

3-methylbutyl acetate, ethanol and ethyl acetate were commonly detected in Tephritidae, whereas acetic acid, ethyl acetate and 3-hydroxybutan-2-one were commonly detected in Drosophilidae. Synthetic blends of these shared antennally active compounds in combination with uniquely detected compounds from each species were used for further behavioral analysis.

Compounds that evoked distinctive antennal responses contributed to the differential attractiveness of 1- and 8-days old yeast culture. Adding uniquely detected compounds found in active (1-day old) or aged (8-days old) volatiles to the base blend restored attraction, despite variable responses between different species of fruit flies.

For B. dorsalis, adding uniquely detected compounds from the 8-days old yeast volatiles to active cultures (1-day old) was as attractive as uniquely detected compounds from the 8-days yeast volatile added to the base blend (p = 0.99). The base blend with or without 1-day unique compounds was less attractive, with no difference between the two (Figure 5, p = 0.67). The minimal media was not attractive (p < 0.001).

Ceratitis capitata flies were most attracted to active 1-day old cultures with uniquely detected compounds from both 1- and 8-days cultures, with no statistical difference between the two (p = 0.975). Flies were also attracted to the base blends with or without addition of unique compounds (with no difference between them, p = 0.99). All treatments were more attractive compared to the control (minimal media, p < 0.05).

Drosophila melanogaster was highly attracted to 1-day old unique volatiles added to the base blend compared to volatiles of 8-days old medium added to the base blend (p < 0.001). D. melanogaster was equally attracted to the active yeast culture with unique compounds from 1- to 8-days old yeast cultures (p = 0.433). The base blend alone or with 8-day unique compounds was marginally attractive with no difference between the two (p = 0.98). All treatments were more attractive than minimal media (p < 0.005).

Drosophila suzukii did not uniquely detect volatiles from the 1-day old yeast cultures. Instead, compounds that elicited unique responses in D. melanogaster from 1-day old cultures (ethyl octanoate and ethyl hexanoate) were added, given the reportedly large overlap in olfactory sensitivities between the species. Adding uniquely detected compounds to either the base blend or the active yeast rendered blends equally attractive to D. suzukii flies (Figure 5). Attraction of D. suzukii flies to the base blend was not statistically different from that of minimal media (p = 0.99).