Ceropegia dolichophylla Schtlr. is a climbing herb that grows in forests from 500 to 1500 m a.s.l. Typically, it twines on other vegetation up to 1.5 m in height. Anthesis of individual flowers lasts for 1 day (opening in the morning; Heiduk et al., 2010), and the flowering season spans from July to early September (Zhou and Xie, pers. comm.; eFloras, 2015). Fruits can be found beginning at the end of September (Zhou and Xie, pers. comm.). In Ceropegia, the pollen is packed into discrete packages, the pollinia, two of which are connected via caudicles and the corpusculum to form a pollinarium. Due to the complicated pollination mechanism with fused and highly synorganized reproductive organs (gynostegium), where pollinia of a previously extracted pollinarium need to be inserted between “guide rails” (Vogel, 1961), C. dolichophylla depends on pollinators for successful reproduction.

Investigations were conducted in both China and Germany. In July and August 2012 bioassays were performed at the non-native location in Bayreuth, Germany, where previous studies took place on greenhouse plants (Heiduk et al., 2010). In the native range, plants of C. dolichophylla were studied in the Mt. Fanjing area in northeast Guizhou province, China (27°49′N-27°50′N, 108°44′E-108°46′E). This area bears vegetation characterized by broad-leaved evergreen forests of high diversity, in a subtropical, humid monsoon climate. C. dolichophylla plants were studied at two sites (henceforth Area 1 and Area 2) approximately 5 km apart from each other. At both sites, floral scent, flower visitors/pollinators, and data on pollen transfer efficiency (PTE) were collected in September 2013. Bioassays were performed in August 2012 and September 2013 at Area 1 and Area 2 and additionally in South China Botanical Garden (SCBG), Guangzhou (distance to Area 1 and Area 2: ca. 1000 km), where C. dolichophylla does not naturally occur.

Voucher specimens collected in China were deposited in the herbarium of the University of Bayreuth [Voucher/Accession: China: Guizhou, Tongren, Fanjing Mt., 874 m, A. Heiduk, I. Schäffler and Y. Hong, Sep. 2012 (UBT)].

To obtain information about pollination of C. dolichophylla in its native range in China, ca. 400 flowers from 12 individual plants (Area 1: 5 plants; daily collection from 7 to 13th September 2013; Area 2: 7 plants; collection on 8th September 2013) were picked in the evening and checked for trapped insects. Insects trapped inside the flowers were examined for pollinaria and number of pollinia carried, and only those carrying pollinaria/pollinia were designated as pollinators. To determine pollination success, 267 flowers from Area 1 and 58 flowers from Area 2 were examined in the field using a 10× hand lens, and the gynostegia were checked for pollinaria removal and pollinia insertion. For both areas, the mean number of removed pollinaria as well as the mean number of inserted pollinia was calculated. These data were used to calculate PTE separately for Area 1 and Area 2. PTE was calculated as the percentage of removed pollinia that were inserted between guide rails. Since each pollinarium consists of two pollinia, the mean number of inserted pollinia was divided by twice the mean number of removed pollinaria (Johnson et al., 2005; Coombs et al., 2009, 2011).

Floral volatiles were collected during daytime from newly opened flowers in situ using dynamic headspace methods (Dötterl et al., 2005b). Flowers were enclosed in a polyester oven bag (6 × 5 cm; Toppits®, Germany) for 10 min to allow accumulation of floral scent. Subsequently, volatiles were trapped by pulling the air from the bag through small adsorbent tubes (Varian Inc. ChromatoProbe quartz microvials; length: 15 mm, inner diameter: 2 mm) for 5 min using a membrane pump (G12/01 EB, Rietschle Thomas Inc., Puchheim, Germany; flow rate: 200 ml/min). The tubes contained 1.5 mg Tenax-TA (mesh 60–80) and 1.5 mg Carbotrap B (mesh 20–40; both Supelco) fixed by glass wool plugs.

In Area 1 seven samples were collected from seven different plants. Five of the samples were collected from a single flower, one sample was collected from two flowers, and one from three flowers (in cases where flowers grew closely together, they were enclosed in a single bag to avoid any injury of flowers). In Area 2 seven samples were collected from five plants (from two of these plants, two samples each were collected). Four of the samples were collected from single flowers, two samples from two flowers, and one sample from three flowers. At each location samples of the surrounding air were also collected as controls.

To obtain solutions of natural scent (19 in total) for bioassays and electrophysiological analyses (see below), floral scent from 17 individual flowers (Area 1, Fanjing Mt., China), and two individual flowers (Bayreuth, Germany) was collected for at least 4 h into large adsorbent tubes (glass capillaries; length: 8 cm, inner diameter: 2.5 mm) containing 15 mg Tenax-TA (mesh 60–80) and 15 mg Carbotrap B (mesh 20–40). The trapped volatiles were eluted with 70 μl of acetone (SupraSolv, Merck KgaA, Germany; following Dötterl et al., 2005a) per adsorbent tube. Subsequently, 2 × 5 and 1 × 7 samples collected in China, and both samples collected in Bayreuth, were combined to provide three samples (2 × 350 μl, 1 × 490 μl) from field plants (China) and one sample (140 μl) from greenhouse plants (Bayreuth), for further experiments (see below).

The volatiles trapped in small adsorbent tubes were analyzed by GC/MS using an automatic thermal desorption (TD) system (TD-20, Shimadzu, Japan) coupled to a Shimadzu GC/MS-QP2010 Ultra equipped with a ZB-5 fused silica column (5% phenyl polysiloxane; 60 m, i.d. 0.25 mm, film thickness 0.25 μm, Phenomenex). The samples were run with a split ratio of 1:1 and a constant helium carrier gas flow of 1.5 ml/min. The GC oven temperature started at 40°C, then increased by 6°C/min to 250°C and held for 1 min. The MS interface worked at 250°C. Mass spectra were taken at 70 eV (EI mode) from m/z 30 to 350. GC/MS data were processed using the GCMSolution package, Version 2.72 (Shimadzu Corporation 2012).

The solvated scent samples were analyzed by GC/MS using a Shimadzu GCMS-QP2010 Ultra equipped with an AOC-20i auto injector (Shimadzu, Tokyo, Japan) and again a ZB-5 fused silica column (5% phenyl polysiloxane; 30 m long, inner diameter 0.32 mm, film thickness 0.25 μm, Phenomenex). One μl of the samples was injected (injection temperature: 220°C; split ratio: 1:1), and the column flow (carrier gas: helium) was set at 3 ml/min. The GC oven temperature was held at 40°C for 1 min, then increased by 10°C/min to 220°C and held for 2 min. The MS interface worked at 220°C. Mass spectra were again taken at 70 eV (in EI mode) from m/z 30 to 350 and data processed as described above.

Identification of the compounds was carried out using the NIST 11, Wiley 9, FFNSC 2, Adams (2007) databases, the database available in MassFinder 3, and published plotted spectra (Francke et al., 1981; Bergström et al., 1982; Francke and Kitching, 2001). Structures of several compounds were confirmed by comparing mass spectra and retention times with those of synthetic reference samples. The assignment of 8-methyl-2-propyl-1,7-dioxaspiro[5.5]undec-3-ene, was based on the mass spectrum of the natural product and the general fragmentation pattern of spiroacetals (Francke and Kitching, 2001).

Double bond positions of alkenes were determined by reaction with dimethyl disulfide (DMDS) (Buser et al., 1983) and subsequent separation of the adducts on a 30 m × 0.25 mm i.d. 0.25 μm film thickness HP5-MS fused silica capillary column (Agilent Technologies, Santa Clara, CA, USA), starting at 60°C for 3 min, increased at a rate of 3°C/min to 300°C, held for 70 min.

Total scent emission was estimated by injecting known amounts of monoterpenoids, aromatics, and aliphatics (added to small adsorbent tubes). The mean response of these compounds (mean peak area) was used to determine the total amount of each compound extracted from the small adsorbent tubes (Dötterl et al., 2005b).

To screen for quantitative differences in absolute amounts of scent between Area 1 and Area 2, the total amount of scent per sample and flower was compared between areas using a t-test (StatSoft Inc., 2005). Normality was tested by Shapiro-Wilk test and homogeneity of variances by Hartley's test (StatSoft Inc., 2005).

To screen for semi-quantitative (percentage amount contributed per compound) differences in scent among samples of plants from Area 1 and plants from Area 2, the Bray-Curtis (BC) similarity index was calculated using Primer 6.1.11, including the add-on package Permanova + 1.0.1 (Clarke and Gorley, 2006; Anderson et al., 2008). If more than one sample was taken from the same individual plant, the mean scent composition was calculated and used for analyses. Based on the BC matrix a PERMANOVA (Factor: Area; 10,000 permutations) was performed using the same software package to test for an Area effect.

The scent components from C. dolichophylla flowers that were perceived by flower visitors/pollinators, were identified by gas chromatography coupled to electroantennographic detection (GC/EAD) and GC/MS (see above). Altogether 34 GC/EAD measurements with 14 flies from China and seven measurements with five flies from Bayreuth, Germany, were performed. The five flies (female Desmometopa sordida) from Bayreuth were collected from flowers of C. dolichophylla greenhouse plants. Four (one Oscinella frit, one Desmometopa varipalpis, two Neophyllomyza sp.) of the 14 Chinese flies were collected at SCBG while feeding on dead honey bees. The other 10 flies were attracted during bioassays performed in China with synthetic compounds of the C. dolichophylla flower scent (see bioassays). All flies were kept separately in Eppendorf® tubes (1.5 ml) with a piece of humid paper towel and stored in the dark at 4°C until electrophysiological measurements were performed.

For measurements, the head of a fly was cut off at the base of the thorax, mounted between two electrodes filled with insect Ringer's solution (8.0 g/l NaCl, 0.4 g/l KCl, 0.4 g/l CaCl₂) and connected to silver wires. The reference electrode was placed in contact with the cutting surface of the head while the recording electrode was brought into contact with the tip of the funiculus (cf. first flagellomere) of an antenna.

For measurements we either used a Carlo Erba Vega 6000 Series 2 (Rodano, Italy) or an Agilent 7890A (Santa Clara, California, USA) gas chromatograph, both equipped with a flame ionization detector (FID) and an EAD setup (heated transfer line, 2-channel USB acquisition controller) provided by Syntech (Kirchzarten, Germany). For each measurement 1 μl of an acetone solution of the C. dolichophylla scent was injected (injector temperature at 250°C) in splitless mode at 40°C oven temperature. The oven of both systems was heated at a rate of 10°C/min to 220°C, and the split vent was opened 0.5 min after injection. A Zebron ZB-5 column was used for analysis (5% phenyl polysiloxane; 30 m × 0.32 mm i.d. film thickness 0.25 μm, Phenomenex) in both GCs. The column of the Carlo Erba GC was split at the end by the four-arm flow splitter GRAPHPACK 3D/2 (Gerstel, Mühlheim, Germany) into two deactivated capillaries (length 50 cm × 0.32 mm i.d.) leading to the FID and to the EAD setup. Nitrogen was introduced as a make-up gas through the fourth arm of the splitter. The column of the Agilent GC was split at the end by a μFlow splitter (Gerstel, Mühlheim, Germany) into two deactivated capillaries leading to the FID (2 m × 0.15 mm i.d.) and EAD (1 m × 0.2 mm i.d.) setup. In both systems the outlet of the EAD was placed in a cleaned and humidified airflow directed over the fly antenna. Acetone solutions of the scent of C. dolichophylla were tested on antennae of five female D. sordida (3 × 1 and 2 × 2 runs per specimen), one female D. sp. nr. sordida (1 × 3 runs), two female D. varipalpis (5 and 3 runs), six female Neophyllomyza sp. (2 × 1, 2 × 2, and 2 × 3 runs), three female N. leanderi (1, 2, and 3 runs), one female Conioscinella sp. (2 runs), and one female Oscinella frit (3 runs). After the measurements, head and body of each fly were stored in a 4% solution of glycerin in ethanol (99.8%) for identification of genus and/or species.

Racemic spiroacetals were synthesized according to established methods (Phillips et al., 1980; Jacobsen et al., 1982; Doubský et al., 2004). To prepare (2S,6R,8S)-8-methyl-2-propyl-1,7-dioxaspiro[5.5]undecane (S4a) (Figure 1), commercially available 2-ethoxycarbonylcyclopentanone (a) was alkylated to yield the disubstituted cyclopentanone b, which after acidic hydrolysis was decarboxylated to produce 2-propylcyclopentanone (c). Baeyer-Villiger oxidation of c produced the racemic lactone d. Alkynylation of (2S)-2-methyloxirane (e) using lithium acetylide yielded (2S)-4-pentyne-2-ol (f) which was benzylated to g. Racemic d and the anion of g were linked to form the intermediate h (not isolated) which upon hydrogenation furnished the (8S)-configured spiroacetal S4 as a mixture of the three stereoisomers S4a-c (see Figure 2A).

To a solution of 2.67 g (15.4 mmol) (S)-benzyloxypent-4-yne (g) in 30 mL abs. THF, cooled to −78°C, were dropwise added 7.50 mL (18.8 mmol) of a 2.5 M-solution of n-BuLi in hexane. After stirring for 90 min at −78°C, 2.00 mL (16.2 mmol) BF₃•Et₂O, dissolved in 20 mL abs. THF, were slowly added, followed by a solution of 2.39 g (16.8 mmol) 6-propyltetrahydro-2H-pyran-2-one (d) in 10 mL THF. Over a period of 3 h, the mixture was warmed to room temperature, and the reaction was quenched by the addition of a mixture of 20 ml water, 20 mL diethyl ether, and ammonium chloride/ammonia (2:1). After separation of the layers, the aqueous phase was extracted 4 times with 20 mL portions of diethyl ether. The combined organic solutions were washed with brine and dried over magnesium sulfate. Filtration over silica and removal of the solvent in vacuo yielded 4.35 g of crude h. This was dissolved in 10 mL methanol and hydrogenated for 21 h at 20 bar, using 5% Pd-C catalyst. After removal of the catalyst by filtration over silica, the crude product (see Figure 2A) was purified by chromatography on silica using a 50:1-mixture of pentane and diethyl ether. A further chromatographic step using benzene as the eluent yielded 113 mg (0.53 mmol, 3.5%) highly pure S4a (see Figure 2B).

NMR-Spectra were run on a Bruker (Billerica, MA, USA) AMX-400 instrument. For the numbering of structural elements see Figure 2B.

¹H-NMR, based on ¹H-¹H-COSY, HSQC, HMBC (400 MHz, CDCl₃): δ [ppm] = 0.92 (t, ³ J_H14−H13 = 7.1 Hz, 3H, CH₃ C14), 1.10–1.20/1.51–1.59 (2m, 2H, CH₂ C3 ax/eq), 1.12–1.23/1.54–1.61 (2m, 2H, CH₂ C9 ax/eq), 1.13 (d, ³J_H15−H8 = 6.3 Hz, 3H, CH₃ C15), 1.30–1.39/1.46–1.55 (2m, 2H, CH₂ C13), 1.33–1.43/1.56–1.64 (2m, 4H, CH₂ C5 C11 ax/eq), 1.31–1.40/1.43–1.51 (2m, 2H, CH₂ C12), 1.50–1.58/1.88 (m/ddddd, ³J_H−H = 13.9, 13.2, 13.2, 4.0, 4.0 Hz, 4H, 2 × CH₂ C4 C10 eq/ax), 3.54 (dddd, ³J_H−H = 11.0, 8.7, 4.0, 2.0 Hz, 1H, CH C2ax), 3.70 (dqd, ³J_H−H = 11.4, 6.3, 2.0 Hz, 1H, CH C8ax).

¹³C-NMR, based on HSQC, HMBC (126 MHz, CDCl₃): δ [ppm] = 14.39 (q, C14), 19.09 (t, C10), 19.16 (t, C13), 19.28 (t, C4), 21.99 (q, C15), 31.55 (t, C3), 33.01 (t, C9), 35.50/35.69 (2t, C5 C11), 38.89 (t, C12), 65.21 (d, C8), 68.85 (d, C2), 96.12 (s, C6).

The 70 eV mass spectrum of S4a was identical to the plotted one published earlier (Francke et al., 1981).

Due to the double anomeric effect (Deslongchamps et al., 1981) and the equatorial orientation of both alkyl substituents (Francke et al., 1980), S4a was the highly dominating stereoisomer. The two thermodynamically less stable (E,Z)-isomers S4b and S4c were formed as by-products, whereas the diequatorially linked highly unstable 4d was not obtained in detectable amounts. The same synthetic approach, but using racemic 2-methyloxirane yielded a mixture of all eight possible stereoisomers of 8-methyl-2-propyl-1,7-dioxaspiro[5.5]undecane (rac-S4), dominated by the racemate of the (2E,8E)-isomer. Enantioselective gas chromatography, employing a home-made 30 m × 0.25 mm i.d. fused silica capillary coated with a 1:1-mixture of OV-1701 and heptakis-[2,3-di-O-methyl-6-O-(tert-butyldimethylsilyl)]-β-cyclodextrin as the stationary phase, separated the enantiomers well. Hydrogen as the carrier gas at a constant oven temperature of 90°C produced an α-value of 1.22 = (ret. time 2S,6R,8S):(ret. time 2R,6S,8R).

The attractiveness of acetone solutions of the scent of C. dolichophylla (see above) was tested in China (for experiments in Germany see Heiduk et al., 2010). Samples were assayed on six different days (4× Fanjing Mt. Area 1, 2× SCBG). Each time a glass vial containing an acetone solution of natural scent (see before) was offered against a similar glass vial filled with pure acetone (control). Within a distance of 30 cm to each other the vials were tucked into the ground and offered for at least 30 min and up to 60 min. The amount of scent available in a sample was sufficient for two assays. The attractiveness of single EAD-active compounds was tested both in China (Fanjing Mt., SCBG) and in Germany (Bayreuth). The compounds were chosen based on preliminary assays with fractions of the complete flower scent. We used the major EAD-active (see below) volatile compounds N-(3-methylbutyl)acetamide (1), stereochemically pure (2S,6R,8S)-8-methyl-2-propyl-1,7-dioxaspiro[5.5]undecane (S4a) and its racemate (including ca. 5% of the three other stereoisomers, which slightly differs from the natural proportions, see 2.8). Racemic (E,E)-2,8-diethyl-1,7-dioxaspiro[5.5]undecane was tested in addition. Since this compound eluted shortly after S4a, it was potentially considered also EAD-active. (E,E)-8-Methyl-2-propyl-1,7-dioxaspiro[5.5]undecane was also used as a racemic mixture. The substances were diluted (dilution: 10⁻³; v/v; final volume: 400–600 μl) in acetone (SupraSolv, Merck KgaA, Germany) and offered in glass vials similar to those used for tests with natural samples. As determined by dynamic headspace and GC/MS for S4a, the amount of scent released from the vials resembled the amount of scent released from single flowers as described by Heiduk et al. (2010).

We tested (a) the single components, 1, S5a, and S4/S4a, (b) the three possible two-component mixtures, (c) the three component mixture, and (d) all three components against each other. When using mixtures, proportions were adapted to the ratios found in C. dolichophylla flowers as indicated by dynamic headspace and GC/MS analysis. Vials containing the samples were tucked into the ground with a distance of 20 cm to each other and offered for at least 40 min and up to 60 min. In each bioassay a glass vial with pure acetone was offered as the control. Approaching flies showed a characteristic zig-zag flight with abrupt landing. They were caught when arriving within a maximum distance of 10 cm to the vial containing the sample. Due to their fast and frantic behavior, not all approaching flies could be caught.

The flowers of C. dolichophylla collected in China altogether contained 119 dipteran individuals, 107 thereof were collected in Area 1 and the other 12 in Area 2. The flies belonged to the families Milichiidae, Chloropidae, Phoridae, and to taxa of lower Diptera (Table 1). Chloropidae were only present in flowers of Area 1.

Different taxa from lower Diptera were the most abundant visitors, however, they did not carry pollinaria - nor did the phorid flies. Milichiids were the second most abundant group, and many of these flies carried pollinaria (60.5%). They were determined as Desmometopa microps (13 females, 10 males; 12 with pollinaria), D. varipalpis (one female), Neophyllomyza sp. (9 females, 7 with pollinaria) and N. leanderi (10 females, 7 with pollinaria). With seven individuals, chloropid flies were not very abundant, however, four (57%) of them carried pollinaria. Chloropids were determined as Conioscinella sp. (2 females, 1 with pollinarium), Polyodaspis sp. (1 male, 2 females, all with pollinaria), and Tricimba spp. (2 females of different species, both with pollinaria).

Among the 267 flowers analyzed in Area 1, 51 pollinaria were found to be removed and 4 pollinia inserted, resulting in a PTE of 4%. The percentage of flowers with removed pollinaria was 13%, and 1% of flowers had pollinia inserted. On average and per flower, 0.19 pollinaria were removed and 0.02 pollinia inserted. Of the 58 flowers collected in Area 2 altogether 47 pollinaria were removed and 2 pollinia inserted, yielding a PTE of 2%. The percentage of flowers with removed pollinaria was 28%, and 3% of flowers had pollinia inserted. On average 0.81 pollinaria were removed per flower, whereas 0.03 pollinia were inserted.

In the floral scent of C. dolichophylla 53 different components were detected: 14 spiroacetals (40.3%), 6 alkanes (16.8%), 4 alkenes (9.4%), 4 other aliphatics (0.1%), one nitrogen containing compound (7.4%), and 23 unknown compounds (1.4%) (Table 2, Figure 3). The most abundant scent components were (2S,6R,8S)-8-methyl-2-propyl-1,7-dioxaspiro[5.5]undecane (S4a) (27%), tridecane (15%) and N-(3-methylbutyl)acetamide (1) (7,4%), contributing 49% to the total scent. Apart from (E,E)-2,8-diethyl-1,7-dioxaspiro[5.5]undecane (S5a) (8%) all other compounds did not exceed 5%.

The total amount of scent per flower (ng/15 min; 10 min accumulation + 5 min sampling) was highly variable and ranged from 15 to 67 ng in Area 1 (median: 32 ng), and 16 to 117 ng in Area 2 (median: 80 ng), respectively. There was no significant difference in total amounts of scent between Area 1 and Area 2 [t₍₁₀₎ = −2.0, p = 0.074]. Furthermore, no significant differences in scent profiles (relative scent composition) were found [PERMANOVA: Pseudo-F_{(1, 10)}: 1.4, p = 0.198] between plants of Area 1 and Area 2.

A comparison of scent from field and greenhouse (Heiduk et al., 2010) plants revealed eight compounds exclusively present in field plants and two compounds only present in greenhouse plants (Table 2).

Only two (Desmometopa sp. nr. sordida and D. varipalpis) of the 14 flies from China and three of the five female D. sordida from Bayreuth gave obvious antennal signals. Antennae of the other flies had too much noise in their signals and, thus, were not included in the analysis.

Of the 53 components found in scent samples of C. dolichophylla collected in the field, 22 compounds (Table 2) were electrophysiologically active in the two species tested. The antenna of D. varipalpis responded to 19 compounds, most strongly to N-(3-methylbutyl)acetamide (1), (2S,6R,8S)-8-methy-2-propyl-1,7-dioxaspiro[5.5]undecane (S4a) + (E,E)-2,8-diethyl-1,7-dioxaspiro[5.5]undecane (S5a), (E,Z)-2,8-diethyl-1,7-dioxaspiro[5.5]undecane (S5b), and an unknown compound (Figure 4A). Of the 19 components with EAD-activity in D. varipalpis, the antenna of D. sp. nr. sordida responded only to 1 and to 6,9-pentadecadiene (8) as well as to 6- + 7-pentadecene (9) (Figure 4A).

All three female D. sordida from Bayreuth responded to the same seven compounds (Figure 4B). Five of them were also active on flies from China and scent samples from plants collected in the field. In each run, 1 and/or S4a + S5a elicited the strongest antennal responses (Figure 4).

During bioassays in China (Fanjing Mt. and SCBG) and Germany (Bayreuth) only Diptera were attracted, and no fly responded to the negative controls (Table 1).

Samples of natural flower scent were tested in China (for experiments in Germany see Heiduk et al., 2010) at Fanjing Mt. Area 1 (4 replicates), and at SCBG (2 replicates), and only at SCBG were flies attracted. The first flies approached in zig-zag flight within a minute after opening the sample vial. Altogether, 12 of the approaching flies (seven female Neophyllomyza sp., five female N. leanderi) were caught, all of them from taxa that occur as natural pollinators of C. dolichophylla.

In bioassays with synthetic samples (China and Germany), altogether 137 attracted flies were collected, the majority of them in Bayreuth (N = 71) and SCBG (N = 53), and a few (N = 13) in Fanjing Mt. Their behavior in approaching the sample vials was identical to that elicited by samples of natural flower scent. Milichiidae were the most numerous attracted flies (97%). They represented eight different species, among them three species pollinating C. dolichophylla. Two different species of Chloropidae were represented by four flies, and one of these species was identified as a pollinator of C. dolichophylla. The remaining flies were from non-pollinating taxa.

Flies were attracted mostly by (2S,6R,8S)-8-methyl-2-propyl-1,7-dioxaspiro[5.5]undecane (S4a) and mixtures containing this spiroacetal. Pollinating species were attracted to S4a, N-(3-methylbutyl)acetamide (1), to the mixture of S4a + 1, and to the mixture of (E,E)-2,8-diethyl-1,7-dioxaspiro[5.5]undecane (S5a) + 1. Non-pollinating species responded to the same lures except the mixture of S5a + 1.

In the four-choice assays which offered 1, S5a, S4a, and acetone, all flies responded to S4a. Overall, most flies were attracted to S4a, and the majority of them were the pollinating milichiid D. microps. The majority of non-pollinating species were attracted to the mixture of S4a + 1.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.