Sex pheromone of a coccoid insect with sexual and asexual lineages: fate of an ancestrally essential sexual signal in parthenogenetic females

Abstract

Sex pheromones play a central role in intersexual communication for reproduction in many organisms. Particularly in insects, reproductive isolation that leads to speciation is often achieved by shifts of pheromone chemistries. However, the divergence and evolution of pheromones remain largely unknown. This study reveals a unique evolutionary consequence for terpenoid pheromones in coccoid insects. Coccoids, such as mealybugs, show clear sexual dimorphism: males are dwarf and short-lived, whereas females are wingless and almost immobile. Female pheromones are therefore indispensable for males to navigate for sexual reproduction, but some females can reproduce asexually. Interestingly, a derived asexual lineage that reproduces by parthenogenesis coexists with its ancestral lineage that reproduces sexually in a population of the pineapple mealybug, Dysmicoccus brevipes. Here, we isolated, characterized and synthesized a novel monoterpene, (−)-(anti-1,2-dimethyl-3-methylenecyclopentyl)acetaldehyde, as a pheromone of the sexual females of D. brevipes. This monoterpene aldehyde, with an irregular linkage of isoprene units, is notable, because all mealybug pheromones previously reported are carboxylic esters of terpenols. This compound was, however, never produced by the asexual females. As a consequence of acquiring parthenogenetic reproduction, the asexual females appear to have abandoned the production of the sex pheromone, which had been essential to attracting males in their ancestors.

1. Introduction

Sex pheromones, which mediate chemical communications for copulation, are ubiquitous among animals as an essential secondary sexual characteristic [1–5]. In some insect taxa, females produce and emit volatile pheromones to attract males from a long distance. These pheromones generally elicit drastic behavioural or physiological responses of conspecific males even in very small amounts [1–3]. Because pheromonal communication is the primary step in the reproduction of many insects [6], the divergence of pheromone systems should have profoundly affected premating reproductive isolation and speciation in insects, which constitute more than 58% of described species on the Earth [7]. Thus, pheromones have been a central topic for chemical ecologists since the dawn of this interdisciplinary field [8].

Typical examples of long-range volatile pheromones are found in moths [9]. Although moths are one of the largest groups of insects and comprise approximately 160 000 described species, their pheromone chemicals, which support species-specific mate finding, are, unexpectedly, rather simple with limited structural classes: most are fatty alcohols or their derivatives with an unbranched chain of an even number (mostly 12, 14 or 16) of carbons [10]. These components are often shared with other species, but combinations and blend ratios confer species specificity to the signal, avoiding the cost of encountering the wrong mates [11–13]. Recent studies have uncovered genetic and biochemical mechanisms for shifts in the pheromone blends that drive the diversification of pheromonal communication [14–17]. However, the evolution and diversity of pheromone chemistry remain to be illustrated in other insects with different pheromone chemistries.

Coccoids, including scales and mealybugs (Hemiptera: Coccoidea), which are small sap-sucking insects related to aphids and whiteflies, also use volatile pheromones with long-range attraction [18]. Adult females hardly ever move, lacking wings and often even legs, and instead produce protective secretions such as shields (figure 1), and as adults can live for several months [19]. In contrast, adult males are winged and mobile, but they are tiny and fragile (figure 1) and have a limited lifespan of a few days at most [19]. Sex pheromones emitted by females are thus essential and are considered to be under strong selection pressure to facilitate mating and reproduction by serving as a key navigation tool for males [20].

Figure 1.

Male and female adults of the pineapple mealybug, Dysmicoccus brevipes, with completely different appearance and biology. (Online version in colour.)

Most coccoid pheromones identified so far are terpenes with unique structures. Currently, terpenoid pheromones have been isolated and identified in seven scale insects and 16 mealybugs, all of them carboxylic esters of sesquiterpene, monoterpene or hemiterpene alcohols [10]. Monoterpene pheromones of mealybugs commonly have an unusual structural feature: most monoterpenes are composed of two isoprene units coupled by a ‘head-to-tail’ (1 → 4′) connection [21], but the two units of the alcohol moieties of mealybug pheromones are linked with irregular non-head-to-tail connections [22–24]. Despite this common motif, all of these pheromones are species specific [22,25]. Unlike moths, which often share the same pheromone chemicals with other species, mealybugs are therefore considered to have evolved a characteristic biosynthetic pathway in each species that generates specialized terpenoids [22–25]. Thus, the sex pheromone systems of mealybugs provide a good opportunity to study chemical communications supported by monoterpene-based structural diversity.

Although pheromone-mediated copulation is indispensable for reproduction by most coccoids, females of some species can reproduce asexually without copulation [26]. Asexual reproduction is assumed to confer advantages over sexual reproduction, which includes a ‘cost of males’ [27–32]. The small size and short lifespan of coccoid males have promoted the evolution of reproductive systems that depend less on males [26]. In fact, males are either very rare or unknown in many coccoids, particularly in mealybug species, 10% of which reproduce either partially or completely by parthenogenesis [33], by which embryos develop without fertilization.

The pineapple mealybug, Dysmicoccus brevipes (Cockerell), is a known parthenogenetic species that carries pineapple wilt-associated viruses which reduce pineapple yields [34]. Although many previous studies [35–37] in a variety of areas indicated that this species reproduces asexually by obligate apomictic thelytokous parthenogenesis, we recently found simultaneous and sympatric sexual and asexual reproduction in a Japanese population [38]. Molecular phylogenetic analyses indicated that these mealybugs are genetically isolated and are likely to represent diverged lineages, yet they are estimated to have diverged in relatively recent times (about 1.3 million years ago) and are very closely related with no diagnostic morphological features [38]. Thus, the sexual and asexual lineages of D. brevipes offer a unique example of the evolutionary consequence of a specialized terpenoid pheromone that is an ‘aphrodisiac fragrance’ essential for sexual females but unnecessary for their asexual sisters.

We isolated and characterized the sex pheromone from airborne volatiles emitted by virgin females of the sexual lineage by means of gas chromatography (GC)–mass spectrometry (MS) and nuclear magnetic resonance (NMR) spectrometry. We then synthesized the compound to confirm its structure and activity. Finally, we quantified and compared diel and daily production of the pheromone in the sexual and asexual mealybugs. Based on these findings, we challenged to elucidate an evolutionary consequence of a formerly essential sexual signal after abandonment of sexuality. Furthermore, we deduced an evolutionary scenario of mealybug pheromones by means of phylogenetic comparative analyses; the pheromone systems and structures of mealybugs are apparently discordant with their phylogenetic relationships and appear to have expanded under saltatorial evolutionary processes.

2. Material and methods

2.1. Analytical instruments

GC–MS analyses were performed on an HP5890 gas chromatograph (Hewlett-Packard, Avondale, PA) equipped with an SX-102A mass spectrometer (JEOL, Tokyo, Japan). The temperatures of the interface and the ion source were 210 and 220°C, respectively. A polar column (DB-23; 0.25 mm internal diameter × 30 m length, 0.15 µm film thickness; J&W Scientific, Folsom, CA) and a non-polar column (DB-1; 0.25 mm internal diameter × 30 m length, 0.25 µm film thickness; J&W Scientific) were used at a constant flow rate (1.0 ml min⁻¹) of helium as the carrier gas. The column oven temperature was maintained at 60°C for 1 min, raised to 220°C at 8 °C min⁻¹, and held there for 5 min. Mass spectra were obtained in electron-impact (EI) mode at 70 eV. For certain runs, n-alkanes were added as references for the calculation of retention indices (Kováts indices, KI). The natural pheromone was quantified by an Agilent 6890N gas chromatograph (Agilent Technologies Inc., Wilmington, DE) equipped with a flame ionization detection (FID) system maintained at 220°C, using a calibration curve generated by 0.05 to 5 ppm solutions of the synthetic pheromone (described below). The coupled GC–electroanntenographic detection (EAD) was performed on an HP5890 gas chromatograph equipped with a biological amplifier for electrophysiological responses (AB-651 J; Nihon Kohden Co., Tokyo, Japan). Six replicates of GC-EAD analyses were performed and confirmed to show consistent responses. The other conditions and equipment for GC were the same as above. NMR spectra were obtained at 30°C with a JNM-A600 spectrometer (JEOL; ¹H: 600.05 MHz; ¹³C: 150.80 MHz). C₆D₆ (min. 99.96%; Sigma-Aldrich, St. Louis, MO) was used as the solvent. Fourier transform infrared (FT-IR) spectra were obtained with an FT/IR-410 spectrometer (Jasco, Tokyo, Japan). Polarimetry was performed with a P-1020 polarimeter (Jasco).

2.2. Collection of volatiles from insects

Sexual and asexual lineages of the mealybug were collected in a pineapple field at the Okinawa Prefectural Agricultural Research Center (OPARC: 26.6° N, 127.9° E; Nago City, Okinawa Prefecture, Japan) on 7 March 2012 [38]. Mealybugs of each lineage were separately reared on germinated broad beans (Vicia faba) in a rearing room (16 h light, 8 h dark; 23°C; 50% relative humidity).

For large-scale volatile collection, we placed approximately 200 newly eclosed virgin females of the sexual lineage into a 1 l glass jar with germinated broad beans. Ambient air passed through activated charcoal (200 g) was pulled over the females and then passed through an adsorbent (1.0 g of HayeSep Q, 60/80 mesh, Alltech, Deerfield, IL) at 1 l min⁻¹ using a vacuum pump. Every 3–4 days, the volatiles were extracted from the adsorbent with 15 ml of pentane, concentrated in an evaporator at room temperature, and stored at –20°C. Over six weeks, volatiles from approximately 672 000 female day equivalents were prepared in a total of 80 rounds of collections.

For small-scale volatile collection to quantify and compare sex pheromone production, we placed virgin females into a 20 ml glass vial with a germinated broad bean, and air passed through activated charcoal (15 g) was pulled over them and then passed through an adsorbent (50 mg of HayeSep Q) at 0.1 l min⁻¹ using a vacuum pump. For investigation of diel rhythms of pheromone production, we prepared 17 vials (sexual lineage) and 16 vials (asexual lineage) containing 10 14 day old virgin females and sampled their volatiles every 2 h over a 24 h period. For investigation of daily fluctuations, 12 vials, each with one newly eclosed virgin sexual female, one newly eclosed mated sexual female (given 1 day to mate in the vial with a newly emerged male), or one newly eclosed asexual female and a germinated broad bean, were prepared, and their volatiles for 1 day were sampled six times over two weeks. Samples were eluted from the adsorbent, using 0.1 ml of hexane containing 100 ng of nonyl acetate (Sigma-Aldrich) as an internal standard.

Voucher specimens, partial mitochondrial DNA sequences of which are deposited in GenBank/DDBJ/EMBL with accession numbers LC121500-LC121505, are stored at the National Agriculture and Food Research Organization (Tsukuba, Japan).

2.3. Preparative liquid chromatography

Column chromatography was carried out on silica gel (0.2 g, Wakogel C-200; Wako Pure Chemicals, Osaka, Japan) by the class-separating method for lipids. The column was successively eluted with 2 ml each of 0%, 5%, 15% and 50% diethyl ether in pentane. High-performance liquid chromatography (HPLC) was performed with a Hewlett-Packard series 1050 HPLC system using a silica gel column (4.6 mm internal diameter × 250 mm length, 5 µm particle size; Inertsil SIL, GL Science Inc., Tokyo, Japan). The solvent for elution was 5% ether in pentane at a flow rate of 1 ml min⁻¹, and the eluate was monitored by a UV detector (λ = 210 nm).

2.4. Preparative gas chromatography

Preparative GC was carried out with an Agilent 6890N gas chromatograph with a flame ionization detector (220°C) and a preparative fraction collection system (Gerstel GmbH & Co. KG, Mülheim an der Ruhr, Germany) cooled by a dry ice–acetone bath. A polar column (TC-FFAP; 0.53 mm internal diameter × 15 m length, 1 µm film thickness; GL Science Inc.) was installed, and the GC oven temperature was held at 40°C for 5 min and then increased to 200°C at 8°C min⁻¹. A fully programmable injector (Optic 3, ATAS GL International, Eindhoven, The Netherlands) maintained at 40°C for 10 s and then raised to 200°C at 5°C s⁻¹ was used for large volume injection; 10 µl of the sample was injected in splitless mode for 4 min. Helium was used as the carrier gas at 5 ml min⁻¹.

2.5. Bioassay for evaluating attractiveness

The attractiveness of each chromatographic fraction to males was assayed as follows [23,24]: 10 newly emerged adult males were transferred into a glass dish (9 cm diameter, 2 cm height) on a fine paintbrush. Then, a sample (10 female day equivalents) dissolved in 5 µl of pentane was loaded onto a piece of filter paper (1 cm²; Toyo Roshi Co. Ltd., Tokyo, Japan) and placed at the centre of the dish. The number of males that mounted the sample and tried copulating with it within 5 min as a percentage of the number released was used as an index of attractiveness. Three replicates were performed for each sample in the assay in the morning, when males showed active mate-finding behaviour, and the scores were averaged.

The attractiveness of the natural pheromone and the synthetic pheromone (§2.7) was tested by a trap bioassay in a pineapple field (approx. 0.5 ha) at OPARC. Pheromones (0.1 mg) dissolved in 100 µl of hexane were impregnated into red rubber septa (8 mm outside diameter × 19 mm height; Sigma-Aldrich) overnight at room temperature while the solvent evaporated. White delta-traps with 24 × 30 cm sticky boards (SE-trap; Sankei Chemical Co., Kagoshima, Japan) baited with the septa were placed 0.6 m above the ground at 5 m intervals. Two sets of traps with five treatments—0.1 mg of the natural pheromone, 0.1 mg of synthetic (±)-1, 0.1 mg of synthetic (±)-2, a mixture of each 0.1 mg of (±)-1 and (±)-2 and control (solvent only)—were prepared (§2.7). They were checked once a week, when their locations were changed, four times, from 16 July to 15 August 2015. Data were analysed by generalized linear mixed models (GLMMs) with a Poisson error distribution to assess the effects of treatments and the effects of trapping date, and random effects to account for overdispersion of scores among the traps. The calculations were performed by the glmer function of the ‘lme4’ package in R software v. 3.2.5 [39]. Multiple comparisons were performed with Bonferroni correction.

2.6. Microchemical reactions of the natural pheromone

For reduction of a possible carbonyl group, the pheromone compound (0.1 mg) was dissolved in 1 ml of 1 mM NaBH₄ in ethanol; the mixture was stirred for 2 h at room temperature, poured into 2 ml of H₂O and extracted three times with pentane (2 ml). The combined extracts were dried over sodium sulfate and concentrated in vacuum with cooling, and an aliquot was analysed by GC–MS as above. For hydrogenation of a possible olefin structure, the pheromone compound (0.1 mg) was dissolved in 0.2 ml of ethanol; the solution was stirred in the presence of platinum black catalyst (Wako) in an H₂ atmosphere for 10 min and then centrifuged, and an aliquot of the supernatant was analysed by GC–MS analysis. The resulting saturated product was further reduced by using NaBH₄ and was then analysed by GC–MS as described above.

2.7. Preparation of synthetic pheromone

A synthetic 1 : 2 stereoisomeric mixture of the deduced pheromone compound, (±)-1 and (±)-2, was synthesized by the previously reported method [40]. These diastereomers were separated with a Hewlett-Packard series 1050 HPLC system using an Inertsil SIL silica gel column (4.6 mm internal diameter × 250 mm length, 5 µm particle size; GL Science, Inc.) impregnated with silver nitrate. The solvent for elution was 5% ether in pentane at a flow rate of 1 ml min⁻¹, and the eluate was monitored by a UV detector (λ = 210 nm) to prepare pure (±)-1 and (±)-2.

2.8. Phylogenetic analysis

To infer a scenario for diversification of mealybug pheromone chemicals, we constructed a maximum-likelihood (ML) phylogenetic tree based on partial sequences (more than 1000 bp) of mitochondrial cytochrome oxidase subunit I (CO1) genes deposited in GenBank/ENA/DDBJ. We compared 10 taxa from four genera (Dysmicoccus, Planococcus, Pseudococcus and Phenacoccus). The sequences were aligned in ClustalX software [41]; the final alignment was inspected and corrected manually, and only unambiguous nucleotide sites without gaps were used for analyses. An ML tree with bootstrap values based on 1000 resamplings of the heuristic search was constructed using PAUP* and ModelTest [42] software with the TIM + I + G evolutionary model, selected by Akaike's information criterion with a correction for finite sample sizes. We used the sequence from Phenacoccus solenopsis, which belongs to a different subfamily (Phenacoccinae), as an outgroup to show the phylogenetic positions of D. brevipes and the other species (Pseudococcinae). The sequences were aligned, and the ML tree were deposited at TreeBASE with an accession ID 20533 (http://purl.org/phylo/treebase/phylows/study/TB2:S20533).

Three discrete characters of the pheromone structures were defined: functional group (acetate, butyrate, senecioate, 3-methyl-3-butenoate or aldehyde), carbon number of terpene moiety (10 or 9) and cyclic/acyclic structure (acyclic, cyclobutane or cyclopentane). Diversification of these characters was analysed by phylogenetic comparative methods [43–46]. Three macroevolutionary models, a model of rate constancy through time, a punctuational model and an accelerating/decelerating model, were fitted to patterns of the structural diversity of the mealybug pheromones with an assumption of an equal rate of character transitions and were compared by a likelihood ratio test approximated by a chi-squared distribution [46,47]. Then the phylogenetic signal—the degree to which similarity in trait values between species can be predicted on their relatedness [46]—was tested by calculating Pagel's lambda [46–50]. These calculations were performed in R software with the fitDiscrete function in the ‘geiger’ package [47]. Data of the asexual D. brevipes were excluded.

3. Results

3.1. Isolation of pheromone and elucidation of its structure

The crude extract of volatiles was fractionated, and the male-attractiveness of the 5% diethyl ether/pentane fraction (56.7% ± 3.3%; mean ± s.e.m.) was comparable to that of the crude extract (70.0% ± 5.8%). This fraction was further separated into eight fractions by HPLC (electronic supplementary material, figure S1). Males showed the strongest attraction toward the third fraction (93.3% ± 6.7%), eluted at a retention time of 7.5–8.0 min. Male attractions to the other HPLC fractions were less than 10.0% ± 5.8%. This was then separated into three fractions by preparative GC (electronic supplementary material, figure S1). Only the second fraction, a single compound eluted at a retention time of 7.5–7.9 min, showed significant attractiveness (83.3% ± 8.8%). No attractions were observed to the other GC fractions. The pheromone compound was therefore isolated. A total of approximately 2.1 mg was acquired.

In GC–MS analyses, the isolated compound appeared at KI = 1563 on a DB-23 column and at KI = 1123 on a DB-1 column. High-resolution EI MS data (m/z 152.1276) showed that the molecular formula was C₁₀H₁₆O (calculated m/z 152.1201), indicating the presence of three double bonds or rings. A characteristic mass fragment was observed at m/z 108 (figure 2a), which could be interpreted as [M⁺ – 44 (CH₃CHO)]. The hypothesized presence of a formyl group was supported by FT-IR spectroscopy showing strong absorption at 1721 cm⁻¹ (C=O) and weak absorption at 2731 cm⁻¹ (C–H). Moreover, the compound was easily converted to the corresponding alcohol, the EI MS spectra of which showed a characteristic ions at m/z 154 [M⁺ = 152 + 2 (H₂)] and m/z 136 [M⁺ − 18 (H₂O)], by NaBH₄ in microreduction (figure 2b), indicating an aldehyde.

Figure 2.

Mass spectra of (a) the natural pheromone compound, (b) an alcohol derivative reduced by NaBH₄, (c) the Pt-catalysed hydrogenated product and (d) the reduced and hydrogenated product. Broken arrows, hydrogenation; solid arrow, reduction.

In addition, the microhydrogenation product gave two GC peaks (4 : 3 ratio) with similar mass spectra, including the molecular ion at m/z 154 [M⁺ = 152 + 2 (H₂)], suggesting the presence of one C = C double bond and one ring structure (figure 2c). Because a pair of diastereoisomers in different amounts was generated during Pt-catalysed heterogeneous hydrogenation, which occurs with syn addition in favour of one side of the ring structure, we supposed the double bond to be either located in the ring or connected to the ring with other asymmetric carbons. FT-IR data showing moderate absorption at 3074 cm⁻¹, which is unique to compounds with a terminal olefin (>C = CH₂), indicated an exomethylene structure attached to a ring. The molecular ion was found at m/z 156 [M⁺ = 152 + 4 (2 × H₂)] in a very small signal following the reduction of carbonyl group of the hydrogenated products (figure 2d).

We used ¹H-NMR signals and ¹H–¹H correlation spectroscopy to determine the structure of the ring and the location of the double bonds. ¹H-NMR analysis showed a characteristic spectrum indicating one formyl proton, two olefinic protons, two sets of methyl protons, three sets of methylene protons and one methine proton (figure 3). The formyl proton was found at 9.46 ppm in a triplet signal with a correlation to the methylene protons (1.68–1.73 ppm; J = 2.4 Hz), which were not correlated with any other protons, indicating a substituted acetaldehyde structure connected with a quaternary carbon constituting a ring with a methyl group found at 0.86 ppm in a singlet. The other methyl group (0.70 ppm) was found in a doublet and was coupled with a methine proton (1.78 ppm), which was not correlated with any other protons and was indicated to be located between two quaternary carbons. Two olefin protons at 4.77 and 4.85 ppm were coupled only weakly (J = 1.8 Hz), indicating an exomethylene attached to a ring as hypothesized above. We therefore considered the other two sets of methylene protons at 1.10–1.59 ppm and 2.11–2.19 ppm to constitute a cyclopentane ring structure. Thus, according to the NMR analyses and the chemical data above, we considered the pheromone compound to be (1,2-dimethyl-3-methylenecyclopentyl) acetaldehyde. We used ¹³C NMR signals (12.0, 24.8, 28.9, 35.5, 43.3, 47.2, 50.4, 105.7, 155.7 and 201.4 ppm) and their correlations with ¹H NMR in heteronuclear multiple quantum correlation and heteronuclear multiple bond correlation to confirm the structure.

Figure 3.

¹H-NMR spectrum (600 MHz) of the pheromone. Chemical shifts are expressed in parts per million (ppm) by frequency relative to residual C₆H₆ (δ 7.15 ppm) in the solvent.

We synthesized a stereoisomeric mixture of the pheromone [(±)-1 and (±)-2] (figure 4) for further confirmation of its chemical structure and biological activity. The two synthetic diastereomers were eluted at retention times of 10.5 min [(±)-1] and 12.7 min [(±)-2] in HPLC with an AgNO₃-impregnated silica gel column and purified. The configurations of the two methyl groups of (±)-1 and (±)-2 were determined to be anti and syn, respectively, by nuclear Overhauser effect (NOE) spectroscopy; (±)-2 showed a clear NOE between the proton signals of the two methyl groups. The GC and HPLC retention times as well as the EI MS and NMR spectra of the natural pheromone and (±)-1 were identical. The natural pheromone displayed an optical rotation of [α]^26.3_D = –39.45 (c 0.0752, hexane), and is therefore considered to be (–)-1, whose absolute configuration we are currently trying to determine. In the field trap experiment, the males showed strong and selective responses to 1; the natural pheromone [(–)-1] and (±)-1 as well as a 1 : 1 mixture of (±)-1 and (±)-2 showed equivalent significant activity (GLMM likelihood ratio tests approximated with a chi-square distribution, χ² = 0.030–0.848, d.f. = 1, p = 0.357–0.863; figure 5). The attractiveness of (±)-2 was very slight and significantly lower than those of the natural and synthetic pheromones including (±)-1 (χ² = 7.84–16.2, d.f. = 1, p < 0.05 after Bonferroni correction).

Figure 4.

Synthetic (1,2-dimethyl-3-methylenecyclopentyl)acetaldehyde [(±)-1 and (±)-2].

Figure 5.

Males captured (mean + s.e.m.) by traps baited with natural and synthetic pheromones [(±)-1 and (±)-2] in a pineapple field. Attractants (0.1 mg) were released from a rubber septum. Bars labelled with the same letters are not significantly different (multiple comparisons using GLMMs followed by Bonferroni correction; p > 0.05). Blank, solvent (hexane) only.

3.2. Pheromone production in sexual and asexual females

We quantified pheromone production and release from female mealybugs by means of small-scale volatile collection and GC analyses. We first investigated diel rhythms of pheromone production by collecting volatiles from 14 day old females every 2 h. Virgin females of the sexual lineage started their pheromone release immediately before light-on and produced it intensively for the first 4–6 h of the photoperiod (figure 6). The pheromone release was then drastically decreased and almost stopped in the last 6 h of the photoperiod. In contrast, no pheromone was detected in volatiles from females of the asexual lineage at any time.

Figure 6.

Diel rhythms of pheromone production (mean ± s.e.m.) by unmated 14 day old female mealybugs of the sexual and asexual lineages.

Next, we examined daily pheromone production by 0 to 14 day old females (figure 7). Pheromone production by virgin females of the sexual lineage increased up to 10 days of adult age and continued by 14 days. Sexual females that mated at 1 day old completely stopped the pheromone production after mating and never reactivated production. Asexual females never produced any detectable pheromone. No EAD-active compounds, i.e. potential pheromones, were found except 1 released from the sexual females (figure 7).

Figure 7.

(a) Chromatograms of volatiles from 14 day old female mealybugs (10 female day equivalents) of the sexual and asexual lineages by coupled GC-FID/EAD analyses and (b) daily pheromone production (mean ± s.e.m.).

3.3. Mealybug phylogeny and pheromones

According to the ML tree, all mitochondrial CO1 sequences formed monophyletic clusters that likely reflect the species' taxonomic relationships (figure 8). However, the pheromone structures [20,23,24,51–56] were discordant with the phylogeny; for example, the species most closely related to D. brevipes (our study species) was Dysmicoccus neobrevipes, whose pheromone is an acyclic acetate similar to that of Planococcus minor and very different from the aldehyde with a cyclopentane of D. brevipes (figure 8). None of the evolutionary models tested showed significantly better fitness than the model of rate constancy through time for transitions of the three characters of the mealybug pheromones (likelihood ratio tests approximated by a chi-squared distribution, p = 0.209–0.901 for the punctuational model and p = 0.140–0.782 for the accelerating/decelerating model). Phylogenetic signals by Pagel's lambda were not significant (likelihood ratio tests approximated by a chi-squared distribution, p = 0.264 for functional group, p = 0.420 for carbon number of terpene moiety and p = 0.133 for cyclic/acyclic structure), indicating that structural similarities of mealybug pheromones are not likely to reflect phylogenetic relationships.

Figure 8.

Phylogeny of mealybugs based on partial sequences of mitochondrial gene (CO1). The tree was constructed by the maximum-likelihood method using unambiguously aligned nucleotide sites (1104 bp). The bootstrap values (>50%) obtained from 1000 resamplings are given at the nodes. The GenBank/EMBLENA/DDBJ accession numbers of each sequence are shown in brackets.

4. Discussion

Sex pheromones that can recruit males from a distance are indispensable for copulation and are essential in the sedentary life cycle of coccoid females for sexual reproduction. We discovered and characterized a monoterpene with unusual non-head-to-tail isoprene connections, or a typical head-to-tail connection followed by cyclization and unusual shifts of methyl groups, (−)-(anti-1,2-dimethyl-3-methylenecyclopentyl)acetaldehyde (1), as the sex pheromone of the sexual lineage of the pineapple mealybug, D. brevipes (figures 2–5). This result substantiates that coccoid insects have developed characteristic and structurally radiated pheromone systems [18,22,25]. On the other hand, the sex pheromone (1) production is completely lost in the asexual sister lineage (figures 6 and 7), which reproduces by obligate thelytokous parthenogenesis. In addition, no pheromone candidates that elicit male-antennal responses were discovered in volatiles from the asexual females (figure 7). As other members of the genus Dysmicoccus reproduce sexually, and only D. brevipes uses obligate parthenogenesis [33], the common ancestor of the two lineages of D. brevipes would have reproduced sexually [38]. Hence, as an evolutionary consequence of acquiring parthenogenetic reproduction, the asexual mealybugs appear to have abandoned production of the sex pheromone, which had been critical to attracting males among their ancestors.

Given the costs required to produce and maintain most characters, useless traits are considered to be disadvantageous and their reduction should be favoured by natural selection [57]. However, male sexual traits that are expected to be under relaxed selection and decay via the accumulation of neutral mutations are generally functional or display only minor shifts since the abandonment of a sexual life cycle [58]. On the other hand, sexual traits of asexual females generally display large-scale shifts relative to sexual females, often indicating decay [58]. For example, thelytokous parthenogenetic females of a parasitoid wasp, Asobara japonica (Hymenoptera: Braconidae), harbouring a reproduction-manipulating bacterium (Wolbachia) do not attract males in courtship behaviour bioassay, suggesting a reduction or absence of female pheromone production [59], although their pheromones have not been chemically analysed and identified. Similarly, females of five asexual species of Timema stick insects (Phasmatodea: Timematidae) are commonly characterized by reduced sexual attractiveness to males of closely related species in behavioural assays [60]. In this case, chemical profiles of cuticular hydrocarbons (methylheptacosanes), which are potential mate recognition cues for stick insects, are more variable in asexual than in sexual females, implying decay of sexual signal production [60]. Our results are consistent with these empirical reports and strongly support theoretical expectation of an evolutionary consequence for sexual traits under selection pressure after the acquisition of asexual reproduction; our chemical analyses combined with behavioural bioassays and phylogenetic analyses have pinpointed a critical pheromone compound for sexual females and its complete loss in asexuals within a period of relatively rapid evolution (approx. 1.3 million years) [38].

Mealybug pheromones have a common structural feature: their terpene moieties are composed of two isoprene units coupled by irregular non-head-to-tail connections, and pheromone structures of mealybug species are considered to be conserved as a whole [22–24]. However, the structural similarities of the pheromones between each mealybug species appear to be inconsistent with the phylogenetic relationships (figure 8). The pheromone of the sexual D. brevipes (1) is a cyclopentylacetaldehyde, whereas that of the allied D. neobrevipes is a lavandulol-related acyclic acetate [23]. Moreover, unusual and similar cyclobutane structures, which are presumably derived from cyclization of intermediates with a common non-head-to-tail (1 → 2′) linkage of isoprene units, are found in three species of different genera—Planococcus citri [51], Pseudococcus cryptus [56] and P. solenopsis (outgroup) [24]—indicating that the cyclization process has been acquired (or lost) multiple times during mealybug speciation. Discordance between phylogeny and pheromone chemistry may indicate that positive selection has forced the genetic background that controls pheromone biosynthesis to generate signals that can be clearly discriminated from those of closely related taxa during evolutionary radiation of mealybug species.

Such saltational shifts of insect pheromones are reported in several taxa, including flies, beetles and moths [61,62]. In Ostrinia moths (Lepidoptera: Crambidae), saltational shifts of pheromone blends, which are composed of similar unsaturated fatty compounds, have been promoted by positive selection acting on orthologous genes encoding pheromone biosynthesis enzymes, leading to avoidance of interspecific interference among closely related taxa [14,17,63]. In Bactrocera fruitflies (Diptera: Tephritidae), which have other mechanisms to ensure reproductive isolation, male-produced pheromone compositions are shown to have evolved through rapid saltational changes associated with speciation, followed by gradual divergence thereafter [62]. The aggregation pheromone blends of bark beetle species of two genera, Dendroctonus and Ips (Coleoptera: Curculionidae), are obviously different between the genera but show a less clear phylogenetic pattern between the species within each genus, suggesting that, within certain phylogenetic constraints, pheromones of bark beetles have evolved with saltational shifts, resulting in sibling species being pheromonally different from one another [64]. Because coccoids are sedentary insects and cannot readily change their habitats, which often play a role in assortative mating and reproductive isolation [62], interference among pheromone signals is likely to be much more serious than for mobile insects, and this may have driven the evolution of species-specific pheromone compounds that ensure unique chemical channels.

Unmated sexual females copiously released the pheromone during the first half of the photoperiod but rapidly ceased to release it after that (figure 6). Moreover, mated sexual females ceased pheromone production (figure 7). These findings indicate that the pheromone is produced only during a limited period even by the sexual mealybugs, which may save a cost of continuous production. Most pheromones are biosynthesized de novo in specific organs [65,66] and thus require nutrient resources [67]. Terpenoid pheromones of coccoids are considered to be de novo biosynthesized, because their structures are unique, and no compounds with similar skeletons are found in host plants [68]. It is therefore favourable to limit their production to necessity. Wasteful metabolism would be critical particularly for coccoids that feed on generally nitrogen-deficient plant sap. Nitrogen is not included in coccoid pheromones but is necessary to operate their biosynthetic cascades. Moreover, pheromones are often perceived not only by conspecific mates, but also by natural enemies as kairomones: some coccoid pheromones are known to recruit predators and parasitoids [18,68,69]. These potential negative effects would have driven the loss of pheromone production in the asexual D. brevipes after the acquisition of parthenogenetic reproduction.

Hitherto known mealybug pheromones are all carboxylic esters of terpenols, but the pheromone of the sexual D. brevipes is a monoterpene aldehyde. It remains unclear whether the aldehyde pheromone is found only in D. brevipes or is present in other species yet to be studied. The sexual D. brevipes presumably has a unique and characteristic step to convert the functional group of the pheromone precursor to an aldehyde, whereas other mealybugs produce their pheromones via esterification of an alcohol precursor with the pheromones' own carboxylic acid. The asexual D. brevipes is likely to have lost or silenced factors for these biosynthetic processes. Comparative studies of the sexual and asexual lineages of D. brevipes and of other mealybugs will provide valuable insights into how the biochemical background generates terpenoid pheromones with notable structural diversity.

The pheromone compound discovered in this study shows strong attractiveness to males of the sexual D. brevipes, but it may be costly to produce sufficient quantities for use in pheromone-based control programmes to protect crops. Moreover, such pheromone-based tactics would not be applicable to the asexual D. brevipes. Instead, another attractant (cyclolavandulyl butyrate) is known to recruit parasitic wasps that attack a broad range of mealybugs, even those that are not typically parasitized under natural conditions [70,71], and to suppress mealybug population increases in orchards [72]. Such attractants for biological control agents may have potential for the management of D. brevipes without dependency on the use of insecticides.

Authors' contributions

J.T. conceived, designed and coordinated the study, collected insects and volatiles, performed chemical analyses, analysed phylogenetic data, and wrote the manuscript; R.T.I. carried out laboratory bioassays, helped design the study and helped draft the manuscript; C.M. carried out field bioassays; K.M. carried out organic synthesis and helped draft the manuscript. All authors gave final approval for publication.

Competing interests

We declare we have no competing interests.

Funding

We acknowledge the grant-in-aid for scientific research from the Japan Society for the Promotion of Science (no. 16K08103) to J.T.