Evaluation of d-Limonene and β-Ocimene as Attractants of Aphytis melinus (Hymenoptera: Aphelinidae), a Parasitoid of Aonidiella aurantii (Hemiptera: Diaspididae) on Citrus spp.

The volatile organic compounds (VOCs) released from herbivore-infested plants can be used as chemical signals by parasitoids during host location. In this research, we investigated the VOC chemical signals for the parasitoid Aphytis melinus to discriminate between Aonidiella aurantii (California red scale)-infested fruit and non-infested fruit on three different citrus species. First, we identified the chemical stimuli emanating from non-infested and A. aurantii-infested citrus fruits via solid phase microextraction (SPME) and gas chromatography-mass spectrometry (GC-MS) analyses and identified 34 volatile organic compounds (VOCs). The GC-MS analysis showed qualitative and quantitative differences between VOCs emitted from non-infested and infested citrus fruit. Two VOCs, d-limonene and β-ocimene, were significantly increased in all infested fruit, regardless of the fruit species. The response of the female adult A. melinus to olfactory cues associated with A. aurantii infested fruit was evaluated using a Y-tube olfactometer. In two-choice behavioural assays, A. melinus females preferred infested citrus cues over non-infested fruit. Females showed positive chemotaxis toward these VOCs in all tested combinations involving two dosages of synthetic compounds, d-limonene and β-ocimene, except for d-limonene at a dosage of 10 μL/mL. The application of these VOCs may help to enhance the effectiveness of bio-control programs and parasitoid mass-rearing techniques.


Introduction
The citrus industry is considered one of the largest fresh fruit industries in Australia, as well as the largest fresh fruit exporter, with an annual average export volume of 170,000 tonnes and a value of AUS$190 million [1]. Aonidiella aurantii (Maskell) (Hemiptera: Diaspididae), is a major insect pest of citrus crops, causing significant production losses [2]. A. aurantii prefers to attack citrus, and it is one of the most important citrus pests worldwide [3]. A. aurantii attacks all citrus species, but there are varying levels of susceptibility, so the order of vulnerability in descending is lemon trees (Citrus limon (L.) Osbeck), grapefruit (C. xparadisi Macf.), orange (C. sinensis (L.) Osbeck) and two different species of

Plant Material
Three different citrus species (lemon, orange and mandarin) were used for behavioural assay and GC-MS analysis. Citrus fruit was collected from 25-year-old trees (spaced 4.5 × 5 m apart) on 15 September 2016, from Murdoch University citrus orchard located in Western Australia, same location as mentioned above. Non-infested or infested fruit from each citrus species was collected manually, stored in glass jars (10 cm diameter, 20 cm high) and transferred to laboratory within 1 h. The infested fruit samples were selected which had the most second and early third instar nymphs. Non-infested fruit samples were chosen to be free of any naturally physical damages as much as possible. Prior to test, the fruit were stored in the laboratory at 25 ± 2 • C and 45-55% RH for 5-7 days. The tested fruit samples were uniformed in weight, each sample consisting of one piece of fruit weighing about 150 g. The infested citrus fruit were selected with 300-400 s and early third instars of A. aurantii.

Extraction of VOCs and GC-MS Analysis
Random pieces of citrus fruit, either non-infested or infested, were placed individually into a 500 mL glass jar and sealed to equilibrate gases for 8, 16 and 20 h for mandarin, orange and lemon, respectively. The differences in the equilibration time for extracting VOCs for each fruit could be attributed to species characteristics and many other factors such as growing area, weather conditions and non-homogeneity of the fruit ripening stage. After this, VOCs were extracted by inserting and exposing a solid-phase microextraction (SPME) fibre with 50/30 µm Carboxen/DVB/PDMS (2 cm) (Sigma-Aldrich, Bellefonte, PA, USA) to the headspace of the glass jar over the fruit samples at room temperature at 25 ± 2 • C and 45-55% RH for 2 h. The fibre was then withdrawn into the needle and transferred to the injection port of the GC-MS system as described by [22]. The desorption time of the SPME fibre at GC-MS injector was 10 min [16]. Six replicates were conducted for each treatment. The VOCs were analysed by GC-MS using an Agilent 7820A GC (Mulgrave, Victoria, Australia), equipped with a DB-35ms (30 m × 0.25 mm × 0.25 µm) fused-silica capillary column (Agilent Technologies, Santa Clara, CA, USA), with a mass spectrometer detector 5977E (Agilent Technologies, USA) under splitless mode. The carrier gas was 99.99% helium supplied by BOC (Sydney, Australia). The operation conditions of GC-MS were 250 • C injector temperature; the oven temperature was initially programmed at 60 • C and increased to 270 • C (by 5 • C/min); the column flow rate and splitless were 20 mL/min at 1.5 min and 1:1; the total GC-MS run time was 45.4 min.
Compound peaks were deconvoluted by AMDIS version 2.72 and identified by searching the NIST 2014 MS database (US National Institute of Standards and Technology, Gaithersburg, MD, USA) with retention index confirmation. The experiment was repeated twice.

Y-Tube Olfactometer Behavioral Experiments
The olfactory response of A. melinus to host-associated cues was tested using a glass Y-tube olfactometer purchased from Volatile Collection Systems Co LLC (Gainesville, FL, USA). The Y-tube olfactometer design was similar to that described in [19] with slight modifications for the purpose of our research. Briefly, the Y-tube olfactometer consisted of a central tube (2.5 cm diameter, 15 cm long) and two lateral arms (2.5 cm diameter, 11 cm arm's length) with ground glass fittings through which humidified air passed (0.5 mL/sec through each arm, controlled by up-stream flow meters). Each arm was connected to an extended glass tube (2 cm diameter, 6 cm long). The extended glass tube with the mesh barrier prevented the insects from escaping. The air was filtered with activated carbon traps, then passed through a chamber into each fruit, then VOCs and insects could be introduced. The air passed through the olfactometer and then the system was left to stabilise for 15 min prior to use.
Olfactometer studies were carried out at 25 ± 1 • C (45-55% RH) in a room illuminated with overhead daylight fluorescent tubes. A light was placed over the Y-tube olfactometer, and the surrounding area (around and below) was shrouded with white paper to block out any visual cues. A. melinus is an ectoparasitoid, so pupae could be isolated from host and host plant material without disrupting their development. From adult emergence until their use for bioassays, parasitoids were held in tubes (1 cm diameter, 4 cm length) and droplets of honey were provided throughout their lives. All wasps used for the bioassays were naive females without any previous contact with infested or non-infested host fruit. After placing parasitoids in the conditioning vial for a day, wasps were removed and placed in an empty container. On the day of bioassay, female wasps were isolated and held individually in glass vials with a drop of honey until tested. A total of 40 replicates (individual A. melinus) were conducted for each of the three citrus fruit species, and the four tested dosages of the synthetic VOCs. For each bioassay, a single female wasp was introduced into the central arm tube of the olfactometer. Once the wasp had reached the point where the arm divides (i.e., when the wasp reached a position to make a choice), a timer was started. If the parasitoid remained stationary for two minutes, it was considered as unresponsive. It was then excluded from the study and substituted with another one.
The olfactometer was reversed after half of the wasps in each replicate had been tested. After all ten parasitoids were assayed, the odour sources were removed, and the apparatus was cleaned with water and acetone, then dried and heated in an oven at 80 • C for at least half an hour between replicates. Overnight between treatments, new sources were introduced into the aeration chambers, and a further replicate of ten parasitoids was tested. For bioassays involving fruit, fresh and ripe fruit samples were tested to reduce the variation in volatile emissions between replicates.
To understand whether infested citrus fruit is attractive to A. melinus, at the first stage, mated females were allowed to choose between non-infested or infested fruit samples for each citrus species. A wasp was considered to have chosen a cue when it remained in the same chamber for at least 20 s, actively searching for a host. The bioassay replicate was considered complete when the wasp left the chamber. However, if wasps are unresponsive after two minutes in the system, they were eliminated from the study.
For assessment of the attraction of synthetic VOCs, two different concentrations (10 and 20 µL/mL) of d-limonene and β-ocimene (d-limonene purity 97%, 98% EE (GLC) Sigma-Aldrich, and β-ocimene purity ≥ 90% Toronto Research Chemicals, North York, ON, Canada) were dissolved in hexane. In each trial, 5 µL of solution was applied onto a filter paper (1.5 cm × 1.5 cm Whatman no. 1) and placed in an Erlenmeyer flask (250 mL) stoppered with a sealed adaptor. After solvent evaporation (20 sec), Erlenmeyer flask was connected to an arm of the Y-tube, and another arm received filter paper treated with 5 µL of pure hexane as a solvent control. Forty mated females were tested for each treatment.
For non-infested and infested citrus fruit bioassays, the female's first choice, and search duration (time spent actively searching inside the arm) were recorded. For the bioassay of synthetic VOCs, only the female's first choice was recorded. The activity recordings of parasitoids were used to assess the overall activity and the preference for each choice. The activity was measured by summing the number of entries into both arms by each wasp.

Statistical Analysis
To compare VOCs emissions between infested and non-infested fruit, the variance between peak areas was analysed for each compound and chemical class. Differences in volatile emissions between non-infested and infested fruit were analysed using non-parametric Mann-Whitney U test. Principal Component Analysis (PCA) was achieved on normalised values of each VOC to derive different variables (principal components) that summarise the original data. PCA analysis was performed using Metaboanalyst 3.0 (a comprehensive online tool suite for metabolomics data analysis). A likelihood chi-square test using Yates correction (with p = 0.05) for each choice-test was used to compare the proportion of parasitoids choosing a given cue. Time spent in the non-infested plant area and the infested plant area was compared using ANOVA, followed by Tukey's post hoc tests. (SPSS version 24.0, SPSS, Inc., Chicago, IL, USA).
The PCA showed that the volatile compound profiles varied with infestation and citrus fruit species (Figure 1). d-Limonene and β-ocimene were the major VOCs released from all tested fruits.
Mated A. melinus females, with no previous oviposition experience, were significantly attracted to the synthetic VOCs in all tests, with the exception of d-limonene at tested dosage 10 µL/mL (χ 2 = 2500; df = 1; p = 0.114). However, mated A. melinus females preferred the reward of associated VOC more than hexane control in the case of d-limonene at the tested dosage of 20 µL/mL (χ 2 = 7.410; df = 1; p = 0. 006), β-ocimene at the tested dosage 10 µL/mL (χ 2 = 4.900, df = 1, p = 0.027) and β-ocimene at the tested dosage 20 µL/mL (χ 2 = 12.100; df = 1; p = 0.001) (Figure 3). Asterisks indicate significant differences between uninfected and infested fruit at p < 0.05; Feature ID includes retention time (min) and m/z ratio z; RI retention index obtained from the NIST database; * Estimated non-polar retention index (n-alkane NIST scale); SE = standard error; n.d = not detected. Each value represents the peak area (mean ± SE) of n = 6 analyses. Insects 2020, 11,      Figure 2. Attractiveness of Aonidiella aurantii-infested citrus fruit towards Aphytis melinus mated females. Two choice bioassays were conducted in a still air arena with citrus fruit, infested or not by California red scale, providing olfactory cues. Forty wasps were tested in each bioassay. For each test, asterisks indicate significant differences in the number of wasps choosing different cue (χ 2 test with Yates correction, p < 0.05). To the left of the bars is the proportion of females that made a choice for one of the two odours, as well as the total number of females that were tested.   . Attraction of female Aphytis melinus to VOCs differentially emitted by A. aurantti-infested citrus fruits. Choice bioassays were conducted in a Y-tube olfactometer presenting one of the two dosage levels (10 and 20 µL/mL) of citrus fruits VOCs vs. hexane. Asterisks indicate significant differences between numbers of VOC and control choices (χ 2 test with Yates' correction, p < 0.05). To the left of the bars is the proportion of females that made a choice for one of the two odours, as well as the total number of females that were tested.

Discussion
This report investigated the capability of an aphelinid parasitoid A. melinus to respond to citrus fruit chemical cues associated with damage induced by a phytophagous pest. The results indicated that A. melinus was able to discriminate between citrus fruit infested with A. aurantii and non-infested fruit by using olfactory cues.
Among 34 VOCs extracted and identified by SPME and GC-MS techniques, only two VOCs, d-limonene, and β-ocimene were increased in all infested three citrus species, which were already known as a constituent of the odours of citrus fruit and citrus oils [16,20,23]. However, since d-limonene and β-ocimene are attractive to many other parasitoid species [19,20,24], these two compounds can be considered as putative VOCs as attractants of A. melinus. PCA analysis has highlighted that VOC emissions are particular for each species and chemical. Fruits differentially emitted the VOCs after insect infestation and the VOCs change depending on the citrus species. Indeed, several plants react to herbivore damage by producing blends of metabolites that change in number or proportion [25]. The increasing in VOCs follow the infestation could be explained for all species, but we also identified some particular compounds emitted under the condition of A. aurantii infestation. Besides d-limonene and β-ocimene, infested mandarin fruits displayed increased emissions of another three compounds (γ-terpinene, alloocimene and alloaromadendrene). Infested lemon fruits increased the emission of just d-limonene and β-ocimene and produced n-hexadecanoic acid. When attacked by A. aurantii, orange fruits similar to other species increased the production of d-limonene, β-ocimene, (3E)-4,8-dimethyl-,1,3,7-nonatriene, hexyl caproate and alloaromadendrene, and exclusively produced five compounds (acetic acid, hexyl ester, alloocimene, α-terpineol, nerolidol and (3E,7E)-4,8,12-trimethyltrideca-1,3,7,11-tetraene). Most of these compounds are known as common floral compounds, but interestingly some of them are also recognised as pheromones for several hymenopteran species [26].
Since the variance in VOCs depends mainly on the fruit species, we cannot exclude the possibility that other VOCs produced specifically by one species or increased in all species may act as an attractant toward A. melinus. Indeed, non-infested fruit also produce VOCs that are attractive to parasitoids [25], and aphelinid parasitoids, such as A. melinus could be able to perceive cues from non-infested plants to locate a host [19,20]. Therefore, short-range VOCs produced by the plants or as feeding larvae secretion could be useful for successful localisation [27]. Overall, since VOCs are known to act as attractive chemicals for several parasitoids [25,28,29], further research is needed to assess the activity of these compounds on parasitoid behaviour. Indeed, understanding tritrophic system communications has potential implications for biological control programs [30].
Olfactory cues from host-infested fruit are known to be essential for parasitoids during foraging for hosts, food and mates in complex environments [13,31]. For A. melinus, the source of volatile cues that the wasp uses to locate the host species were determined when those hosts occurred on more than one host plant species [19]. The evidence that olfactory cues from infested fruit evoke behavioural responses from mated A. melinus females, and the presence of chemicals that were produced exclusively or in higher amounts by infested citrus fruit, supported our hypothesis that VOCs could act as short-range attractants and played a key role during host-seeking. Biological assays conducted in the olfactometer indicated that A. melinus preferentially orientated towards the source of VOCs, that is, the presence of A. aurantii on the fruit is crucial for host location.
The VOCs d-limonene and β-ocimene have been reported as bouquet constituents of several citrus species [16,20]. d-Limonene and β-ocimene were identified as constituents of different infested fruit species that attracted other parasitoids such as Agathis bishopi and Aphidius gifuensis [24,32]. Many experiments have shown that, in the species examined, successful source location requires the presence of an attractive odour [33].
It is clear that A. melinus was not attracted to the pest when fruit material was absent, suggesting the importance of plant VOCs as long-range attractants of parasitoids, this result is consistent with that host insects do not always produce good cues [34]. This study confirmed that the aphelinid parasitoid A. melinus uses common VOCs to locate prey. The volatile profile analysis provides complementary information about the composition of the volatile chemicals potentially involved in the behaviours with 34 compounds varying during infestation by A. aurantii. Further electrophysiology and behavioural assays are expected to provide more accurate identification of the VOCs that contribute towards attracting A. melinus to infested host citrus fruit. Such information may prove useful for further studies aiming at developing semiochemical strategies that could be incorporated into an integrated management approach for enhancing existing pest control techniques for California red scale.

Conclusions
The aphelinid parasitoid A. melinus responded to citrus fruit chemical cues associated with damage induced by California red scale. The analytical data showed that more than 34 different volatile compounds were separated and identified from the infested and non-infested citrus species fruit. Interestingly, the three-species showed two common VOCs increased as a response of A. aurantii infestation: d-limonene and β-ocimene.
Olfactory cues seem to have a key role in host-seeking behaviour for the three citrus species. Synthetic VOCs may be useful to improve biological control programs especially after VOCs are recognized as VOCs of a number of parasitic wasps, but still, further studies are necessary to ensure the behavioural activity of this investigated VOCs toward A. melinus parasitoids under field condition.