Volatile Characterization of Lychee Plant Tissues (Litchi chinensis) and the Effect of Key Compounds on the Behavior of the Lychee Erinose Mite (Aceria litchii)

Herbivore-Induced Plant Volatiles (HIPVs) are volatile signals emitted by plants to deter herbivores and attract their natural enemies. To date, it is unknown how lychee plants, Litchi chinensis, respond to the induction of leaf galls (erinea) caused by the lychee erinose mite (LEM), Aceria litchii. Aiming to reveal the role of HIPVs in this plant-mite interaction, we investigated changes in the volatile profile of lychee plants infested by LEM and their role on LEM preferences. The volatile profile of uninfested (flower buds, fruit, leaves and new leaf shoots) and infested plant tissue were characterized under different levels of LEM infestation. Volatiles were collected using head-space-solid phase microextraction (HS-SPME) followed by gas chromatography-mass spectrometry (GC-MS) analyses. Fifty-eight volatiles, including terpenoids, alcohols, aldehydes, alkanes, esters, and ketones classes were identified. Using dual-choice bioassays, we investigated the preference of LEM to uninfested plant tissues and to the six most abundant plant volatiles identified. Uninfested new leaf shoots were the most attractive plant tissues to LEM and LEM attraction or repellence to volatiles were mostly influenced by compound concentration. We discuss possible applications of our findings in agricultural settings.


Introduction
Plants have sophisticated defense systems to recognize and counterattack herbivorous mites and insects. Such defensive responses involve changes in plant traits that interfere with herbivores directly or indirectly. Direct defenses include the production of toxicants and volatile organic compounds (VOCs) that repel or deter herbivores, while indirect defenses include the production of VOCs that attract natural enemies of the herbivore (for reviews of VOCs as direct and indirect defenses see Howe et al. [1] and Dudareva et al. [2] and the references therein). VOCs with toxic, repellent, or attractive properties include various groups of terpenoids (monoterpenes, sesquiterpenes and homoterpenes), phenylpropanoids and benzenoids, fatty acid derivatives, and amino acid derivatives [3][4][5]. Upon mechanical damage, mite, or insect attack, the plants' immune responses trigger the production of novel VOCs, known as Herbivore-Induced Plant Volatiles (HIPVs), locally and systemically within the plant [4].
The most well-documented function of HIPVs is the attraction of natural enemies [4][5][6], which indirectly promotes plant fitness [7]. Several studies have shown that terpenoids t n+1 and t n retention times of the reference n-alkane hydrocarbons eluting immediately before and after compound "X"; t x retention time of compound "X" [61] to a series of n-alkanes (C 9 -C 21 , Sigma-Aldrich, St. Louis, MO, USA), and compared with their RIs available in the literature [58,[62][63][64]. The relative percentages were obtained by total ion current (TIC) peak areas.
Principal Component Analysis (PCA) and Agglomerative Hierarchical Clustering (ACH) for GC-MS data were performed using XLSTAT 2021 (Addinsoft, New York, NY, USA). Only constituents in a concentration higher than 0.5% were used as variables for the PCA and ACH analysis. Data used for the PCA and ACH analyses were a 21 × 5 matrix (105 data) for samples A1 to A5, and 81 × 4 matrix (112 data) for samples B1 to B4. Pearson's correlation model was used for PCA. Euclidean distance for measure and Ward's method were used for ACH analysis.

Dual-Choice Test between Field-Collected (Non-Infested) Lychee Tissues
Considering the minute size of LEM (~200 nm), it was not possible to carry out conventional dual-choice bioassays using Y-tube olfactometers [65], petri-dish arenas [66], cages or greenhouses [67], as used for other mite species. All known experimental set-up were tested, but without yielding any mite response. Therefore, a new method to conduct dual-choice bioassays was developed. The experimental arena consisted of a microscope glass slide (75mm × 25mm) with markings dividing it into four equal sections (Figure 1a). Double-sided tape (1.25 cm wide) (Scotch ® , 3M, Miami, FL, USA) was placed alongside the edges of the slide to keep the mites in the arena (Figure 1a). Then, an Eppendorf ® tube filled with nutrient solution (10N-5P-14K) (MaxiGro ® , General Hydroponics Inc., Sebastopol, CA, USA) and holding one plant tissue (flower buds, fruits, leaves, or new leaf shoots), was placed on two opposite sides of the arena. Four arenas were placed on a white Styrofoam board (20 × 10 cm) ( Figure 1b) and each board was placed inside a custom-made acrylic cage (30 × 30 × 30 cm) with openings on two sides to allow airflow. The experimental units were kept at 25 ± 2 • C, 50% ± 10% RH, and a 12:12 h (L:D) photoperiod. Fluorescent lamps were placed above each cage to provide uniform light over the experimental arenas. A leafdisc (0.5 cm diam.), cut from an amber-colored erinea (i.e., fully infested with LEM) was placed in the middle of each arena (on the black dot Figure 1a), and after 24 h, the mites that moved towards one side and crossed the midline marking were considered attracted to the plant part offered on the same side. To test whether LEM was able to make reliable choices we offered the same plant tissue on both edges, as controls. The plant tissues tested were leaf, new leaf shoot, fruit, and quiescent flower bud. All possible combinations of plant tissues were offered to LEM in a factorial design (N = 8). Only LEM found in Sections 3 and 4 and TAPE ( Figure 1a) were considered in the analyses. Differences in the proportion of mites between treatments were tested with generalized linear mixed-effects models (GLMM) using the lme4 package [68]. Binomial error distribution and including the position of the slide (left or right) and Styrofoam board (replicate) as a random factor [69]. Residual plots of data revealed no major deviation from the normality and variance homogeneity assumptions. Analyses were performed separately for each plant tissue/combination and using the R program version 4.2.1 [70].

Dual-Choice Test on Single Volatile Compounds
Using the same experimental unit as above, dual-choice bioassays were used to evaluate LEM attraction to the six major volatile compounds identified in potted plants and field-collected tissues (nonanal, decanal, sabinene, limonene, β-caryophyllene and ar-curcumene; see Result Section 3.1). Eight concentrations were tested for each compound: 3%, 5%, 7%, 10%, 25%, 50%, 75% and 100%. Compounds were diluted in absolute EtOH, which was also used as a control. All dual-choice tests were made by offering a single compound concentration against the control (EtOH). To test whether LEM was able to make reliable choices we offered the clean filter paper or ethanol on both sides, as controls. Forty-five minutes prior to testing, 5 µL of each compound was applied to a circular filter paper (d = 1 cm), allowing enough time for the EtOH to evaporate. Then, the filter paper containing the test compound was placed on one end of the slide, and the filter paper containing EtOH (control) was placed on the other end ( Figure 1a). The assessment of LEM choice between the two odor sources was performed as above. Replicates in which LEM made no choice (remained in the middle of the arena, Sections 1 and 2) were discarded. To correct for the potential positional bias, the treatment and the control sources were switched from left to right in all choice experiments. The number of replicates per compound concentration are shown in their respective figures (see Section 3). The same analysis described above was performed separately for each compound, as well as each concentration.

Dual-Choice Test on Single Volatile Compounds
Using the same experimental unit as above, dual-choice bioassays were used to evaluate LEM attraction to the six major volatile compounds identified in potted plants and
Principal component analysis (PCA) was carried out to identify the grouping pattern among potted (infested) lychee plants ( Figure 3) and field-collected (non-infested) samples ( Figure 4) based on volatile compounds obtained from GC-MS data ( Table 1). For better distinction between the groups, we focused on the proportions of the volatile compounds. Based on the Eigenvectors criterion, principal components with an eigenvalue greater than one are considered important. First, PCA analysis was applied based on the GC-MS data from samples A1 to A5. Principal components F1 and F2 had higher Eigenvectors of 9.58 and 5.85, respectively ( Table 2). Two macro principal components accounted for 45.60% (F1) and 27.87% (F2) representing 73.47% of the total variance, indicating that the F1 and F2 had good support for the difference in the plant tissues ( Figure  3a).     Table 3 shows the percent contribution and squared cosine values (cos 2 ) of each principal component. The percent contribution of each variable shows the most important variables. Also, the cos 2 values indicate the potential importance of components for a given observation. Taking the values of cos 2 and the percent contribution, nonanal (7.63%; 0.73), decanal (7.44%; 0.71), ar-curcumene (9.57%; 0.92), α-zingiberene (8.59%; 0.82) and δ-cadinene (8.60%; 0.82) showed a strong impact on the F1, while α-copaene (16.13%; 0.94) showed the strongest impact on the F2. Cos 2 for F1 was the highest impact on the observation of initial infestation A1 (0.758) and overexploitation A4 (0.614); heavy infestation A3 (0.724) was the highest impact on F2.    Table 3 shows the percent contribution and squared cosine values (cos 2 ) of each principal component. The percent contribution of each variable shows the most important variables. Also, the cos 2 values indicate the potential importance of components for a given observation. Taking the values of cos 2 and the percent contribution, nonanal (7.63%; 0.73), decanal (7.44%; 0.71), ar-curcumene (9.57%; 0.92), α-zingiberene (8.59%; 0.82) and δ-cadinene (8.60%; 0.82) showed a strong impact on the F1, while α-copaene (16.13%; 0.94) showed the strongest impact on the F2. Cos 2 for F1 was the highest impact on the observation of initial infestation A1 (0.758) and overexploitation A4 (0.614); heavy infestation A3 (0.724) was the highest impact on F2.

LEM Behavioral Responses to Field-Collected (Non-Infested) Lychee Tissues and Single Volatile Compounds
In almost all cases, when given a choice between two field-collected (non-infested) lychee tissues (control groups), there was no significant difference in the number of LEM that reached both sources. Except when presented with a choice between leaf shoot vs. leaf shoot there was a slightly significant difference (p = 0.04) towards one of the sides ( Figure S2). For the other groups tested, fruit vs. fruit (p = 0.1), leaf vs. leaf (p = 0.1) and bud vs. bud (p = 0.31) there was no difference between choices made by LEM. When different plant tissues were offered, LEM showed a preference for leaf shoot over buds (p = 0.04, Figure 5). No clear choice was observed for the other treatments, leaf vs. fruit (p = 0.61), leaf vs. bud (p = 0.12), leaf vs. leaf shoot (p = 0.31), fruit vs. bud (p = 0.31) and fruit vs. leaf shoot (p = 0.1).
In our subsequent experiment using potted plants (A1-A5), 14 unique compounds were found, mostly aldehydes and alkanes (Table 1, Figure 2). In the PCA analysis from samples A1 to A5, the abundant aldehydes nonanal and decanal were found to have a strong discrimination effect among the LEM infestation levels, separating them into two clusters ( Figure 3). The first cluster included the initial infestation (A1), heavy infestation (A3) and uninfested plant (A5) treatments, and the second cluster included the intermediate infestation (A2) and overexploitation (A4) treatments. In addition, the terpenoids limonene, β-caryophyllene, ar-curcumene and α-zingiberene also contributed greatly to discriminate among LEM infestation levels. Hence, the terpenoids β-caryophyllene, ar-curcumene and α-zingiberene were found to play an important role in the discrimination among both the infested and uninfested lychee plants (A), besides their participation in the discrimination among the field-collected tissues (B). Importantly, while the levels of the most abundant aldehydes (nonanal, decanal) decreased with the progression of the LEM infestation, the most abundant terpenoids (sabinene, limonene, ar-curcumene and β-caryophyllene) and alkanes (tetradecane, hexadecane, heptadecane) increased in the presence of LEM, suggesting that those major compounds may play an important role in lychee plant defense and LEM attraction/repellence for these compounds might differ.
Recently, Gunpal and Patni [54] showed that lychee leaflets infested by LEM were rich in sugars, terpenoids, linoleic acid esters, vitamins, antioxidants, glycerol, and steroids. Our results were different from those reported by Gunpal and Patni [54] and we attribute these differences to the sampling methods. They used dried, powdered leaves extracted by Soxhlet apparatus with methanol. Our analyses using the HS-SPME method possess unsurpassed sensitivity and specificity in volatile sampling. In addition, we used authentic standards to confirm the identity of the compounds, while Gunpal and Patni [54] based their identification solely on the mass spectra of peaks. Second, our study offers further knowledge, demonstrating for the first time that the volatile profile of LEM infested lychee plants changes according to LEM infestation level. It has been shown by Popitanu et al. [71] that the erineum-forming mite Aceria erinea induced significant emissions of terpenoids in Persian walnut leaves (Juglans regia) and these emissions scaled positively with the percentage of the infested leaf area. In agreement with these findings, our study showed that terpenoids correlated positively with the LEM progression of infestation. These findings raised the hypothesis that LEM responses to the main terpenoids emitted by lychee are dose dependent. Aiming to address this question, we designed a dual-choice arena ( Figure 1) to investigate the LEM preferences when offering the terpenoids β-caryophyllene, ar-curcumene, limonene and sabinene on different concentrations. In fact, we confirmed our hypothesis by showing that key terpenoids played an important role in LEM attraction or repellence, depending on their concentration (Figures 8-11).
Terpenoid VOCs play an important role in direct and indirect plant defenses [3]. These compounds can be toxic and repellent to the attacking insects [72] and mites [73], and can attract natural enemies of herbivores [72,74]. For example, the terpenoids α-pinene, eugenol, limonene, and terpinolene repel several mosquito species [75,76], and α-pinene repels the tick Ixodes ricinus [77]. Additionally, the sesquiterpene β-caryophyllene has been shown to repel mosquitoes [78] and the Asian citrus psyllid, Diaphorina citri Kuwayama (Hemiptera: Liviidae) [79]. Terpenoids can also act as elicitors that trigger plant defense mechanisms to target pests. For instance, tomato plants exposed to the volatiles of transgenic tobacco enriched in β-ocimene decreased aphid, Macrosiphum euphorbiae, development and reproduction (direct defense) and increased attraction of its natural enemy, the parasitoid Aphidius ervi (indirect defense) [72]. Similarly, by inhibiting the emission of terpenoid HPIVs, Mumm et al. [74] demonstrated that the inhibition of those terpenoids in mite-infested plants decreased the attractiveness of the predatory mite Phytoseiulus persimilis.
Like terpenoids, aldehydes can also act as specific elicitor molecules emitted after herbivory, pathogen infection, or oxidative stresses [80,81]. For example, Yi et al. [81] showed that Lima bean plants growing adjacent to plants chemically induced with benzothiadiazole up-regulated the production of nonanal and were more resistant to infection by the pathogen Pseudomonas syringae. In insect-plant systems, infestation by the peach aphid (Myzus persicae) and the Colorado potato beetle (CPB), (Leptinotarsa decemlineata) drastically increased the amounts of nonanal in attacked plants [82]. The increased amounts of nonanal in potato plants infested with CPB played an important role in the attraction of the two predators, Podisum maculiventris and Perillus bioculatus [80]. However, nonanal has also been shown to increase the attraction and oviposition of Helicoverpa assulta (Lepidoptera: Noctuidae) moths in tobacco plants [83]. In the current study, the emissions of nonanal and decanal decreased with increasing LEM infestation (Table 1). Although the results from the behavioral assays were inconsistent with increasing doses of nonanal and decanal (Figures 6 and 7), in the trials with lower concentrations of decanal (3, 5, and 7%) the treated-side was more attractive. Taken together, these data may suggest that decanal may initially elicit attraction to LEM, and as infestation progresses and decanal emissions increase, infested tissues may become less attractive to LEM.
The attractiveness of volatile odors to most insects and mites is dose dependent [84][85][86][87]. Similarly, in the present study, we found that the compound concentration was the most important factor determining attraction or repellence of LEM. A trend towards attraction of LEM at low and intermediate concentration and repellence at high concentration was found in three out of the six volatiles tested. Decanal (Figure 7), β-caryophyllene ( Figure 8) and limonene ( Figure 10) were mostly attractive to LEM at low and intermediate concentrations. Sabinene ( Figure 11) promoted attraction of LEM in almost all concentrations tested, and nonanal ( Figure 6) promoted repellence or had no effect on LEM preferences. Importantly, since we tested individual compounds, the outcome might be different when testing them in a mixture. Further research should investigate if volatile emissions of these compounds are associated with natural enemies of LEM.
Several studies have shown that the ratio of odor blends are important in insect attraction [88][89][90]. Blends of volatiles in the proper ratios rather than single compounds can mimic natural conditions. Thus, the next step in understanding the attractive role of the lychee volatiles identified in this study to LEM is testing different odor blends at different ratios. Nevertheless, the findings that the LEM in this study significantly preferred some volatile compounds at lower concentrations is an important first step in understanding the attractive behavior of this pest. Further, to our knowledge, this is the first description of a bioassay for testing LEM odor preferences. It is important to note that because LEM hides in the erineum it was not possible to assess the number of mites released in the arena. In an erineum disc of 0.5 mm dimeter the LEM population may vary from 0 to more than 8000 mites. This information, however, can only be confirmed after the mites exit the erineum. In addition, most LEM cannot survive outside the erinea for more than 24 h [39], meaning that mites that did not make their decision within the period of observation (24 h), ended up dead in the middle of the arena. As this method allows mites to walk freely, it is uncertain how long a mite stayed on either side of the slide. Hence, we did not evaluate the activity of the mites, but the proportion of mites choosing one specific odor over another and whether the compound repelled or attracted those mites. By assessing only their final choice, this experimental set-up might have decreased the chances of having undecided mites influencing the results. One may wonder why there was a variation in LEM response when offered the choice between leaf shoot vs. leaf shoot (control group). The non-infested lychee tissue used was field-collected. Individual lychee trees may be exposed to different biotic and abiotic conditions that in turn may influence nutrition or synthesis of defensive compounds. These factors can also affect mite behavior. In our study we were not able to control for this variation.
Finally, LEM is a host specific eriophyid mite known to attack several lychee tissues [32]. Consistent with what has been observed in the field, here we showed that LEM's most preferred tissue were the new leaf shoots ( Figure 5). The attacked young tissue and the erinea progress simultaneously, meaning that the infested old plant tissue only dies months after exploitation by the mite (Ataide et al. in preparation). Once new growth starts to form, mite infestation in new tissue prevails and soon the whole plant is infested. This is important information for the management of LEM. It is known that lychee trees undergo two to three waves of flushing each year, usually in the summer and early in the autumn in southern China [91]. Thus, this is a critical period for prophylactic treatments with pesticides. Once the new leaf shoots are protected, LEM spread within and between plants and therefore yield losses are expected to be suppressed.
Here, we showed that there is the potential to use key terpenoids and aldehydes in the attraction or repellence of LEM to lychee plants. In the future, lychee VOCs or HIPV can be used to develop volatile-based management strategies against LEM. HIPVs, especially elicitors such as the plant hormones salicylic acid and jasmonic acid, could induce the plant defense system before insect attack [91][92][93][94][95]. In addition, a "semiochemically assisted trap cropping", which consists of the use of trap crops supplemented with lures, such as pheromones or kairomones, can enhance the effectiveness of pest control [96]. All these strategies have contributed to the detection, monitoring, and manipulation of populations of native and invasive insect species around the world. However, these strategies have been barely investigated for mite control [97] and there are still significant knowledge gaps in the chemical communication in the acari compared to insects. As they are wingless organisms, we envision a management system for LEM focused on the use of attractant semiochemicals to lure and target them outside the erinea. Alternatively, repellents could potentially be used to protect new shoots from LEM infestation. Overall, by addressing LEM preferences to plant volatiles and specific plant tissues, we hope that new control strategies can be developed.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/biom13060933/s1, Figure S1: Images of the four erineum stages associated with different levels of LEM infestation. Photos of the leaflets showing the erinea were taken using an Apple iPhone 11 (f/1.8; 1/120 s; ISO-125). Photo background has been removed using Adobe ® Photoshop CS 5.0. Figure