Rootstocks with Different Tolerance Grade to Citrus Tristeza Virus Induce Dissimilar Volatile Profile in Citrus sinensis and Avoidance Response in the Vector Aphis gossypii Glover

The citrus tristeza virus (CTV) is an agent of devastating epidemics of the citrus plant grafted on Citrus aurantium, one of the main rootstocks still used in the Mediterranean area. Consequently, CTV-tolerant alternative citrus rootstocks are considered necessary to manage this disease and/or its vector; that in Mediterranean countries is the aphid Aphis gossypii. In this study, we analyzed the VOCs emitted from Citrus sinensis plants grafted on the CTV-susceptible C. aurantium and on the CTV-tolerant Volkamer lemon, Forner-Alcaide no. 5, and Carrizo citrange. Furthermore, the aphid preference/avoidance response toward these combinations was evaluated in a semi-field experiment. The VOC profiles recorded on the leaves of C. sinensis grafted on the four rootstocks listed above showed significant differences in the abundances and ratios of the compounds emitted. The behavioral experiments indicated that A. gossypii prefers to orient and establish on the C. sinensis plants grafted on C. aurantium rather than on that grafted on the three CTV-tolerant varieties. The possibility that this avoidance mechanism is triggered by the different profile of the VOC emitted by the different combinations and the consequent susceptibility/tolerance shown toward CTV is discussed.


Introduction
The citrus tristeza virus (CTV) is an aphid-borne closterovirus (genus Closterovirus, family Closteroviridae), an agent of devastating epidemics that drastically modified the course of the citrus industry [1]. CTV infection causes a decline syndrome of different citrus species such as sweet oranges (Citrus sinensis L.), mandarins (C. reticulata Blanco), and grapefruits (C. paradisi Macf.), propagated on sour orange (C. aurantium L. Osb.) [1,2]. The sour orange was formerly the most important rootstock in almost every citrus-growing region worldwide [3] and still is highly widespread in the countries of the Mediterranean basin. The CTV epidemics determined the death of almost 100 million trees propagated on sour orange and became a limit for their use as a rootstock [1,4]. CTV attacks on the trees grafted on sour orange induce a quick plant decline, which results from a virus-induced graft incompatibility between the scion and rootstock [5].
This generated the necessity of finding alternative CTV-tolerant rootstocks that, similarly to sour orange, produce good yields of high-quality, tolerate abiotic stress factors, such as frost and salinity, and resist biotic constraints, such as gummosis and foot rot caused by Phytophthora spp. and the viroids of citrus exocortis (CEVd) and hop stunt (HSVd) [6,7].
In consideration of the efficient CTV dispersals by vectors, the propagation of citrus on CTV-tolerant rootstocks is nowadays considered the only viable option to manage CTV and avoid the tristeza decline [8,9].
In this context, among the CTV-tolerant rootstock candidates to substitute sour orange, only a few genotypes have shown promising potential. Among them, there are Volkamer 6.33 ± 1.88 leaves in plants of C. sinensis grafted on C. aurantium, a value significantly higher than that found in C. sinensis grafted on Carrizo citrange, where only 1.58 ± 0.57 leaves evidenced the presence of aphids (p < 0.01; ANOVA followed by Tukey's test), and on Forner-Alcaide no. 5 and Volkamer lemon with, respectively, 2.33 ± 0.65 and 2.33 ± 0.61 leaves infested (p < 0.05; ANOVA followed by Tukey's test). No statistical differences were recorded in the number of aphid infested leaves between C. sinensis grafted on Carrizo citrange, Forner-Alcaide no. 5, and Volkamer lemon ( Figure 1; Table S1).  Table S1.

VOC Chemical Analysis
VOC collection with headspace Solid-Phase Micro-Extraction (SPME) method followed by GC-MS analysis revealed the presence of a total of fifty-six VOCs released from the C. sinensis leaves grafted on the four different rootstocks. Ninety percent of Asterisks indicate significant differences among the treatments (* p < 0.05; ** p < 0.01; ANOVA, followed by Tukey's test). The values are reported in Table S1. In each of the three timepoints, it was observed that aphids infested a mean (±SE) of 6.33 ± 1.88 leaves in plants of C. sinensis grafted on C. aurantium, a value significantly higher than that found in C. sinensis grafted on Carrizo citrange, where only 1.58 ± 0.57 leaves evidenced the presence of aphids (p < 0.01; ANOVA followed by Tukey's test), and on Forner-Alcaide no. 5 and Volkamer lemon with, respectively, 2.33 ± 0.65 and 2.33 ± 0.61 leaves infested (p < 0.05; ANOVA followed by Tukey's test). No statistical differences were recorded in the number of aphid infested leaves between C. sinensis grafted on Carrizo citrange, Forner-Alcaide no. 5, and Volkamer lemon ( Figure 1; Table S1).

VOC Chemical Analysis
VOC collection with headspace Solid-Phase Micro-Extraction (SPME) method followed by GC-MS analysis revealed the presence of a total of fifty-six VOCs released from the Table 2. Volatile emissions 1 of Citrus sinensis (CS) leaves grafted on the different rootstocks Citrus aurantium (CA), Volkamer lemon (CV), Carrizo citrange (CC), and Forner-Alcaide no. 5 (FO) collected using headspace SPME. The compounds found also in the rootstock without scion reported in Guarino et al. [16] are in italic. The leaf VOC emission recorded underlined quantitative differences across the different C. sinensis combinations investigated. The two first components in the PCA analysis showed around 89% to the overall variability among the four scion/rootstock combinations analyzed (see Material and methods section), with a strong separation among them ( Figure 2). Most of the compounds extracted were positively correlated with the first component (Dim1), while five compounds, three sesquiterpenes, one ester, and one alcohol (Z)-3-hexenol, the only one compound recorded in this category; Table 2), respectively, were negatively correlated. A clear discrimination across combinations according to their rootstocks and susceptibility to CTV infection was highlighted ( Figure 2). Indeed, the four samples are located in four different quadrants: the C. sinsensis samples (CS) grafted on CTV-tolerant rootstocks obtained from trifoliata parents, Carrizo citrange (CC) and Forner-Alcaide no. 5 (FO), were in the right quadrant, in the upper and lower one, respectively, while the samples grafted on Volkamer lemon (CV), and on the CTV-susceptible rootstock C. aurantium (CA), were in the upper left and lower left quadrant, respectively. Fifty-five metabolites, about 90% of the total extracted, were shared between FO and CC ( Figure 2; Table 2). Interestingly, two sesquiterpenes, α-farnesene and cis-β-farnesene, were correlated to the quadrant where CA was located, while β-sesquiphellandrene, hexenyl-acetate, and (Z)-3-hexenol were found in the CV's quadrant ( Figure 2).  A heatmap based on metabolite profiles isolated was able to separate the four combinations in two main distinct groups ( Figure 3). The first cluster enclosed the samples grafted with two CTV-tolerant genotypes CC and FO, while in the other one, in two different branches, clustered the samples grafted on CA and CV, the CTV-susceptible and -tolerant samples, respectively ( Figure 3). Limonene, sabinene, carene, trans-β-ocimene, linalool, and citronellal, all compounds belonging to monoterpene class, were the compounds showing the highest difference in the amount across samples (Figure 3).  PCA and heatmap analyses were able to underline a panel of compounds (sabinene, carene, limonene, trans-β-ocimene, linalool, citronellal, camphene, β-pinene, hexenylacetate, cis-β-farnesene, and α-farnesene) showing the highest discrimination power among combinations. ANOVA followed by Tukey's test carried highlighted that CV + CS emitted a significantly lower amount of sabinene (F = 3.61; df = 3; p < 0.05) and trans β-ocimene (F = 4.29; df = 3; p < 0.05) in comparison to the other combinations. Finally, comparing the VOC profiles obtained from C. sinensis leaves of the four combinations investigated here (CA + CS, CC + CS, CV + CS, and FO + CS) to the compounds extracted from the leaves of each rootstock used (CA, CC, CV, and FO), previously reported in Guarino et al. [16], a modulation of rootstocks on the products emitted from C. sinensis (CS) was highlighted ( Figure 4). Indeed, an overlap between the profiles of rootstocks and their combinations with CS was underlined ( Table 2), but without a correlation in the amount of molecules emitted. Among combinations, CA + CS showed the highest level (40.3%) of shared compounds with its rootstock, followed by CV + CS (36.1%), FO + CS (23.4%), and CC + CS (22.2%) (Figure 4).  [16], a modulation of rootstocks on the products emitted from C. sinensis (CS) was highlighted ( Figure 4). Indeed, an overlap between the profiles of rootstocks and their combinations with CS was underlined ( Table 2)

Discussion
The finding of tolerant alternative citrus rootstocks to Citrus Tristeza Virus (CTV), maintaining an adequate high yield and quality guaranteed by Citrus aurantium, is a priority to manage and overcome this disease and/or its vectors.
In the present study, it was found that CTV-susceptible and -tolerant rootstocks in C. sinensis scions elicited a different VOC profile from leaves and determined a specific attraction response toward A. gossypii, the main virus vector in Mediterranean areas, with

Discussion
The finding of tolerant alternative citrus rootstocks to Citrus Tristeza Virus (CTV), maintaining an adequate high yield and quality guaranteed by Citrus aurantium, is a priority to manage and overcome this disease and/or its vectors.
In the present study, it was found that CTV-susceptible and -tolerant rootstocks in C. sinensis scions elicited a different VOC profile from leaves and determined a specific attraction response toward A. gossypii, the main virus vector in Mediterranean areas, with an Recently, other studies have attempted to evaluate host preference in terms of the feeding and oviposition of other disease vectors toward un-grafted different citrus rootstocks. For example, it was observed that the psyllid Trioza erytreae (Hemiptera: Triozidae), a vector of Huanglongbing (HLB), prefers to orient toward Carrizo citrange, while it avoids Forner-Alcaide no. 5 [32]. Urbaneja-Bernat and co-workers [33] indicated that the Carrizo citrange rootstock is also highly vulnerable to the psyllid Diaphorina citri Kuwayama (Hemiptera: Liviidae) in comparison with other HLB vectors. Unlike the previous studies reported, our observations were carried out on grafted plants, the stage more likely to be attacked by the CTV vectors.
The behavioral observations on A. gossypii indicated that the aphids prefer to orientate and develop more frequently toward C. sinensis plants grafted onto C. aurantium rather than the other rootstocks. Due to the marked differences in the VOCs profiles of scion/rootstock combinations investigated, it is possible to argue that such variations could be the basis of the avoidance mechanism (antixenosis) observed toward the aphid. However, we cannot establish if other morphological or physiological factors might concur in the observed response. Indeed, our behavioral observations were carried out in semi-field conditions; therefore, the aphids were exposed to olfactory, visual, and morphological cues rather than only to the plant's VOCs through an olfactometer instrument.
The relationships between aphids and their host plants are complex. In fact, the recognition of a plant as a suitable host and subsequent feeding initiation by an aphid depends on a composite interaction between aphid and plant traits [34]. This process of selection, settling, and feeding establishment on a new host plant is critical from the perspective of an aphid's fitness, but is also a focal point in the evolution of plant resistance toward piercing-sucking pests [34,35]. Plants rely on unique mechanisms of recognition, signaling, and defense to cope with the specialized mode of phloem feeders as aphids [34]. The ability to locate and recognize host plants is essential for the survival of aphids, and this ability can be mediated by several factors, and among them, olfactory cues play a key role [36]. In particular, it has been observed that winged aphids exploit both visual and volatile cues to detect potential host plants [37], and the landing responses may be elicited by plant volatiles detected by antennal olfactory sensilla [38][39][40].
After settling on a host plant, aphids may assess its suitability based on surface molecules, including lipids and secondary metabolites [41]. Aphids then broadly probe and salivate into the plants, and often reject non-hosts after initial sampling of epidermal cell contents, or subsequent sampling of mesophyll cell contents [42]. Previous studies suggested that aphids do not show clear discrimination between host and non-host plants until they have landed and inserted the stylets [43,44]. Powell and colleagues [42] observed that, even if field and laboratory studies indicate that the major features influencing plant preference by aphids are perceived after stylet insertion, the cues detected by aphids before stylet insertion undoubtedly play a key role in the host selection process. Over the past decades, aphids have also been shown to acquire olfactory information on plant quality in contrast to in flight or when walking toward a plant, so host selection does indeed already begin pre-alighting [37,45]. In some cases, individual VOCs are used by aphids as host cues, and in others, specific VOCs act as non-host cues during host finding; for example, isothiocyanates can attract brassica-related aphids and discourage the other species [45][46][47]. However, in the majority of cases, the relative concentration of chemicals in a mixture of VOCs influences the host plant searching behavior of aphids, rather than one or two VOCs acting as key attractive compounds [48][49][50]. In fact, past studies in which insects have been exposed to plant-produced chemicals alone or in combination have showed that appropriate blends or combinations of VOCs determine stronger behavioral responses than single compounds [49,50].
In accordance, the data obtained in our study can suggest that the changes in the ratio of the VOCs emitted from different scion/rootstock combinations might influence the aphid preference behavior toward plants grafted on C. aurantium. In fact, the VOC profiles recorded from the leaves of C. sinensis grafted on the four different rootstocks showed differences in the compounds produced, mainly quantitative. Interestingly, among the compound observed, the volatiles belong mostly to terpenes, in agreement with the finding reported by Guarino and co-workers [16] from the rootstocks' leaves investigated here. Similarly, a previous study [51] analyzing citrus leaves obtained from the Sugar Belle mandarin hybrid on three different rootstocks suggested that rootstock choice influences the overall scions VOCs profile, and this can enhance its defense by repelling vectors or for signaling to their natural enemies or parasitoids. In addition, in agreement with our findings, the differences were mainly related to the monoterpenes [51].
A possible explanation of aphids' preference orientation toward plants grafted on C. aurantium could also be linked with their ability to use plant VOCs to discriminate between the suitability of different plants within the same species [45]. In fact, C. sinensis plants grafted on sour orange might have a higher vigor and a consequently higher number of young sprouts and leaves, the tissue mainly attacked by aphids, rather than the plants grafted on the other rootstocks tested in this study. The incidence of A. gossypii can be influenced by anatomical traits such as the age of shoots, or metabolic profiles, which can influence plant-insect-natural enemy interactions [50][51][52][53][54].
The higher presence of some VOCs in C. sinensis plants grafted on CTV-tolerant rootstock rather than the ones grafted on sour orange could suggest that such chemicals might concur in the avoidance mechanism observed toward A. gossypii. Sabinene, less abundant in the VOCs of the C. sinensis plant grafted on Volkamer lemon, is reported in the essential oils of Schinus terebinthifolius Raddi leaves, exhibiting anti-acetylcholinesterase activities toward the related Aphis nerii Boyer de Fonscolombe [55]. As mentioned above, we cannot establish if the avoidance mechanism exhibited by A. gossypii toward plants grafted on the CTV-tolerant rootstocks is determined by the not-landing of the individuals on the plants (antixenosis) or by their escape after probing/feeding of the plants (antibiosis). It is possible that both mechanisms can occur in synergistic combination as evidenced in Brassica juncea-fruticulosa toward the aphid Lipaphis erysimi (Kaltenbach) [56].

Plant and Insects
Two-years-old plants of C. sinensis (CS) grafted on C. aurantium (CA), Volkamer. lemon (CV), Carrizo citrange (CC) (C. sinensis × P. trifoliata), and Forner-Alcaide no. 5 (FO) were provided from Vivai Maimone Giuseppe Alessio located in Milazzo (Messina-Italy). Citrus plants were produced and maintained in a greenhouse at 25 ± 5 • C and 50-70% RH under natural photoperiod. Four plants of C. sinensis per different rootstock were grown on a substrate consisting of sand and peat (1:1) in 20 L cylindrical plastic containers and were watered 3 times per week and fertilized using alternating ratios of 3.1.1 and 1.3.1 (N.P.K). One month before starting the experiments, the plant material was transferred to IBBR Institute (Palermo-Italy) and maintained in shadow recovery condition until being used for collection and analysis of plant secondary metabolites.
An A. gossypii colony was established from infested Hibiscus rosa-sinensis L. plants. Insects were maintained in an environmentally controlled room (28 ± 2 • C, 70 ± 10% relative humidity, photoperiod 16L:8D) in wooden cages (25 × 25 × 40 cm) with two 5 cm diameter mesh-covered holes for ventilation. Aphids used in the experiments were both winged and wingless, mixed-instar individuals and were collected from the cultures immediately prior to bioassay.

VOC Chemical Analysis
Chemical analysis of the plant secondary metabolites for C. sinensis grafted on the different rootstocks (see "4.1 Plant and Insects" section) was carried out through the headspace SPME method [58], followed by gas chromatography and mass spectrometry (GC-MS). VOCs emitted from CS grafted on CA, CV, CC, and FO were separately collected in the headspace by using SPME. The stationary phase used as the coating was polydimethylsiloxane (PDMS, 100 µm) (Supelco, Bellefonte, PA, USA). A manual SPME holder from the same manufacturer was used for injections. Fibers were conditioned in a gas chromatograph injector port as recommended by the manufacturer at 250 • C for 30 min.
For the headspace collections, a one-year-old leaf chosen randomly from each plant grafted on the different varieties investigated (CA + CS, CC + CS, CV + CS, and FO + CS) was cut at the base of the petiole and immediately covered with parafilm to minimize the VOC emission from the point of cutting. Leaves were then gently placed by forceps into 22 mL glass vials, weighted using a precision balance, and sealed with a polytetrafluoroethylene silicon septum-lined cap (Supelco, Bellefonte, PA, USA). Subsequently, an SPME needle was then inserted through the septum and volatiles were absorbed on the exposed fiber for 5 min at controlled room temperature (22 ± 1 • C). Therefore, the experiments were carried out using four biological replicates by sampling leaves each from individual trees of C. sinensis grafted on the four different rootstocks (CA + CS, CC + CS, CV + CS, and FO + CS), for a total of 16 samples (4 plants x 4 rootstock).
In order to perform the chemical analysis of the collected VOCs, the loaded fiber was desorbed in the gas chromatograph inlet port for 1 min immediately after the end of the sampling time. Coupled GC-MS analyses of the headspace collections from the four plant species were performed on an Agilent 6890 GC system interfaced with an MS5973 quadruple mass spectrometer equipped with a DB5-MS column in splitless mode. Injector and detector temperatures were 260 • C and 280 • C, respectively. Helium was used as the carrier gas. The GC oven temperature started at 40 • C and then increased by 10 • C/min to 250 • C, with initial and final hold times of 5 and 20 min, respectively. Electron impact ionization spectra were obtained at 70 eV, recording mass spectra from 40 to 550 amu. Peak integration was carried out using ChemStation software.

Behavioral Bioassays
To investigate the Aphis gossypii preference response to C. sinensis grafted on the four different rootstocks, a semi-field bioassay was carried out in an experimental cage. The cage setup for the experiment, size 2.00 × 1.05 × 1.60 m, was made by a wood structure covered with a tissue-not tissue 17 g/mq (VERDEMAX ® , Boretto (RE)-Italy). Sixteen plants, four per rootstock, were used by dividing them into four blocks with one treatment per block with a distance of 20 cm to each other ( Figure S2), while, inside the block, the distance between the pots was 10 cm. Each plant was randomly placed in each block and their position followed a ten-days clockwise rotation.
At the start of the experiment and every ten days thereafter, cohorts of mixed-stage aphid individuals (approx. N = 100) were placed at the center of each block. Plants were then inspected every ten days (three samplings in total) to count the number of leaves per plant infested by A. gossypii individuals. After each inspection, aphids were not removed from the infested leaves, to resemble more realistically a field condition. The experiment lasted 30 days, from 15 May to 14 June 2022.

Statistical Analysis
VOC profiles recorded from each scion/rootstock combination were used to develop a PCA (Principal Component Analysis) using the R package FactoMiner [59] and factoextra (https://CRAN.R-project.org/package=factoextra; accessed on 13 July 2022). PCA allowed us to investigate what variables were able to separate the four samples investigated. A heatmap with the metabolites profiles for each variety was carried out by heatmap.2 in R package gplots (https://github.com/talgalili/gplots; accessed on 13 July 2022). The compounds that, in PCA and heatmap analyses, were able to discriminate the four combinations were further investigated by using a one-way ANOVA followed by Tukey's test. A two-way ANOVA and Tukey's multiple pairwise-comparisons were developed using R studio (https://www.rstudio.com; accessed on 13 July 2022) to verify the significance levels of behavioral bioassays results. Finally, a VENN diagram was also used to easily highlight the number of metabolites shared across C. sinensis grafted on 4 specific rootstocks studied here (see section "4.1 Plant and Insects") and the VOC profiles previously extracted by Guarino and colleagues [16] from the rootstocks investigated. The diagram was developed using the R package VennDiagram (https://cran.r-project.org/package=VennDiagram; accessed on 13 July 2022).

Conclusions
This study showed that the C. sinensis plant grafted on CTV-tolerant rootstocks exhibited different VOC profiles in comparison with plants grafted on the CTV-susceptible sour orange. The behavioral bioassays with A. gossypii, the main CTV vector in Mediterranean countries, indicated that aphids prefer to orient and develop on plants grafted on CTV-susceptible rootstocks. This avoidance mechanism induced by the rootstocks Volkamer lemon, Forner-Alcaide no. 5, and Carrizo citrange could represent one of the co-factors at the basis of the tolerance toward the virus. Future studies will focus on the response of C. sinensis plants grafted on the CTV-tolerant rootstock after aphid feeding, through high-throughput approaches, also using other combinations with new promising alternative rootstocks [6], in order to provide useful solutions for the A. gossypii control and, consequently, the CTV disease in citriculture.