Jasmonic Acid and Salicylic Acid Levels in Defense Response of Azalea ( Rhododendron simsii Hybrid) to Broad Mite ( Polyphagotarsonemus latus )

: The broad mite ( Polyphagotarsonemus latus (Banks)) is an important pest in many crops, including azalea ( Rhododendron simsii Planch. hybrid). Broad mites cause the malformation of shoot tips, leaves and ﬂowers in azalea. It is known that the jasmonic acid (JA)- and salicylic acid (SA)-dependent signaling pathways are related to the presence of herbivorous mites. Here, we describe the levels of the two main plant defense-related hormones, SA and JA, in mite-infested plants. The plant hormones were analyzed using liquid chromatography combined with tandem mass spectrometry (LC-MS/MS). We studied both short-term hormonal responses under controlled conditions with artiﬁcial inoculation, and long-term responses under culture conditions with natural infestation. The long-term development of broad mite populations and hormone response were studied during two subsequent growing seasons on 3 and 18 different cultivars, respectively. During the experiments on 18 azalea cultivars under natural infestation, the presence of different species of tarsonemid mites was also examined. JA concentrations only showed variation in the early phase of infestation. Subsequently, the SA levels increased signiﬁcantly for all the cultivars where broad mites were detected. Based on the observed timing of the defense responses, we suggest that the interaction of the JA and SA pathways as a defense response for pot azalea against P. latus involves a primary plant response through the JA pathway. In the presence of the mites, the production of SA increased in the plants in a later phase as the P. latus population grew. Our results also show that the hormone response depends on type of mite. Changes in hormone levels were found upon infestation with P. latus , but not in the presence of another frequently occurring tarsonemid mite, Tarsonemus confusus Ewing.


Introduction
Pot azalea, commercially sold as a flowering indoor pot plant, is the main ornamental plant produced in Belgium. In Europe, the main producers of azalea and Rhododendron are Belgium and Germany, with export values of EUR 36 and 7 million, respectively. The modern pot azalea traces its roots to at least four species, the most important being Rhododendron simsii Planch. The introduction of R. simsii in Belgium dates back to 1818 [1]. Other parental species are Rhododendron indicum (L.) Sweet, R. mucronatum G.Don and R. scabrum G.Don [2,3]. The production area in Belgium is around 230 ha, and almost all companies are located around Ghent [4,5]. Pot azalea is propagated by cuttings and its production cycle takes almost two years, combining greenhouse cultivation with outdoorcontainer field cultivation during summer.
The broad mite, Polyphagotarsonemus latus (Banks) (Acari: Tarsonemidae), is an important pest on many commercial crops and this mite has been reported in most parts of the world. The first record in Europe dates back to 1961 when the broad mite was found in Italy [6]. Since then, broad mites have been found in other European countries, and in Belgium, it was reported on pot azalea in the mid-1980s [7]. Restrictions on the use of broad-spectrum pesticides are the main reason for the significant increase in the pest status of broad mites in pot azalea in recent years. The broad mite causes esthetic damage; on azalea, it is responsible for the bronzing, browning and curling of apical leaves, and can cause leaves to become brittle, rigid and stunted with the edges curled downwards. Flowers also become malformed and discolored. Broad mites are very small (<200 µm) and can only be detected using a microscope. A small number of 20 mites or less can cause serious economic damage in crops such as pepper [8,9], cucumber [10], lime [9] and chili [9,11]. Also in pot azalea, it has been shown that low mite numbers can lead to high damage rates [12]. Even after treatment, symptoms may continue to develop [13] and damage initiated in a plant progresses even when mites are no longer present. In his review on broad mites, Gerson [13] suggested that the symptoms are caused by toxins. This conclusion was based on the persistence of symptoms for some weeks post-treatment. These putative toxins would explain the appearance of injuries very shortly after the initiation of a mite attack, and the later occurrence of symptoms at a distance from the site of mite feeding.
Differences in susceptibility to broad mites have been observed in pot azalea, along with the damage rate, and in relation to the population sizes of P. latus on the tested azalea cultivars [12]. Trichomes (plant hairs) are potentially involved in broad mite resistance in pot azalea [14]. Additionally, in Capsicum species, trichomes are known to act as a limiting factor for the development of broad mites [15]. In plants, trichomes can act as a physical barrier and chemical repellent to insect or mite attacks [16]. In pot azalea, four types of trichome are described. All pot azaleas have long non-glandular trichomes; in addition, there are three types of short trichomes. They are thorn-like trichomes, short non-sticky trichomes and short trichomes with glandular ends (plants with these trichomes also have sticky leaves). There are indications that the latter are related to broad mite resistance. However, this is not the only discriminative trait, as in certain susceptible cultivars, trichomes with glandular ends were observed [17]. Recently it was shown that wounding or exogenous jasmonic acid (JA) treatments are able to promote trichome development and enhance insect resistance [16].
For decades, researchers have already known that plants have a defense network dependent on phytohormones, which translates stresses of biotic or abiotic origin into plant defense responses. Jasmonic acid (JA) and salicylic acid (SA) are known to serve as the primary signals of the plant's immune response. SA and JA work antagonistically and use other hormones to fine-tune immune responses built on SA and JA [18].
SA is most recognized as primarily being induced in the mediation of attacks by biotrophs [19], whereas JA is involved in defense against necrotrophic pathogens [20] and herbivores [21]. As reviewed by Pieterse et al. [18], well-known trade-offs exist between SAdependent resistance and JA-dependent defense in disease and pest resistance. However, neutral and synergistic interactions have also been described [18]. Most studies focus on plant pathogens, but in recent years, a large number of studies have also examined plant defense responses for herbivores. For example, in a study on tomato and the two-spotted spider mite, different types of interaction have been described. These different interaction types, including direct and indirect defenses, were reviewed by Blaazer et al. [22]. Their review shows that both JA and SA are related to the presence of herbivorous mites. In brief, (1) the accumulation of SA and JA can be triggered in tomato as a plant response at the start of mite infestation; (2) the levels of both JA and SA can decrease as a result of defense suppression by the mites; and (3) intraspecific variation in natural mite populations, especially in polyphagous mites, can have an effect on the plant-mite interaction. Not only is the herbivore-plant relationship involved, but so is the attraction of natural enemies by the emission of volatile compounds. Specifically, for P. latus in cucumber leaves, both the JAand SA-dependent pathways are activated, and possibly an oxidative stress response [23]. To examine the possible involvement of the JA signaling pathway in the defense response against P. latus, [23] evaluated feeding on wild-type (WT) tomato 'Castlemart' and on def-1, a mutant impaired in JA biosynthesis. In the WT plants, broad mite feeding did not affect growth or leaf development, but in the def-1 plants, 50-60% inhibition was observed. Mite populations developed only on def-1 plants, as was shown by progeny counts. Furthermore, P. latus prefers JA-defective mutants, and therefore, is able to actively discriminate between resistant and susceptible plants [23,24]. Therefore, it was concluded by [23] that the JA pathway is likely to play an important role in the resistance response against P. latus [23].
Different JA and SA biosynthesis marker genes in azalea have been studied [25]. To obtain insight into the molecular mechanisms behind the interaction between P. latus and the R. simsii hybrid, the expression levels of genes involved in the infestation of leaves with broad mites were analyzed. The focus was on JA and it was shown that transcript levels of marker genes for the JA biosynthesis pathway were significantly induced upon P. latus infestation in the R. simsii hybrid.
In this research, we aimed to study the relationship between the presence of broad mites and plant defense hormone levels in pot azalea (R. simsii hybrid). The hormonal responses were evaluated shortly after infestation, but the long-term responses were also analyzed. Another aim was to evaluate the variation in susceptibility to broad mites in different pot azalea cultivars. Additionally, we also wanted to observe other mites, such as Tarsonemus confusus, on pot azalea, for which it is uncertain if they should be considered harmful. In a controlled experiment and immediately after infestation with broad mites, the JA and SA levels in azalea plants were analyzed. Two cultivars with well-known resistance-'Nordlicht' (susceptible cultivar) and 'Elien' (resistant cultivar)-were used [12]. The levels of SA and JA were compared between infested and uninfested plants.

Materials and Methods
The plants were grown on ebb and flow tables. The plants were infested artificially by surrounding them with Hedera helix L. infested with broad mites, as described in Luypaert et al. [25]. The samples for LC-MS/MS analysis and for the quantification of the number of mites were taken as shoot tips at 4, 6, 8 and 11 days post-inoculation (dpi). Five independent biological replicates were taken at each time point (n = 5), with one shoot tip per replicate for LC-MS/MS analysis and another for mite counts. The samples for LC-MS/MS analysis were flash-frozen in liquid nitrogen and stored until analysis at −80 • C. The broad mites were extracted from the sampled shoots following the three-step detection method including (1) sampling of shoot tips, (2) isolation of the mites in 70% ethanol and (3) sieving and vacuum filtration to facilitate counting [26]. Subsequently, the mites were counted using a binocular microscope (Olympus SZX16, Antwerp, Belgium; max 115×). In experiment 2, the natural occurrence of broad mites was monitored and correlated to the hormone levels. Three azalea cultivars were studied during a growing season of 32 weeks. The three cultivars used-'Nordlicht' (susceptible cultivar), 'Mevrouw Gerard Kint' (intermediate response) and 'Elien' (resistant cultivar)-differed in their susceptibility to P. latus: [12].
Each cultivar had one replicate with 100 plants and was placed on a separate greenhouse bed comparable to the setup in commercial cultivation. Luypaert et al. [12] tested a randomized cultivar setup and a setup grouped per cultivar. They concluded that randomization leads to a bias caused by the attractiveness of different azalea cultivars to broad mite; thus, in the present experiment, the plants were grouped according to cultivar. In experiment 2, no tarsonemid mites were artificially introduced into the crop. Instead, the natural distribution of P. latus was monitored closely.
The plants were sampled between 19 January (W3) and 13 September (W37). Every two weeks, samples were taken for the estimation of mite numbers. The dates and week (W) numbers of sampling were: 19 January According to standard cultivation procedures, the plants were cut back in W5 (after sampling), W20 and W35 (after sampling). Sampling for mite counts stopped with the last cut in W35. One extra sample for defense hormone analysis was taken from the fresh shoots in W37.
Broad mites were extracted and counted as described in experiment 1. The morphological identification of mites was conducted using a light microscope (Leica DM2000, Diegem, Belgium; max 400×).

Experiment 3-Defense Hormone Analysis under Natural Infestation in 18 Cultivars of Pot Azalea and Occurrence of Mite Species during the Growing Season (30 Weeks)
In total, 18 commercial cultivars of pot azalea were used in the experiment. Eight of these cultivars were previously tested for susceptibility [12] (Table 1). For each replicate, 24 plants of one cultivar were placed in a greenhouse. To avoid dispersion of the mites between cultivars, sticky plates (Biobest Bug Scan IVOG system-yellow) were installed between the blocks of plants. No tarsonemid mites were introduced in the crop, but the natural distribution of the different species of tarsonemid was monitored closely.
To estimate mite numbers, the plants were sampled every two weeks between 20 October (W42) and 9 May (W19). On four dates-1 February (W5), 6 March (W10), 28 March (W13) and 27 April (W17)-shoot tips were sampled for JA and SA analysis. Sampling for defense hormones was conducted according to experiment 2: three independent biological samples were taken per cultivar and per sampling date with 3 to 5 shoot tips per sample of the different plants (n = 3). Tarsonemid mites were sampled (3 shoot tips per cultivar) and counted using a binocular and identified morphologically using a light microscope, as described in experiment 2. Damage was calculated per cultivar by counting the number of plants with damage.
In this experiment, we looked in more detail at the different mite species known for their occurrence on pot azalea. Besides Polyphagotarsonemus latus, identification keys were used for four Tarsonemus species: Tarsonemus confusus, T. bilobatus Suski, T. floricolus Canestrini & Fanzago, T. lacustris Schaarschmidt, species of the genus Xenotarsonemus Beer, and the species Phytonemus pallidus (Banks). For the identification of these mite species, all the collected specimens were mounted on slides for microscopic observation [27][28][29]. Table 1. Rhododendron simsii hybrids used in the different experiments; cultivars marked with * are also used in Luypaert et al. [12].

JA and SA Analysis
In experiment 1, the extraction and analysis of phytohormones SA and JA were conducted according to Bosco et al. [30]. In brief, at −80 • C, frozen leaf samples were kept in liquid nitrogen and were ground using a mortar and pestle before 20 mg of homogenized sample material was extracted with 2 mL solvent mixture methanol:water:acetic acid (10:89:1) and spiked with 2H 4 -SA (Olchemim, Olomouc, Czech Republic) and 2H 6 -JA internal standards (Olchemim, Olomouc, Czech Republic). The samples were extracted at 4 • C on a shaker (210 rpm) for 16 h. Subsequently, the samples were vortexed and filtered using a 0.22 µm Millex-GV filter (Millipore, Billerica, MA, USA), and the extract was collected in glass tubes. Then, 1 mL of the extract was loaded onto Oasis MCX 1 cm 3 cartridges (Waters, Milford, MA, USA) (preconditioned with 1 mL pure methanol and 1 mL milliQ water). After loading the samples, 0.5 mL milliQ water was used to wash the cartridges. The flow-through was discarded and the columns were eluted under vacuum with 1 mL methanol. In a warm water bath (35 • C, nitrogen atmosphere), the eluates were evaporated to dryness. For resuspension of the eluates, 0.5 mL 75% methanol was added. The extract was again filtered through a 0.22 µm Millex-GV filter and transferred to a micro-vial prior to the injection of 10 µL into the LC-MS/MS. External calibration curves were used to calculate the final concentrations of the samples for JA (Olchemim, Czech Republic) and SA (Sigma-Aldrich, Bornem, Belgium).
In experiments 2 and 3, extraction was based on Sanchez-Bel et al. [31] using diethyl ether as solvent. Before analysis, the frozen samples were ground and lyophilized. All the steps were performed on ice. Per sample, 20 mg of lyophilized plant material was placed in a glass tube and spiked with internal standard 2H 4 -SA (Olchemim, Olomouc, Czech Republic) and 2H 6 -JA internal standard (Olchemim, Czech Republic) prepared in MeOH/water (75/25). Subsequently, 2.5 mL ultrapure water was added. The samples were shaken on ice for 30 min (150 rpm) and centrifuged (cooled; 5000× g) for 40 min. The supernatant was adjusted to pH 2.8 with 6% acetic acid. The samples were partitioned twice against diethyl ether. Subsequently, the total eluate was evaporated using nitrogen gas at 35 • C, and the residues were dissolved in 1 mL water/methanol 25:75 and put on a 0.22 µm PVDF filter. The filtrate was put in a micro-vial and stored at 4 • C prior to injection (10 µL) into the UHPLC−MS/MS system. In experiments 2 and 3, the level of JA present was calculated using a calibration curve (relative), but without the use of an internal standard.
The advantage of the protocol based on Sanchez-Bel et al. [31] in experiments 2 and 3, compared to Bosco et al. [30] used in experiment 1, is that the extraction is conducted on dry plant material. This results in lower variation in the measured values. Moreover, the protocol using dry plant material is more simple and makes no use of expensive solid-phase extraction columns that can also compromise the recovery of some compounds. Therefore, in experiment 1, the hormone levels are expressed as fresh weight (fw), and in experiments 2 and 3, the hormone levels are expressed as dry weight (dw).
For all the experiments, the analysis was run on an Acquity UHPLC system coupled with a Xevo TQ-S triple quadrupole MS detector (Waters, Milford, MA, USA), as described in Bosco et al. [30].

Statistical Data Analysis
The LC-MS/MS data were acquired using MassLynx, version 4.1 (Waters, Milford, MA, USA), and data processing of the acquired LC-MS/MS data was performed using Targetlynx (Waters, Milford, MA, USA). The types of statistical test used are mentioned with the different experiments in the Section 3 and in tables and graphs.

Experiment 1-Defense Hormone Analysis after Broad Mite Infestation under Controlled Conditions in Two Cultivars of Pot Azalea Shortly after Infestation (Up to 11 Days)
In experiment 1, the phytohormone level was determined in R. simsii 'Elien' and 'Nordlicht' at four time points upon artificial infestation. On 'Nordlicht', the first mites were observed at 6 dpi, while on 'Elien', the first observation was at 8 dpi. At 11 dpi, the final day of observations, the number of mites on 'Nordlicht' increased to 12.4 ± 4.4, while on 'Elien' at the same time point, 1.4 ± 1.2 mites were counted. Figure 1 shows the levels of JA over the course of 11 days. The infestation of R. simsii 'Nordlicht' had no effect on the JA level ( Figure 1). In contrast, 'Elien' had a significantly higher level of JA when comparing the P. latus-infested plants to the control plants (mock) at 6 dpi. The SA levels are expressed as the relative response: area analyte/area internal standard. As sample preparation required ground fresh leaves, the hormone levels are expressed as fresh weight. The levels of free SA in 'Nordlicht' were increased by 1.6, 10.4, 6.3 and 18.7 at 4, 6, 8 and 11 dpi, respectively ( Table 2). Statistical differences were only The SA levels are expressed as the relative response: area analyte/area internal standard. As sample preparation required ground fresh leaves, the hormone levels are expressed as fresh weight. The levels of free SA in 'Nordlicht' were increased by 1.6, 10.4, 6.3 and 18.7 at 4, 6, 8 and 11 dpi, respectively ( Table 2). Statistical differences were only observed at 6 and 11 dpi (Mann-Whitney U, p < 0.05). In contrast, in 'Elien' the level of free SA remained relatively stable. Across all the time points after P. latus infestation, the increase in the level of free SA was 1 to 1.6 times baseline; these differences were not statistically significant (Mann-Whitney U, p < 0.05). The plants were examined for the presence of mites from January onward, but it took until W19 for the first P. latus mite to be observed on 'Mevrouw Gerard Kint'. A large number of 28 mites was found on 'Nordlicht' in W25 (Figure 2). In W29, large numbers of mites, 36.3 (SD ± 48.5), were found on 'Mevrouw Gerard Kint'. On 'Elien', during the whole experiment, only a limited number of two mites was counted (W29). No mites were extracted on the other dates. When symptoms of mite infection were observed in a cultivar, this was always locally at first, and subsequently spread to a larger area.
A significant increase in SA was found in 'Mevrouw Gerard Kint' and 'Nordlicht' after observations of mite infestations. In 'Elien', the levels of SA were very low during the whole growing period (Figure 2). The differences between the plants of 'Nordlicht' with and without visual symptoms were not significant in W31. In W33 and W35, the differences between symptomatic and asymptomatic plants were significant for both 'Mevrouw Gerard Kint' and 'Nordlicht' (t-test; p < 0.05). The levels of JA were low in April but increased in all three cultivars from W19 onward. From W35, a decrease in JA levels was observed.
until W19 for the first P. latus mite to be observed on 'Mevrouw Gerard Kint'. A lar number of 28 mites was found on 'Nordlicht' in W25 (Figure 2). In W29, large numbers mites, 36.3 (SD ± 48.5), were found on 'Mevrouw Gerard Kint'. On 'Elien', during t whole experiment, only a limited number of two mites was counted (W29). No mites w extracted on the other dates. When symptoms of mite infection were observed in a cu var, this was always locally at first, and subsequently spread to a larger area. A significant increase in SA was found in 'Mevrouw Gerard Kint' and 'Nordlic after observations of mite infestations. In 'Elien', the levels of SA were very low duri the whole growing period (Figure 2). The differences between the plants of 'Nordlic with and without visual symptoms were not significant in W31. In W33 and W35, t differences between symptomatic and asymptomatic plants were significant for both 'M vrouw Gerard Kint' and 'Nordlicht' (t-test; p < 0.05). The levels of JA were low in Ap but increased in all three cultivars from W19 onward. From W35, a decrease in JA lev was observed.

Experiment 3-Defense Hormone Analysis under Natural Infestation in 18 Cultivars of P Azalea and Occurrence of Mite Species during the Growing Season (30 Weeks)
Different mite types were detected during the experiment. Only four mite spec occurred; in order of prevalence (total numbers counted during the experiment are give P. latus (counted mites: 3889), T. confusus (counted mites: 552), Xenotarsonemus sp (counted mites: 34), P. pallidus (counted mites: 10) and T. floricolus (counted mites: 4). N specimens were found for T. bilobatus or T. lacustris.
No correlations (Pearson) were found between the mite numbers of P. latus and confusus. Nevertheless, we observed that the cultivars with no observations of P. latus a

Experiment 3-Defense Hormone Analysis under Natural Infestation in 18 Cultivars of Pot Azalea and Occurrence of Mite Species during the Growing Season (30 Weeks)
Different mite types were detected during the experiment. Only four mite species occurred; in order of prevalence (total numbers counted during the experiment are given): P. latus (counted mites: 3889), T. confusus (counted mites: 552), Xenotarsonemus spp. (counted mites: 34), P. pallidus (counted mites: 10) and T. floricolus (counted mites: 4). No specimens were found for T. bilobatus or T. lacustris.
No correlations (Pearson) were found between the mite numbers of P. latus and T. confusus. Nevertheless, we observed that the cultivars with no observations of P. latus also showed low numbers of T. confusus. This results in a moderate Spearman's rank correlation: r = 0.55 (p < 0.01). Plant damage was caused by the occurrence of P. latus, as shown by a high Pearson correlation between the total number of P. latus mites and the number of plants showing damage: r = 0.89 (p < 0.0001). The same correlation calculated between T. confusus and plants showing damage was low and not significant.
The first P. latus mites were found at the start of the experiment in W42 on 'Leonardo'. The first symptoms were only observed in W47 on 'Franziska R.'. This cultivar showed increasing damage afterwards. Later, symptoms were also seen on the cultivars 'Michelle Marie', 'Fluostern', 'Mevrouw Gerard Kint' (red), 'Otto', 'Leonardo' and 'Renato'.
The highest total number was counted on 'Otto' (counted mites: 741), followed by 'Leonardo' (counted mites: 521) and 'Mevrouw Gerard Kint' (red) (counted mites: 451). All these cultivars showed infestation at an early stage of the experiment. On 'Michelle Marie', infestation was already present in December (W51), although the mite numbers did not increase considerable compared to the other cultivars. An overview of the occurrence of P. latus on the different cultivars is given in Figure 3.  In all the cultivars, the levels of SA increased in March (W9 to W13) compared to W5. A Pearson correlation was calculated between the maximum number of mites and the SA levels in W10. The correlation was high, with, respectively, r = 0.75 (p = 0.0002) and r = 0.84 (p < 0.0001) when the mite numbers of W9 and W11 were correlated with the SA levels on W10. Moreover, for all the cultivars, the correlation was very high, with r = 0.84 (p < 0.00001) when the total mite numbers across the entire experiment were correlated to the SA level found in W10.
Although in W13, the non-infected cultivars showed lower levels of SA, the correlation between the SA levels and mite numbers for all cultivars at this time point was still very strong (r = 0.85; p < 0.0001). Two cultivars showed symptoms only on part of the plant. Therefore, the samples taken from symptomatic vs. asymptomatic plants were analyzed separately. In both cultivars, huge differences in the SA level between symptomatic and asymptomatic plants were observed.

Discussion
The study of plant-arthropod interactions lags behind the study of plant-pathogen interactions. However, in recent years, plant-herbivore interactions have received increasing attention. For mites, plant reactions to the presence of the two-spotted spider mite (Tetranychus urticae Koch) have been studied most, as reviewed by Agut et al. [32] and Blazer et al. [22]. In the present study, we have focused on the relationship between the two main plant defense hormones, SA and JA, in response to broad mite presence in pot azalea. Our results are in line with the general accepted idea that the JA pathway is involved in defense against herbivores [21]. It is known that endogenous bioactive plant defense hormones such as JA are present in low levels in the plant, but accumulate upon herbivore feeding or wounding [33]. The results of experiment 1 under controlled conditions and of experiment 3 under natural infestation during the growing season showed that an increase in JA is only found at specific time points and early after infestation. In experiment 2, no significant changes in JA levels were observed. As suggested by Luypaert et al. [17], the timing of sampling is essential while measuring hormonal changes. Additionally, infestation pressure at the time of sampling plays a role in the observed hormone levels. The present study confirms the findings of Luypaert et al. [25], that P. latus infestation activates JA-dependent pathways most strongly at earlier time points (5-12 dpi) in R. simsii hybrids. In preliminary laboratory tests using floating leaf discs, it was shown that after the artificial application of 50 or 100 µM MeJA, the population growth of P. latus on R. simsii 'Nordlicht' significantly decreased [17]. Luypaert et al. [25] isolated different JA biosynthesis marker genes in azalea to obtain insight into the molecular mechanisms behind the interaction between P. latus and the R. simsii hybrid. Transcript levels of marker genes for the JA biosynthesis pathway were significantly induced upon P. latus infestation in the R. simsii hybrid [25]. This strengthens our hypothesis of an early-induced plant defense based on the JA response upon P. latus infestation in pot azalea.
The high increase in SA levels upon broad mite infestation is in contrast with the expected plant defense response. The SA pathway is often associated with defense against biotrophic pathogens [19]. Arena et al. [34] found, in Arabidopsis thaliana (L.) Heynh., a simultaneous increase in the levels of SA and JA when the plants were infested with falsespider mites of the genus Brevipalpus Donnadieu. They also evaluated mite performance on Arabidopsis mutants impaired in their SA or JA response and showed a function for the SA-mediated pathway in improved mite reproduction. Therefore, it is suggested that Brevipalpus mites manipulate the plant defensive response via the SA-mediated pathway for their own benefit [34].
In our study, significant increases in SA level were strongly related to the presence of broad mites. Levels remained low when no mites were present. In experiment 3 with 18 cultivars tested, an especially strong correlation was found between the rising number of mites in W9 (r = 0.75) and W11 (r = 0.84) and the SA level in W10. An increase of up to almost 100 times the SA level was measured in the susceptible cultivar 'Nordlicht' (experiment 2). These high SA levels where still observed during the following months of infestation. At the same moment, in 'Mevrouw Gerard Kint', a cultivar with moderate susceptibility, the increase in SA was lower, while in 'Elien', a resistant cultivar, no change in the SA level was observed.
In preliminary tests with a floating disc assay, the application of 50 or 100 µM SA had no significant effect on the broad mite population size [35]. Other preliminary data suggest that the application of SA can lead to increasing population sizes of broad mites, but this depends on the SA concentration applied. Additionally, the expression of SA marker genes isolated after P. latus infestation in the R. simsii hybrid was not clear [25]. Our SA analysis, however, could not confirm their conclusion that the SA biosynthesis pathway, led by either RsPAL or RsICS, was suppressed at later time points in response to P. latus infestation in pot azalea. If no local increase in transcript levels is observed, a systemic response of signaling molecules in plant defense for stronger and faster defense responses could be a hypothesis [36]. JA, for example, will not only accumulate at the sites of local wounding, but also in undamaged, more distal leaves. This increase can be found within two minutes [33]. For SA, more data are available in relation to pathogen-induced SA, which was also shown to act systemically in the plant and was found in the phloem [37].
In tomato, P. latus prefers plants deficient in the JA pathway compared to wild-type plants [38]. Treatment of the JA-deficient plants made them as unattractive as the wild-type plants. The wild-type plants infested by broad mites showed an increase in JA-related transcripts. In a study by Grinberg et al. [9] using Northern blot analysis, it was shown that both the JA and SA pathways are activated in response to broad mites on cucumber. In citrus, P. latus infestation activated both the JA-and the SA-dependent pathways in sour orange but no activation was observed in Cleopatra mandarin [39].
From our results, it is probable that in an early phase of infestation, the plant responds by activating the JA-dependent pathway; subsequently, the JA levels drop quickly and a switch is made to the production of SA. Because of the small timeframe, it is possible that the increase in JA was not observed in some of our experiments under natural infestation, as the exact moment of mite infestation is difficult to determine. In experiment 1, under controlled infestation conditions, it is shown that the JA level can change in a time period of only two days. Under culture conditions, as applied in experiments 2 and 3, not all cultivars and not all plants within one cultivar experience the same infestation pressure. Differences in JA content could be observed when symptomatic and asymptomatic plants were compared at the moment the mite population start to build up, as was the case for 'Michelle Marie' and 'Leonardo' (experiment 3). Because timing is crucial, further research would be needed to elucidate the switch from the JA-to the SA-dependent pathway.
The key question to be elucidated is that of whether P. latus is able to suppress plant defense by modulating the action of SA in response to JA. If so, a plausible answer might be that P. latus mites inject effectors, interfering with host immune responses. To date, no data for P. latus are available, but this phenomenon is known to occur in phytopathogens [40], aphids [41] and nematodes [42]; additionally, spider mites might secrete effectors via their saliva [43,44]. For example, strains of the spider mites Tetranychus urticae and T. evansi Baker & Pritchard can suppress the expression of SA and/or JA marker genes [21,[44][45][46]. The genomic data of P. latus, in accordance with the whole spider mite genome, would provide evidence if this mite species is also able to encode putative salivary proteins [47]. Another alternative might be that P. latus is a vector for phytopathogens that have been implicated in the suppression of plant defenses. The hypothesis of virus transmission by P. latus feeding on plants is experimentally rejected [48,49], still leaving the possibility of transmitting bacteria or fungi.
For whitefly, Bemisia tabaci (Gennadius), 2 under laboratory and semi-field conditions, examples are known of JA suppression in correlation with the induction of SA [50][51][52]. Negative crosstalk between the SA and JA defense pathways is often displayed. For example, the suppression of JA can be an effect of antagonistic crosstalk through strong SA induction [18]. However, suppression can also occur downstream of phytohormone accumulation and independently of JA-SA crosstalk [53].
The conclusion on the role of JA in the defense of pot azalea against broad mite is supported by the findings that the application of MeJA had a negative effect on the fitness of P. latus in the pot azalea cultivar 'Nordlicht' when compared to a control treatment [35]. This is in accordance with the reports of decreases in spider mite population growth after the application of MeJA [32,[54][55][56][57][58][59]. In pot azalea, the negative effect of MeJA on the population of P. latus was related to the concentration applied: for 50 and 100 µM MeJA, the population growth decreased by a factor 3.2 and 2.3, respectively [35]. Choh et al. [60] reported a similar effect as the number of eggs produced by T. urticae on lima beans (Phaseolus lunatus L.) differs in relation to the JA concentration applied. Additionally, in the T. urticae-sensitive Cleopatra mandarin, MeJA treatment significantly reduced the number of eggs produced by the spider mite, whereas salicylic acid had no effect on the oviposition of T. urticae [32].
For the eight cultivars used in the study of Luypaert et al. [12] and in experiment 3, correlations between the experiments were found for mite numbers and for damage rates, but the correlations were not significant. Previous studies on pot azalea and broad mite interaction showed that leaf damage does not always correlate well with the number of mites found on the plant [12]. However, our results regarding susceptibility confirmed the study of Luypaert et al. [12], with 'Elien' as the most resistant and 'Otto' being the most susceptible cultivar. Similar results were found for cultivars with a shared genetic background. 'Huelsten', 'Inka', 'Lunterra', 'Mevrouw Edmond Troch' and 'Nordlicht' all are sports of 'Helmutt Vogel'. In these five cultivars, the levels of SA were relatively low upon infestation by broad mites. For 'Michelle Marie' and its sport 'Franziska R.', both mite numbers and SA levels were high, suggesting that these cultivars are susceptible to broad mites. Similar results are also found for 'Thesla' and its sport 'Amélie' in mite numbers and SA levels. Both cultivars showed infestation later in the growing season. 'Mevrouw Gerard Kint' (red) and (picotee) and 'Renato' share a common genetic background and show a relatively low increase in SA levels; nevertheless, higher numbers of mites can occur. Only on 'Sachsenstern' were the level of SA and mite number low, in contrast to the results obtained for its sport 'Fluostern'. 'Sachsenstern' probably avoided infestation. Additionally, 'Nordlicht' had lower mite numbers and a lower SA level compared to experiment 2. Additionally, in other crops, a differential response towards broad mite was found, e.g., in citrus, sour orange supported larger densities than Cleopatra mandarin but the levels of injury were similar [39], and in 36 genotypes of common bean (Phaseolus vulgaris), differences in P. latus infestation were found [61].
The present study showed that not all types of tarsonemid mites cause damage in pot azalea. In culture conditions, the herbivorous and damage-causing mite P. latus is most prevalent. Nevertheless, not all pot azalea cultivars are equally attractive to this mite. 'Elien', the most resistant cultivar towards P. latus mites, possibly exhibits an antixenotic effect based on the presence of glandular trichomes with a sticky end [17]. For the other cultivars, i.e., 'Huelsten', 'Inka' and 'Sachstenstern', the nature of their resistance to P. latus is less clear. Another frequently occurring mite, T. confusus, was present on all the azalea cultivars without causing any symptoms. Until now, it was not fully understood whether this mite species is herbivorous. This study could only relate the presence of P. latus to plant damage. P. pallidus could be harmful, but so far, this has not been observed and the number of mites found in our experiments was probably too low.

Conclusions
In conclusion, it is clear that the presence of broad mite on pot azalea increases both JA and SA levels. A primary plant response is through the JA pathway, but later, this changes to an increase in the levels of SA. Further efforts, e.g., experiments under controlled conditions, are needed to better understand whether the increase in SA is a result of crosstalk between JA and SA and whether it can help to elucidate the timing and factors influencing the JA and SA responses.
Another conclusion has importance for azalea cultivation. In 18 cultivars, a large variation in susceptibility to broad mite was found, with 'Otto' being the most susceptible and 'Elien' the most resistant cultivar. This variation in susceptibility to broad mite opens up perspectives for using more resistant genotypes in cultivation and suggests that more genotypes with improved resistance can be developed though breeding.