The Use of Metabolomics to Elucidate Resistance Markers against Damson-Hop Aphid

Phorodon humuli (Damson-hop aphid) is one of the major pests of hops in the northern hemisphere. It causes significant yield losses and reduces hop quality and economic value. Damson-hop aphid is currently controlled with insecticides, but the number of approved pesticides is steadily decreasing. In addition, the use of insecticides almost inevitably results in the development of resistant aphid genotypes. An integrated approach to pest management in hop cultivation is therefore badly needed in order to break this cycle and to prevent the selection of strains resistant to the few remaining registered insecticides. The backbone of such an integrated strategy is the breeding of hop cultivars that are resistant to Damson-hop aphid. However, up to date mechanisms of hops resistance towards Damson-hop aphids have not yet been unraveled. In the experiments presented here, we used metabolite profiling followed by multivariate analysis and show that metabolites responsible for hop aroma and flavor (sesquiterpenes) in the cones can also be found in the leaves, long before the hop cones develop, and may play a role in resistance against aphids. In addition, aphid feeding induced a change in the metabolome of all hop genotypes particularly an increase in a number of oxidized compounds, which suggests this may be part of a resistance mechanism. Electronic supplementary material The online version of this article (10.1007/s10886-018-0980-y) contains supplementary material, which is available to authorized users.


Introduction
Hop is an indispensable raw material for brewing beer and is almost exclusively used by the brewing industry, in order to impart increased microbiological stability, to strengthen foam stabilization, and to improve taste and aroma of the product (Nance and Setzer 2011). The annual worldwide production of hop is approximately 100,000 metric tons on average Electronic supplementary material The online version of this article (https://doi.org/10.1007/s10886-018-0980-y) contains supplementary material, which is available to authorized users. (Barth-Haas Group 2016), which underlines the worldwide economic significance of this globally traded crop. Until recently, the general goal of hop breeding programs was to obtain cultivars high in yield and in α-acids content. Since the first decade of the twenty-first century, the interest of hop breeding programs has shifted to the development of new flavours, which are increasingly demanded by craft brewers. Surprisingly, the objective to develop new disease-and insectresistant accessions, is currently of minor interest to hop breeders (Čerenak et al. 2009;Darby and Campbell 1996;Kralj et al. 1998;Seigner et al. 2005Seigner et al. , 2009, notwithstanding hop is cultivated in monoculture, thus the risk of pest and disease outbreaks is increased. Several pathogens and pests are able to damage the hop plants significantly, thus causing serious yield losses, and sometimes the complete destruction of a crop (Mahaffee et al. 2009). In the northern hemisphere the Damson-hop aphid Phorodon humuli (Schrank) is a major pest of hop. It is an obligate holocyclic/heteroecious aphid with four species of Prunus − sloe, damson, plum, and cherry plum − serving as primary (winter) hosts and hop as its sole secondary (summer) host (Eppler 1986). It can also cause significant loss of yield, but even light infestations of the cones can already compromise their quality and reduce their economic value (Barber et al. 2003;Weihrauch et al. 2012).
Farmers control P. humuli with insecticides, although due to environmental concerns the number of registered active ingredients is steadily decreasing. The pesticide selection pressure often results in the development of resistant aphid genotypes (Hrdý et al. 1986). An integrated approach to pest management in hop is therefore badly needed in order to break this cycle and to prevent the selection of strains resistant to the few remaining registered insecticides. One cornerstone of such an integrated strategy consists in the breeding of hop cultivars that are, partially, resistant to P. humuli. Differences in varietal susceptibility to the hop aphid have been demonstrated previously (Campbell 1983; Darby and Campbell 1988;Dorschner and Baird 1988;Kralj et al. 1998;Weihrauch and Moreth 2005), but the mechanisms underlying these resistances are unknown. To cope with herbivorous insects, plants have evolved various and complex mechanisms of defence. In principle, two categories can be distinguished: constitutive and induced defence, which can be further sub-divided into direct and indirect defence. The first barrier is mechanical protection against insect attacks, such as the presence of hairs or a thick cuticle. However, the chemistry of plants plays a much more important role in defence/resistance mechanisms than these mechanical barriers (Smith and Boyko 2007;Züst and Agrawal 2016). Although extensively studied, the attractiveness, repellence and resistance mechanisms of plants against aphids are far from being fully unraveled (Dogimont et al. 2010;Miles 1999;Smith and Boyko 2007;Walling 2008;Züst and Agrawal 2016). In general aphids select their host plants based on plant colour and odor via complicated, multiple stage processes (Döring 2014;Mehrparvar et al. 2014;Pitino and Hogenhout 2012;Powell and Hardie 2001;Powell et al. 2006). Hop is very rich in secondary metabolites such as bitter acids (α-acids), prenylflavonoids and terpenoids. These families of molecules may play a role in host selection by aphids (Kryvynets et al. 2008). Farag et al. (2012a, b) using LC-MS and NMR techniques, managed to identify and quantify 46 hop metabolites, including 18 bitter acids, 12 flavonoids, three terpenes, three fatty acids and two sugars. Nance and Setzer (2011) and Yan et al. (2017) analysed the volatile components extracted from hop cones. Some of these compounds like β-myrcene, germacrene D and α-selinene were demonstrated to be not only important for hop aroma but also to play a role in resistance or susceptibility to hop powdery mildew (Čerenak et al. 2009). The bracts of the cones contain most of the polyphenols, while the terpenoids and bitter compounds are contained in the lupulin, a yellow sticky resin secreted by glands in the cones (Biendl et al. 2014). However, as the winged P. humuli morphs migrate to hop as their summer host long before the cones have been formed, the leaf metabolites are more likely explaining host selection behavior during the aphid's spring migration. Plant metabolomics, a powerful tool, first described in 2000 (Fiehn et al. 2000) was developed to identify biomarkers responsible for metabolic characteristics and revealing metabolic mechanisms. In literature we can find that the tool was successfully used for example to enhance understanding of the mechanisms of induced defense response in rice plant against C. suppressalis (Liu et al. 2016) or to identify secondary plant compounds involved in hot plant resistance ). Additionally, metabolomics or metabolite profiling was used to study foxglove aphids (Aulacorthum solani Kaltenbach) on leaves of soybean (Sato et al. 2013), or to explain variation in acceptance or discrimination between plant species of pea aphid (Hopkins et al. 2017) The present study objective was to identify key features of the leaf metabolome that correlated with P. humuli resistance (or susceptibility) in a broad panel of hop genotypes. For this purpose we used GC-MS metabolic profiling of apolar secondary metabolites in hop leaves before and after P. humuli infestation. The hop genotypes selected for this study were verities already used commercially or important parental lines with a full spectrum of resistance. Firstly, the resistance level of 20 field-grown genotypes was compared with the metabolic profile of leaves collected in early summer, when plants had not yet been colonized by aphids, therefore representing the possible constitutive defence compounds. Secondly, we assessed the metabolites induced by aphid feeding on six genotypes covering the full range of resistance. This second experiment was performed at two different time points in a controlled environment. Taken together, our results show that sesquiterpenes and lupulon contribute to the metabolic differentiation of hop genotypes and show a good correlation with resistance or sensitivity to aphids. Hop Genotype Classification into Resistance Categories In 2011, so before the metabolite profiling experiments, the aphid infestation on the hop genotypes used in this study was monitored and quantified in an experimental hop field and used to categorise the hop genotypes into resistance classes, as follows (also see Table 1).
Experimental Set-up A schematic overview of the experimental setup is shown in the Supplementary Materials, Fig. S1. The plants analyzed in this study were grown under two different conditions. Firstly, 20 hop genotypes were grown in an experimental hop field (Wolnzach, Bavaria, Germany). These plants were free of aphids and were used to analyse the constitutive and Damson-hop aphid infested (red) genotypes are shown. Arrows indicate a number of the compounds that changed upon aphid infestation leaf metabolome, without aphid infestation. Secondly, to study the effect of aphids on the leaf metabolome, six genotypes were grown under greenhouse conditions and infested with P. humuli at two relevant time points for the plant-insect interaction.
For the former experiment, the plant material consisted of two (vS1, vS3, iR3, R1, R2), six (vS2, PA, TU, PE, OL, SR, SD, HT) and 12 (HM, HS, iR2, WH49, SE, R3, BO) root cuttings per hop genotype. Plants (leaves) were harvested 36 days after planting. Each genotype was analysed as four (vS1, vS3, iR3, R1, R2) or six (remaining genotypes) replicates, where 10-13 leaves per replicate were harvested, immediately frozen in liquid nitrogen, ground to a fine powder with a mortar and pestle, and stored at −80°C until analysis. Prior to harvesting, all genotypes were carefully investigated for the presence of aphids on leaves and aphid infestation could be excluded in all cases.
For the aphid infestation experiment, six representative hop genotypes were chosen that significantly vary in their resistance against aphids (HM, HS, iR1, SE, R3, BO). At the beginning of the experiments (early and later summer), plants (60 pots per treatment) were divided into two groups, one marked as control ( c ) and second as treatment ( t ) and kept apart to avoid cross-contamination. In early summer, plants were grown for 47 days and leaves of plants marked as treatment were exposed to aphid infestation (n = 5 per leaf) 13 days prior to harvest. In late summer, plants were grown for 45 days and leaves of plants marked as treatment were exposed to aphid infestation (n = 20 per leaf) 10 days prior to harvest. From each plant four comparable leaves without (control) or with aphids (treatment group) were collected. Prior to harvest all leaves of the treatment group were cleaned from aphids with a paintbrush. Leaves from five plants were combined to one replicate, immediately frozen in liquid nitrogen, ground to a fine powder with a mortar and pestle, and stored at −80°C until analysis (five replicates per treatment were collected).
Chemical Analysis For GC-MS analysis, 500 mg (FW) of the frozen ground leaf material was extracted twice with 1 mL of dichloromethane (Sigma-Aldrich, The Netherlands) containing  4.92 mM (Z)-nerolidol (Sigma-Aldrich, The Netherlands) as internal standard. The material was then sonicated for 15 min in an ultrasonic water bath (3510 Branson, USA) and centrifuged for 10 min at 1609 ×g (Harrier 15/80, MSE, United Kingdom). The dichloromethane extracts were dried through anhydrous sodium sulfate (Sigma-Aldrich, The Netherlands).
Samples were analysed on an Agilent 7890A gas chromatograph connected to the 5795C mass selective Triple-Axis Detector (Agilent Technologies, United States). For that purpose, 1 μl of extract was injected at 250°C in splitless mode on a ZB-5MS column (Phenomenex, 30 m × 0.25 mm; ID 0.25 μm) with 5 m guard column with a constant flow of helium at 1 mL/min. The oven was programed for 1 min at 45°C and then subsequently ramped at 10°C/min to 300°C, at which it was kept for 7 min. The ionization potential was set at 70 eV, and scanning was performed from 45 to 450 amu, with a scanning speed of 3.99 scans/s.

Data Processing and Statistical Analysis
The GC-MS data from all experiments were baseline corrected and aligned using Metalign (Lommen 2012) followed by mass/peak filtering using METOT (https://wiki.nbic.nl/index.php/METOT), an in-house developed web-based tool. Filtering was based on the observation of a mass/peak in at least three samples (minimal number of replicates) and a peak height above a certain threshold value considered being the noise level (selected empirically).
The obtained filtered data were clustered using MSClust (Tikunov et al. 2012), yielding a matrix with sample names in columns and putative metabolites (named centrotypes) in rows. The matrix consisting of the corresponding peak heights was then normalized to 1 mg FW leaf material and to the peak height of the IS (relative abundances). The MS spectra of the centrotypes that were reconstructed by MSClust were compared with the NIST14 spectral library and in house MS spectra/retention index libraries.
The data (relative abundances of centrotypes) from all chemical analyses were first analysed with 1-way ANOVA with Bonferroni correction (P < 0.05) to select the metabolites that show significant differences between treatments. Next, exploratory multivariate analyses (principal component analysis, PCA and hierarchical cluster analysis, HCA) were performed first with GeneMaths XT (ver. 2.12, Applied Maths NV, Belgium), followed by Multibase: Excel Add-ins for PCA and PLS (ver. 2015, Numerical Dynamics) and MetaboAnalyst 3.0 (Xia et al. 2015). Canonical redundancy analysis (RDA) was performed with Canoco for windows 4.5 and/or Canoco5 software (http:// www.canoco5.com). Prior to PCA, HCA and RDA, data were normalized by median value, log2 transformed and auto scaled. HCA analysis was performed using Pearson clustering and the complete aggregation procedure. The RDA analysis was performed using four resistance groups as environmental factors. For the analysis of aphid infested data, a multivariate classification method, partial least squares -discriminant analysis (PLS-DA) on two groups (control and treatment) was performed with MetaboAnalyst 4.0. Data were pre-processed as for the exploratory multivariate analyses.

Results
The GC-MS Metabolome of Hop Leaves Based on the aphid counting in the field, as summarized in Table 1, the hop genotypes used in the present study were divided into three arbitrary categories: aphid susceptible or very susceptible (S, vS), intermediate resistant (iR) and aphid resistant (R) genotypes. We investigated the compounds responsible for the different level of resistance in the leaves by analyzing the dichloromethane extracts with GC-MS. Figure 1 shows the metabolite profiles (chromatograms) of hop plants belonging to the three mentioned resistance categories (a-c respectively), with (red lines, Binfested^) or without (green lines, Bcontrol^) aphid feeding. Aphid feeding resulted in the increase or suppression of one or a number of metabolites in the attacked plants.
Qualitative assessment of the data showed that the profile of R genotypes changed more than that of iR or S plants when they were exposed to aphid feeding, and that those changes are associated with more than one metabolite. Therefore, taking into account the complexity of the data set, an untargeted metabolomics approach was further used to identify metabolites related with aphid-resistance/susceptibility.
Metabolic Profile of 20 Field-Grown Genotypes During the field experiment leaf samples for 20 genotypes were analysed using untargeted metabolomics. This untargeted approach allowed for the detection of 213 putative metabolites. The 1way ANOVA with Bonferroni correction showed that there was a significant difference (P < 0.05) between the analysed hop samples in 176 mass-signals. After filtering out the noise, 97 mass-signals (metabolites) remained, of which many could be putatively annotated (Table S1). These were used in the subsequent multivariate analyses to identify the relevant differences between the samples.    2002/186/740 (vS3). The separation between the vS and other groups explained 21.7% (PC1) of the variation. At the same time PC2 (15.7%) differentiated further within S, iR and R genotypes. The variation between the biological replicates was larger for the vS (e.g. PA, HS) than for the S, iR and R genotypes (e.g. iR2, SE, R3, BO). This grouping was explained (the loadings plot, Fig. S2) by two main clouds of metabolites that cause the separation along PC1, and three weakly separated groups that cause separation along PC2 (Clusters I-III, Table 2). Along PC1, the largest group consisted of the sesquiterpenes (Cluster III), and compounds annotated as higher alkanes and phytol-like compounds (Cluster II), which are separated from sesquiterpenes along PC2. The remaining group (Cluster I) comprised lupulon-like compounds, sterols and their possible precursors. Sesquiterpenes (Cluster III) generally had higher amplitudes in the R (e.g. 95 germacrene D), and iR genotypes, with the exception of WH49 (e.g. 105 γ-cadinene), or were equally distributed among the analysed hop genotypes (e.g. 78 βcopaene) (Fig. 3). The lupulon-like compounds and sterols (Cluster I) were more defined and usually the vS genotypes had clearly higher levels of these compounds than the other ones (e.g. 307 lupulon, 280 lupulon, 304 laurenan-2-one) (Fig. 3). Alcohols, sterols and their precursors from Cluster II (e.g. 33 1-penten-3-ol, 200 phytol acetate) had the lowest concentration in vS and the highest in R genotypes.
The link between above mentioned metabolites and aphid resistance was further analysed using hierarchical cluster analysis (HCA), which highlights the similarity between different hop samples based on the metabolites they contain. HCA analysis clustered samples into two main groups, coined Branch A and Branch B, with two sub-clusters B1 and B2 (Fig. 4). Branch A contained vS hop genotypes, with exception of vS1 and vS3. Branch B consisted of three out of four R hop genotypes, vS3, iR3 (sub-cluster B1), and vS1, S, remaining iR and R1 hop genotypes (sub-cluster B2).
This clustering was caused by three groups of metabolites (Cluster I and III Fig. 4; Cluster II shown in supplementary materials Fig. S3). The most R genotypes (R2, R3, BO), but also vS (vS2, PA, HM, TU) were characterized by higher levels of metabolites (Cluster III), which were annotated as sesquiterpenes (C15H24) or sesquiterpene alcohols (C15H26O or C15H24O). The vS genotypes (vS2, PA, HM, TU, HS) had high levels of compounds clustered into Cluster I. Metabolites of this cluster were in many cases annotated as humulene, humulene oxides, and plant sterols. Cluster II (Fig. S3) represented metabolites, which were relatively abundant in the S and iR genotypes, and were belonging to the longer chain hydrocarbons. In summary, PCA and HCA analyses allowed to distinguish between hop genotypes based on their leaf metabolome, and to identify that the sesquiterpenes were present in higher levels in the R genotypes. To narrow down individual metabolites that correlate with the resistance, redundancy analysis (RDA) was performed. RDA was carried out using resistance level as an environmental variable (explanatory variable). The first dimension (AX1) separated the vS hop genotypes from the rest (Fig. 5). The second dimension (AX2) separated the genotypes into two groups: iR, (v)S, and R genotypes. Closer analysis of responsible metabolites that were ranked in order of theirs positions on the ordination axes showed that the sesquiterpenes εmuurolene and germacrene D (82, 95/103 respectively) drive the separation of the R genotypes. On the other side of the plot, compounds annotated as lupulon or lupulon-like compounds (e.g. 280, 282, 272, 287) seemed to drive the separation of the vS genotypes. The separate clustering of the iR and S genotypes was driven by higher levels of hydrocarbons and hydrocarbon alcohols or esters such as green leaf volatiles, e.g. isovaleric acid tetracosyl ester, butyl 2-methylbutanoate, (Z)-3-hexen-1-ol and (Z)-3-hexen-1-ol acetate (369,364,4,24). In summary, we identified metabolic features showing good correlation with resistance in a field experiment performed with 20 diverse hop genotypes. However, this experiment explained only the constitutive defences observed in the plants, in the next step we have investigated the molecules that upon induction would correlate with hop resistance to aphids.
Aphid Induced Metabolite Changes Based on the results obtained from the field study, a sub-selection of hop genotypes was further investigated in the greenhouse setting. The aim of this experiment was to detect metabolite changes that occur after aphid infestation, and to identify metabolites that may be involved in induced defence mechanisms. Hereto, the metabolite profile of hop leaves of six varieties; HM, HS, iR1, SE, R3 and BO was investigated for aphid infection, 4 weeks after planting (Fig. S1). When the infestation experiment was repeated in late summer ( Fig. S1 and Fig. S4), no clear metabolite changes occurred upon aphid infestation (see below). As the objective of this study was to determine metabolite changes as a response to aphid spring migration we focus on the spring experiment.  Table  S1 Untargeted GC-MS analysis allowed for the detection of 215 putative metabolites, of which 109 displayed a significant difference between the varieties (P < 0.05, 1-way ANOVA with Bonferroni correction). The subset of significantly different metabolites was used as an input for untargeted statistical analysis.
The results of the supervised PLS-DA analysis between two (control and treatment) groups are shown in Fig. 6 (Fig. 6 shows scores plot, Fig. S5 shows loadings plot). The PLS-DA scores plot (component 1 explains 32.8%, component 2 explains 16.2% of total variation) shows that the first component clearly separates the control plants from the treated (infested) hop genotypes (Fig. 6). The component 2 divides the hop genotypes into two well distinguished groups, containing S (HS, HM) and R (BO, R3), while iR genotypes are divided between these two groups. The corresponding loadings plot displayed three dense clouds of metabolites (Fig.  S5); clusters I, II and III along loadings 1, and clusters II and I/III along loadings 2. Clusters I and III contain compounds with high VIP score values (a measure of a variable's importance in the PLS-DA model) that influence the separation into control and infested genotypes. Cluster II contains compounds annotated as sesquiterpenes, with very low VIP score values, indicating that these compounds do not play a role in the induced resistance mechanisms. Analysis of the first 30 metabolites with the highest VIP score values showed that 22 of these were higher in infested genotypes (Cluster III) than in the respective controls (Cluster I) (Table 3). Figure 7 shows the data for a number of these compounds, where e.g. abundances of squalene and cabreuva oxide E (340 and 184, respectively) were higher, while abundances of α-tocopherol and lupulon (388 and 272, respectively) were lower in infested genotypes. A second PLS-DA of control and aphid infested plants, which was performed on data collected in late summer (Fig. S4) did not show strong shifts between control and aphid-induced plants. In general control genotypes were grouped together with treated ones, with the exception of the S genotypes. Also in this experiment, however, the main resistance groups separated well. Moreover, the separation between genotypes iR1, R3, BO on the one hand and HS, HM, SE on the other was stronger than in early spring.

Discussion
Damson-hop aphid is one of the major pests of hop in the northern hemisphere. Even light infestations of the harvested cones can already damage their quality and reduce their economic value. An integrated approach to pest management focusing on the breeding of hop cultivars at least partially resistant to P. humuli, as well as methods to screen for such resistance, would be very desirable. The use of untargeted metabolomics has contributed to the elucidation of resistance mechanisms in two Senecio species and tomato to Western flower thrips (Leiss et al. 2009;Mirnezhad et al. 2010), and in soybean to foxglove aphid (Sato et al. 2013). However, to our knowledge this approach has not yet been used to study resistance in hops. In this study we have used metabolomics to pinpoint leaf metabolites that correlate with, and are possibly responsible for, hop resistance against P. humuli. Because aphids colonise their host long before the cones have been formed (Born 1968), we decided to focus on the metabolite composition of leaves, as it would add important knowledge to the existing data on cone metabolites. By using multivariate analysis in two independent The results of the untargeted PCA of 20 hop genotypes showed that based on metabolite composition hop genotypes cluster in groups with different resistance levels (Fig. 2). Only the breeding line vS1 did not follow this pattern. However, in contrast to the other genotypes we investigated, the breeding line vS1 has the North American H. lupulus var. pubescens in its pedigree, contrary to all other tested genotypes that were bred from European or East Asian genotypes. Even if pubescence has been associated with insect resistance this was not the case for vS1 (Hanley et al. 2007). Weaker differences between S and iR hop genotypes are supported by the work of Weihrauch et al. (2013), in which the iR cultivar SE grouped either with S or iR genotypes, depending on the season when the assessment was performed. The closer analysis of responsible metabolites by RDA showed that the sesquiterpenes, ε-muurolene and germacrene D (82 and 95/103, respectively), drive the separate clustering of the R genotypes. Both compounds have been shown to play a role in plantaphid interaction (Bruce et al. 2005;Saad et al. 2015). Our results are supported by Čerenak et al. (2009), who showed a link between essential oil compounds of hop cultivars and their resistance/susceptibility to Powdery mildew. Among the markers that they reported, were santalene, germacrene D, α-selinene, and caryophyllene epoxide. In our study, germacrene D was strongly correlated with aphid resistance. Additionally, Kralj et al. (1998) recognized three markers correlating with hop resistance to aphids, where two compounds were positively annotated as αand β-pinene. In our study these metabolites were not found, which can be explained by the fact that Kralj et al. (1998) studied the plant headspace, while we analysed hop leaf extracts. Next, we have identified compounds annotated as lupulon or lupulon-like compounds that seem to drive the separation of the vS genotypes. A high beta acid content of hop leaves has been reported to be a good indicator for aphid susceptibility (Kammhuber 1997). In contrast, these components of the beta acid family of leaf-specific bitter acids have repellent and oviposition-deterring effects on two-spotted spider mites Tetranychus urticae, another wellknown pest of hop (Jones et al. 1996(Jones et al. , 2010. Plant can activate specific defences to deal with pests, the so called induced defences. The current study also examined the influence of aphid feeding on the hop leaf metabolome, to pinpoint the induced metabolites upon aphid infestation. Aphid feeding induced leaf metabolome changes in all hop genotypes. Various classes of molecules were up or down regulated (Table 3). However, this effect was only observed in spring and not in late summer, when the experiment was repeated. These findings are consistent with those of Weihrauch et al. (2013) and suggest that late stage resistance might be driven by another resistance mechanism. The elevated levels of oxidized compounds such as cabreuva oxide E, cedroxyde or sesquicineol-2-one upon aphid infestation would suggest the influence of those metabolites in the spring resistance mechanism, while lupulon-like and α-tocopherol-like compounds were reduced upon infestation. Lupulon has been shown to influence the behavior of aphids on hop and to change between early and late summer (Kryvynets et al. 2009).
In summary, in this study two different experiments were performed. One concentrated on constitutive resistance mechanisms and showed that resistant genotypes contain higher amounts of sesquiterpenes; this class of metabolites could be used as a constitutive resistance marker. The second experiment investigated the induced resistance mechanisms and indicated that oxidized compounds such as e.g. sesquicineol-2one are more important than sesquiterpenes.
With this study we have narrowed down the group of metabolites that are of importance in hop resistance mechanisms towards Damson-hop aphid. Further confirmation with Damson-hop aphid feeding experiments will be needed. However, in conjunction with the hop genome and the understanding of the biosynthesis of sesquiterpenes and bitter acids (Cattoor et al. 2013;Clark et al. 2013;Hill et al. 2017), it should be possible to design and test specific molecular markers related to the molecules we identified here.