# Source paper archive

- Source title: Effects of Florivore Type and Plant Chemotype on Metabolic Responses in Flowers and Leaves of an Aromatic Plant Species
- Author(s): Gaar, S., Brokate, L., Dussarrat, T., Mueller, C.
- Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.13.766439v1
- PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.13.766439v1.full.pdf
- PDF SHA-256: 29a9e64c2b3cfc51014dcca487e299373983de2a672f044d9a2e9ddabc3abfe3
- Converter: pymupdf4llm 0.0.27
- Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version.

---

&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)

# 1 Effects of Florivore Type and Plant Chemotype on Metabolic Responses in 2 Flowers and Leaves of an Aromatic Plant Species


3


4 **Selina Gaar** **&#91;1&#93;** **, Lukas Brokate** **&#91;1&#93;** **, Thomas Dussarrat** **&#91;1, 2*&#93;** **, Caroline Müller** **&#91;1, 3, *&#93;**


5


6 _1_ _Department of Chemical Ecology, Bielefeld University, Bielefeld, Germany_


7 _2_ _Bordeaux University, INRAE, UMR1332 BFP, 33882 Villenave d’Ornon, France_


8 _3_ _Joint Institute for Individualisation in a Changing Environment (JICE), University of Münster and_


9 _Bielefeld University, Bielefeld, Germany_


10


11 **Correspondence:** Caroline Müller (caroline.mueller@uni-bielefeld.de), Thomas Dussarrat


12 (thomas.dussarrat@inrae.fr)


13


14 **ORCID**


15 Selina Gaar: https://orcid.org/0009-0000-0475-2603


16 Lukas Brokate: https://orcid.org/0009-0006-7604-7907


17 Thomas Dussarrat: https://orcid.org/0000-0001-6245-3652


18 Caroline Müller: https://orcid.org/0000-0002-8447-534X


19


20 **Funding:** This research was funded by the German Research Foundation (Deutsche


21 Forschungsgemeinschaft, DFG), project MU1829/28-2 and DU 2440/1-2, as part of the research unit


22 FOR 3000 (project number 415496540).


23


24 **Keywords**


25 Chemotypes; feeding mode; florivory; induced responses; intraspecific chemodiversity; metabolomics;


26 plant-insect interactions; plasticity; _Tanacetum vulgare_, volatile organic compounds.


1


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


27 **ABSTRACT**


28 Florivory is an important biotic challenge in plant-insect interactions, evoking induced responses in


29 plant specialised metabolism, the extent of which depends on phenotypic plasticity. While responses


30 to leaf herbivores are well studied, floral responses to florivores with different feeding modes, their


31 systemic effects on leaves and the role of intraspecific chemodiversity remain poorly understood. We


32 used clonal replicates of five _Tanacetum vulgare_ (Asteraceae) chemotypes, which differ in their leaf


33 and flower terpene profiles, to test how florivory by chewing beetles ( _Olibrus_ spp.) and phloem

34 feeding aphids ( _Macrosiphoniella tanacetaria_ ) affects the metabolic responses of flower heads and


35 systemically in leaves. Floral volatile organic compounds (VOCs) and stored terpenes were analysed


36 via GC-MS and metabolic fingerprints via LC-MS. While overall metabolic profiles of flower heads


37 and leaves remained largely stable following florivory, specific metabolites shifted in abundance


38 locally and systemically. These changes primarily involved the shikimate and phenylpropanoid


39 pathways and minor terpenes. The specific metabolic responses varied by florivore species and


40 chemotype, a pattern observed across floral VOCs, as well as stored terpenes and untargeted metabolic


41 fingerprints, differing also between organs. Overall, florivory triggers organ-specific chemical


42 responses in reproductive and vegetative tissues, shaped by a complex interplay between plant


43 chemotype and florivore type.


2


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


44 **1. Introduction**


45 Plants interact with a diverse community of insects that feed on leaves, roots, stems and/or flowers.


46 Flowers, in particular, are often nutrient-rich resources for mutualists such as pollinators, making them


47 also attractive targets for floral herbivores, known as florivores (McCall and Irwin, 2006; Sasidharan


48 et al., 2023). As reproductive organs, flowers contribute strongly to fitness and according to the


49 optimal defence hypothesis, plants should therefore allocate defences preferentially to them. Flowers


50 are expected to be protected by high constitutive concentrations of specialised metabolites, whereas in


51 leaves induced defences are often more pronounced (Brown et al., 2003; McCall and Irwin, 2006;


52 Smallegange et al., 2007). Florivores differ in their feeding modes, which determine how and where


53 they damage flowers and how plants respond. Biting-chewing florivores, such as beetles and


54 caterpillars, consume floral tissue directly and often cause visible damage, whereas piercing-sucking


55 florivores, such as aphids, feed on phloem sap, typically at the floral stalks (Rusman et al., 2019).


56 Chewing mainly activates the jasmonate/ethylene pathway, while piercing-sucking typically activates


57 the salicylic acid pathway, although crosstalk occurs (Erb et al., 2012; Koornneef and Pieterse, 2008;


58 Schweiger et al., 2014). Plant responsiveness to feeding by insects with different feeding modes has


59 mostly been studied in leaves (Mithöfer and Boland, 2012; Schuman and Baldwin, 2016), while


60 knowledge on flower responses is comparatively scarce.


61 Signalling pathways shape downstream defence responses, including changes in specialised metabolite


62 profiles (Walling, 2000). In flowers, florivory can alter profiles of both emitted volatiles and stored


63 specialised metabolites (Lucas‐Barbosa et al., 2016; Zangerl and Berenbaum, 2009). Altered emission


64 of volatile organic compounds (VOCs) potentially acts as indirect defence by attracting natural


65 enemies (Desurmont et al., 2015; Schiestl et al., 2014), whereas changes in the stored metabolite


66 composition may contribute to direct defence against subsequent florivores (Boyer et al., 2016;


67 Rusman et al., 2022). However, volatile and stored metabolites can also be attractive or stimulatory to


68 plant antagonists (Gols, 2014; Schiestl, 2015). Flowers therefore face a trade-off: they must repel/deter


69 florivores while remaining attractive to mutualists, which suggests that floral defences should be


70 selective (Kessler and Halitschke, 2009; McCall and Irwin, 2006; Schiestl et al., 2014). Possible


71 mechanisms include visitor-specific blends of attractive and repellent VOCs (Kessler et al., 2013;


3


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


72 Raguso, 2016) or temporal regulation of VOC emission (Theis et al., 2007). Moreover, defence


73 responses can also be systemic, as exemplified by the observed variation in stored and volatile floral


74 metabolites under leaf herbivory (Bruinsma et al., 2014; Rusman et al., 2022; Schiestl et al., 2014).


75 Because flowers and leaves are connected via the phloem, which determines how far systemic


76 induction extends (Orians, 2005), signals from damaged flowers may also reach leaves through the


77 vascular system. In accordance, florivory has been shown to alter leaf chemistry, such as


78 glucosinolates in _Brassica nigra_ (Brassicaceae) after combined florivore and pathogen attack


79 (Chrétien et al., 2022) and anthocyanins in _Impatiens capensis_ (Balsaminaceae) after artificial


80 florivory (Soper Gorden and Adler, 2016). However, whether and how florivory affects leaf chemistry


81 across plant species remains largely unknown.


82 Plant specialised metabolite profiles also differ constitutively among individuals of the same species in


83 composition and concentration, a phenomenon referred to as intraspecific chemodiversity (Moore et


84 al., 2014; Wetzel and Whitehead, 2020). In some species, this variation is structured into so-called


85 chemotypes (Müller et al., 2026), i.e., groups of individuals that differ in the presence and/or quantity


86 of their dominant metabolites, for example alkaloids or terpenoids in species of Asteraceae (Castells et


87 al., 2014; Grof-Tisza et al., 2022). As a consequence, chemotypes also differ in their interactions with


88 herbivores, affecting, for example, herbivore preference behaviour and performance (Jakobs and


89 Müller, 2018; Kleine and Müller, 2011; Sasidharan et al., 2023; van Leur et al., 2008). Chemotypes


90 may also differ in their inducibility, expressing phenotypic plasticity. For example, chemotype

91 specific differences in induced VOC emissions and stored metabolites were found in _Gossypium_


92 _hirsutum_ (Malvaceae) and _Tanacetum vulgare_ (Asteraceae) after feeding by chewing and phloem

93 feeding herbivores (Clancy et al., 2020; Mamin et al., 2025; Newrzella et al., 2026). Despite the


94 potential ecological relevance, the knowledge about induced chemotype-specific responses is still


95 limited.


96 Among the Asteraceae, _T. vulgare_ (common tansy) stands out for its pronounced intraspecific


97 chemodiversity in leaf and floral terpenes across a broad geographical range (Clancy et al., 2016;


98 Keskitalo et al., 2001; Wolf et al., 2011). Leaf terpene profiles allow individuals to be grouped into


99 chemotypes, each defined by its dominant and minor co-occurring terpenes (Clancy et al., 2018;


4


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


100 Kleine and Müller, 2011). Within its natural range, _T. vulgare_ is colonised by a diverse herbivore


101 community (Schmitz, 1998) and chemotype-specific plant responses have been demonstrated. For


102 example, infestation by the specialised aphid _Macrosiphoniella tanacetaria_, which preferentially feeds


103 on the stems below the inflorescence, alters the nutritional quality of the phloem sap, with effects


104 depending on the colonised plant part and the plant’s chemotype (Jakobs et al., 2019). Infestation by


105 this aphid species on leaves also enhanced shoot monoterpene emissions in a chemotype-specific


106 manner (Newrzella et al., 2026). Flowers of _T. vulgare_ are also visited by oligophagous _Olibrus_


107 beetles (Coleoptera: Phalacridae), which feed mainly on pollen and other parts of flower heads (Eilers


108 et al., 2021; Sasidharan et al., 2023). When feeding on different stems of the same plants under field


109 conditions, _M. tanacetaria_ and _Olibrus_ spp. induced chemotype-specific metabolic responses in the


110 flower heads: aphids mainly decreased and beetles mainly increased the levels of metabolic features,


111 with little overlap in significant features across insect species and plant chemotypes (Gaar et al.,


112 2026). Nevertheless, both insect species increased features related to the shikimate and


113 phenylpropanoid pathways (Gaar et al., 2026). Yet, it remains unclear how chemotypes shape local


114 and systemic plant responses to these distinct florivore types.


115 The aim of this study was to elucidate the effects of florivory by insects differing in feeding mode on


116 the metabolic responses in both flower heads and leaves of different _T. vulgare_ chemotypes. Plants of


117 five chemotypes were cloned and clones were either infested with _M. tanacetaria_ aphids or _Olibrus_


118 spp. beetles or kept uninfested as controls, allowing us to test plant individual-specific responses.


119 Metabolic responses were measured using a multi-platform approach, analysing floral VOCs as well


120 as terpenes and metabolic fingerprints of flower heads and leaves to determine local as well as


121 systemic responses. We expected plant responses to depend on the chemotype and florivore feeding


122 mode, with only modest shifts in overall metabolic profiles restricted to individual metabolites in both


123 flower heads and leaves. Specifically, we hypothesised florivory to cause an increase in individual


124 floral VOCs, but only minor changes in stored flower and leaf terpenes. Moreover, we expected


125 alterations across diverse other classes of specialised metabolites, especially those associated with the


126 shikimate and phenylpropanoid pathways, with metabolites modulated by aphid infestation being


5


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


127 predominantly decreased in abundance, whereas those affected by beetle feeding to be predominantly


128 increased and more pronounced.


129


130 **2 Materials and Methods**


131 **2.1 Plant and Insect Material**


132 Seeds of different _T. vulgare_ plants (considered as different maternal genotypes) were collected in


133 2019 in Bielefeld, Germany (for details see Ziaja and Müller, 2023). Plants were grown in a


134 greenhouse and chemotypes assigned based on leaf monoterpene profiles. Five chemotypes were


135 selected for further studies, based on their abundance and stability and to maximise chemical diversity.


136 Two chemotypes were dominated by a single terpene accounting for &gt;50% of the total terpene content,


137 namely β-thujone (BThu) and artemisia ketone (Keto), while three chemotypes were characterised by


138 two to three major terpenes contributing &gt;50% in total, namely artemisyl acetate, artemisia ketone and


139 artemisia alcohol (Aacet), α- and β-thujone (ABThu) and ( _Z_ )-myroxide, santolina triene and artemisyl


140 acetate (Myrox). As _T. vulgare_ is an outcrossing species, offspring from one maternal individual can


141 belong to different chemotypes. To separate chemotype from maternal genotype effects, we chose


142 plants from multiple maternal origins per chemotype. I n total, the selected chemotypes originated from


143 seeds of ten maternal genotypes, with overlapping genotypes across chemotypes where possible. From


144 each plant individual, three clones were produced from rhizomes and cultivated in separate pots (0.5


145 L) in a 1:1 mixture of sterilised potting soil and river sand. Plants were grown in a climate chamber at


146 23 °C during the light and 18 °C during the dark period, with a relative humidity of 60–70%. The


147 photoperiod was initially set to 12 h light/12 h dark for three weeks, increased to 14 h light/10 h dark


148 for two weeks and then maintained at 16 h light/8 h dark for the rest of the experiment, intending to


149 induce flowering. Nevertheless, plants originating from different individuals started flowering at


150 different time points, but clones within plants were roughly synchronised (mean flowering-onset


151 difference: 21.3 ± 25.4 days). The lights provided an average UV-A radiation of 0.524 W m⁻² and a


152 photosynthetically active radiation of 180 μmol m⁻² s⁻¹. Plants were watered three times per week and


153 positions rotated weekly. Once roots emerged through the pot bottom, plants were fertilised weekly


154 with WUXAL Super 8-8-6 (Hauert MANNA Düngewerke GmbH, Germany) at a concentration of


6


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


155 2 mL L⁻¹ of water. _Neoseiulus cucumeris_ were released preventively every eight weeks to control for


156 potential thrips and _Steinernema feltiae_ were applied every three months against fungus gnats.


157 A colony of _Macrosiphoniella tanacetaria_ aphids was established in May 2024 from individuals


158 collected in Bielefeld and maintained on _T. vulgare_ plants grown from purchased seeds of a random


159 mixture of chemotypes (as described in Gaar et al., 2026). Aphids were reared in tents (60 × 60 × 60


160 cm) in a climate chamber (16 h light/8 h dark; 22 °C/18 °C; 60-70% relative humidity).


161 _Olibrus_ spp. beetles (probably _Olibrus aeneus_ ) were collected in summer 2024 in Bielefeld and


162 Salzkotten from flower heads of wild chamomile species ( _Matricaria_ spp., _Anthemis_ spp., or


163 _Tripleurospermum_ spp., Asteraceae) and maintained in plastic boxes (22 × 13 × 12 cm) lined with


164 paper tissue within the same climate chamber as the aphids (as described in Gaar et al., 2026). They


165 were provided with flower heads of _T. vulgare_ from a mixture of all five chemotypes used in this


166 study (greenhouse-grown) as well as flower heads from wild _T. vulgare_ plants.


167


168 **2.2 Performance of Experiment and Sampling**


169 At peak flowering, each of the three clones was either infested with _M. tanacetaria_, with _Olibrus_ spp.


170 or kept uninfested as control (Figure 1; Table S1). To maintain insects on flower heads,


171 polyethylenetetraphtalate sauce cups (135 mL, Pro-Pac Ostendorf Plastic, Germany) with removable


172 lids and mesh bags were used. A 6 cm diameter hole was cut into the bottom of each cup and a mesh


173 bag with its lower end cut open was glued to the cup outside. The cup was slipped over a flower head


174 bouquet and fixed in position to a stick. Depending on the bouquet size, 5-15 flower heads were


175 enclosed, two apterous adult aphids per flower head or one adult beetle per three flower heads were


176 placed into the cup to standardise florivory pressure and the bag was closed around the stem with a


177 drawstring ribbon. The infestation levels reflect natural florivore densities observed on _T. vulgare_ .


178 Clones of the control treatment received empty cups in the same manner.


179 After three days of treatment, emitted volatiles were passively collected from late morning to


180 afternoon, using polydimethylsiloxane (PDMS) tubes as sorbent material, modified after Eckert et al.,


181 (2023). Two PDMS tubes (Carl Roth, Karlsruhe, Germany) with an inner diameter of 1.0 mm, an


182 outer diameter of 1.8 mm and a length of 0.5 cm were placed into the cups, avoiding any contact with


7


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


183 the flower heads. To sample background volatiles of the climate chamber, one additional cup


184 (hereafter cup blank) with two PDMS tubes per sampling event was fixed on a stick at similar height


185 as the flower heads and positioned in a pot filled with the same substrate. After 6 h collection, the


186 PDMS tubes were removed, placed in glass vials (1.5 mL), sealed with polytetrafluoroethylene tape


187 and stored at −70 °C until chemical analysis. These VOC profiles may also contain VOCs released


188 from the insects, which nevertheless contribute to the overall bouquet that a pollinator or other insect


189 would encounter.


190 After four days of infestation, the insects were removed. Flower heads and the second youngest


191 unfolded leaf were harvested, immediately flash-frozen in liquid nitrogen and stored at −70°C for later


192 analysis of terpenes and metabolic fingerprints. Infestation and harvest were always conducted


193 between 10 am and 12 pm. Since flowering started asynchronously among plants, 24 sampling events


194 were conducted over a period of eight months. As some plants ultimately did not flower, the final


195 number of successfully sampled triples was for Aacet: _n_ = 6, ABThu: _n_ = 7, BThu: _n_ = 9, Keto: _n_ = 4,


196 Myrox: _n_ = 4. In addition, for three Myrox individuals, only one of the three clones flowered, two of


197 these clones treated with beetles, the other with aphids. For one additional Myrox individual, aphid

198 and beetle-infested clones were accidentally swapped during leaf sampling, so only the control leaf


199 was retained for further processing. Prior to extraction, frozen flower heads and leaves were freeze

200 dried for 48 h and ground to a fine powder. The workflow is depicted in Figure 1.


201


202 **2.3 Analysis of Emitted Volatiles and Stored Terpenes**


203 Volatiles collected on PDMS tubes were analysed using a thermal desorption-gas chromatography

204 mass spectrometry (TD-GC-MS) system (TD-30 Series, GC 2010plus – MS QP2020, Shimadzu,


205 Japan) operated in electron impact ionisation mode at 70 eV (similar as in Eckert et al., 2023; Ziaja


206 and Müller, 2025). Thermal desorption of the tubes was carried out at 250 °C for 8 min with a flow


207 rate of 60 mL min⁻¹. The cryogenic trap was held at −20 °C during desorption and subsequently heated


208 to 280 °C for 3 min for analyte release. Injection into the GC was performed at 280 °C in splitless


209 mode, with a helium column flow of 1.54 mL min⁻¹. Compounds were separated on a semi-polar VF-5


210 MS capillary column (30 m × 0.25 mm × 0.25 µm, 10 m guard column; Agilent Technologies, CA,


8


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


211 USA). The oven temperature was initially set to 50 °C and held for 5 min after injection, then


212 increased to 250 °C at a rate of 10 °C min⁻¹, followed by a ramp of 30 °C min⁻¹ to 280 °C, with a final


213 hold for 2 min. The MS was operated with an ion source temperature of 230 °C, an interface


214 temperature of 280 °C and a detector voltage of 1.5 kV. Mass spectra were acquired in quadrupole MS


215 mode using the following scan parameters: from 8.2 to 19 min, a scan speed of 2000 and an _m_ / _z_ range


216 of 30–400; from 19 to 29 min, a scan speed of 625 and an _m_ / _z_ range of 30–500. To calculate retention


217 indices (RI) according to van den Dool and Kratz (1963), an alkane standard mixture (C7–C40;


218 Sigma-Aldrich) was analysed by desorbing 1 µL from a PDMS tube using the same GC method. To


219 prevent potential carry-over, an idle run was performed after the alkane standard and after every


220 twelfth sample.


221 Stored terpenes were analysed following Eilers et al. (2021) with minor modifications. For each


222 sample, 10 ± 2 mg were extracted in 1 mL _n_ -heptane (HPLC grade, Roth, Germany) containing 10 ng


223 µL⁻¹ 1-bromodecane (GC-MS grade, Sigma Aldrich, Germany) as internal standard, homogenised


224 (FastPrep-24 &#91;TM&#93; 5G, MP Biomedicals, Germany) for 30 s at 9 m s⁻¹, sonicated in an ultrasonic bath for


225 5 min at room temperature and centrifuged for 10 min at 20 °C. The supernatants were collected for


226 GC–MS analysis and blanks containing only the internal standard prepared in the same way (injected


227 every 20 plant samples). Samples were analysed using the same GC–MS system as for the VOC


228 measurements, but with liquid injection at 230 °C in split mode (1:10), a final hold time at the


229 maximum temperature of the oven of 3 min and an MS interface temperature of 250 °C. An alkane


230 standard mixture was analysed once a day, using the same GC–MS method to calculate RIs following


231 van den Dool and Kratz (1963). Samples were analysed in two batches: the first after approximately


232 three months of sampling events and the second after all plants had been sampled.


233 Chromatograms of VOCs and stored terpenes were processed using LabSolutions GC-MS Postrun


234 Analysis v4.45 (Shimadzu). Compounds were identified by comparing their RI and mass spectra with


235 those reported in Adams (2007), as well as with entries from the mass spectral libraries of NIST


236 (2014) and FFNSC 3 (Mondello, 2015). A compound was considered identified when its mass


237 spectrum matched a reference spectrum and its RI deviated by no more than 1.5% from the reported


238 value (Adams, 2007), as previously described in Gaar et al. (2026). Remaining features were


9


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


239 annotated as “unknown monoterpene or sesquiterpene”, based on characteristic masses. Compounds


240 were quantified based on peak areas from the total ion chromatogram. For VOCs, the peak area of the


241 cup blank collected on the corresponding sampling event was subtracted from the sample peak area


242 and the values were normalised to the number of flower heads per sample. In approximately 80% of


243 the VOC samples, acetophenone was overloaded and therefore excluded from further analyses. For


244 stored terpenes, peak areas were normalised to the internal standard, mean peak areas of blanks were


245 subtracted from each sample and values were subsequently normalised to the sample dry mass.


246 Compounds detected in fewer than three samples were removed from the final table (Tables S2, S3,


247 S4, S5).


248


249 **2.4 Analysis of Metabolic Fingerprints**


250 Metabolic fingerprinting was performed following Dussarrat et al. (2023) and Schweiger et al. (2021),


251 with minor adaptations. For each sample, 10 ± 2 mg was extracted in 500 µL of 90% methanol (v/v)


252 with hydrocortisone (10 mg L⁻¹; Sigma-Aldrich) as internal standard, vortexed, sonicated for 20 min


253 in an ice-cooled ultrasonic bath and centrifuged for 10 min at 6 °C. The supernatants were filtered (0.2


254 µm, Phenomenex, Torrance, CA, USA) before LC–MS analysis. Extraction blanks containing the


255 internal standard were prepared in the same manner (injected every 20 samples) and pooled quality


256 control samples (QC) were prepared from aliquots of all extracts (injected every 15 samples).


257 Liquid chromatography-mass spectrometry (LC–MS) analyses were performed using a Dionex


258 UltiMate 3000 UHPLC (Thermo Fisher Scientific, CA, USA) coupled to a compact quadrupole time

259 of-flight (QTOF)-MS/MS (Bruker Daltonics, Germany), equipped with a Kinetex XB-C18 column


260 (150×2.1 mm, 1.7 µm, with guard column; Phenomenex), following Schweiger et al. (2021) with


261 minor modifications. Chromatographic separation was carried out at 45 °C with a flow rate of 0.5 mL


262 min⁻¹, using Millipore water with 0.1% formic acid as eluent A and acetonitrile with 0.1% formic acid


263 as eluent B. The gradient began at 2% B, increasing to 30% over 20 min and then to 75% over 9 min,


264 followed by column cleaning and re-equilibration. Mass spectrometric data were acquired in negative


265 electrospray ionisation mode over an _m_ / _z_ range of 50–1300 at an acquisition rate of 5 Hz. All other


266 instrument parameters followed Schweiger et al. (2021). MS/MS were acquired using AutoMS/MS


10


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


267 mode. The isolation widths and collision energies were increased in line with the rising _m_ / _z_ of the


268 precursor ions. Nitrogen served as the collision gas. Before each sample run, a sodium formate


269 calibration solution was introduced to recalibrate the _m_ / _z_ axis.


270 Feature detection, alignment and grouping were carried out with the T-ReX 3D algorithm in


271 MetaboScape (v. 2021b, Bruker Daltonics). Pre-processing of raw LC-MS data were performed as


272 follows: intensity threshold of 1000 counts, minimum peak length of 11 spectra and EIC correlation


273 threshold of 0.8. Features were retained only if eluted between 0.3 and 29 min and present in at least


274 three samples. Recursive peak picking was applied with a minimum peak length of 6 spectra. Feature


275 intensities were first normalised to the internal standard (hydrocortisone, &#91;M+HCOOH−H&#93;⁻). The


276 coefficient of variation (CV) of each feature was then calculated across the QC samples and only


277 stable features with a CV below 30% were retained. The remaining intensities were normalised to


278 sample dry weight and the mean normalised blank intensities were subtracted from each


279 corresponding feature, with negative values set to zero. The resulting datasets are provided in Tables


280 S6 and S7.


281


282 **2.5 Statistical Data Analysis**


283 All data analyses and visualisations were performed in R v4.6.1 (R Core Team, 2026). Data were


284 processed and plotted using the _tidyverse_ collection of packages (Wickham et al., 2019), including


285 _dplyr_, _tidyr_, _stringr_, _tibble_, _readr_, _purrr_ and _ggplot2_ . Heatmaps were created with _pheatmap_ (Kolde,


286 2025), UpSet plots with _ComplexUpset_ (Krassowski et al., 2022) and Venn diagrams with


287 _VennDiagram_ (Chen and Boutros, 2011).


288 To explore the overall variation in floral VOC profiles and to assess how plants of different leaf


289 chemotypes cluster based on those, a principal component analysis (PCA) was performed on the


290 normalised VOC data &#91;relative values, cube root transformation and Pareto scaling performed in


291 MetaboAnalyst v6.0 (Pang et al., 2024)&#93;. To visualise how floral VOCs differ between treatments


292 within each chemotype, a heatmap of the mean normalised abundance of each VOC was generated for


293 every chemotype × treatment combination. To identify individual VOCs that responded significantly


294 to florivory, moderated t-tests and fold changes (FC) were calculated for treatments _versus_ controls,


11


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


295 both across all samples and within each chemotype. Moderated t-tests were performed on the


296 normalised data using the _limma_ package (Ritchie et al., 2015), with plant identity included as a


297 blocking factor to account for the paired design with plant clones. Moderated t-tests borrow


298 information across all volatiles to stabilise variance estimates, which yields more reliable _P_ -values


299 when sample sizes are small, which was the case for the Keto and Myrox chemotypes. Only VOCs


300 present in at least three samples were tested. Fold changes (FCs) were calculated from absolute


301 abundances as the ratio of the insect treatment mean to the control treatment mean. _P_ -values from the


302 moderated t-tests were adjusted for multiple testing using the false discovery rate (FDR) method


303 (Benjamini and Hochberg, 1995). As no VOC remained significant after FDR correction, VOCs were


304 considered significantly increased or decreased based on unadjusted _p_ -values ( _p_ &lt; 0.05) combined


305 with an FC &gt; 2 or &lt; 0.5. These VOCs were visualised in a heatmap based on normalised data. To


306 compare responses across groups (chemotype × treatment combination), the overlap of significant


307 VOCs, irrespective of the direction of change, was visualised in an UpSet plot.


308 The total stored terpene content (summed peak areas of all terpenes, normalised to internal standard


309 and sample weight) was compared between flower heads and leaves within untreated control plants


310 using a two-sided paired Wilcoxon signed-rank test. To explore variation in stored terpene profiles


311 while accounting for the paired design with plant clones, a multilevel PCA was performed on the


312 normalised data (relative values, cube root transformation and Pareto scaling performed in


313 MetaboAnalyst) using the package _mixOmics_ (Rohart et al., 2017). Multilevel PCA centres each


314 terpene within plant clones, so that the remaining variation reflects clone differences. To visualise how


315 the stored terpenes in flower heads and leaves differ among the treatment groups within each


316 chemotype, a heatmap was generated per organ, showing the mean normalised abundance of each


317 terpene. To identify insect species- and chemotype-specific modulations, moderated t-tests and FCs


318 were calculated using the same approach as for the VOC data, separately for flower heads and leaves.


319 Significant terpenes were visualised in a heatmap per organ and the overlap across groups (treatment ×


320 organ × chemotype combination) was visualised in an UpSet plot.


321 Variation in metabolic fingerprints was assessed by multilevel PCA and, separately for flower heads


322 and leaves, by multilevel Partial Least Squares Discriminant Analysis (PLS-DA) on normalised LC

12


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


323 MS data using _mixOmics_ (median normalisation, cube root transformation and Pareto scaling


324 performed in MetaboAnalyst). To illustrate how metabolic profiles differ between flower heads and


325 leaves across chemotypes, a heatmap was created for each organ, showing the mean normalised


326 abundance of each feature per chemotype. Moderated t-tests and FCs were calculated as described for


327 the VOCs. After FDR correction, only one of approximately 1400 features remained significant.


328 Chemical features were considered significantly altered if unadjusted _p_ -values &lt; 0.05 combined with


329 an FC &gt; 2 or &lt; 0.5. To compare responses across groups, overlaps of significant features, irrespective


330 of the direction of change, were visualised in UpSet plots per organ (chemotype × treatment) and


331 across organs and treatments (chemotypes pooled). To compare the direction of responses, the


332 percentage of increased features (the remainder, up to 100%, being decreased features) was shown in a


333 bar plot for each treatment × organ × chemotype combination. To test whether the number of


334 significant features differed among chemotypes, organs and insect species, counts of significant


335 features per treatment × organ × chemotype combination ( _n_ = 20: 5 chemotypes × 2 organs × 2 insect


336 species) were analysed. As the counts were overdispersed (dispersion of a Poisson model = 4.2), a


337 negative binomial generalised linear model (GLM) was fitted using the MASS package (Venables and


338 Ripley, 2002). The model included chemotype, insect species and organ as main effects and the


339 interactions chemotype × insect and insect × organ. The proportion of increased _versus_ decreased


340 features (number of increased features out of all significant features) was analysed with a binomial


341 GLM using the same predictors (dispersion = 1.1). Effects were tested with Type II likelihood-ratio


342 tests ( _car_ package; Fox and Weisberg, 2019). When effects were significant, post hoc pairwise


343 comparisons were performed on estimated marginal means with Tukey adjustment (package _emmeans_ ;


344 Lenth et al., 2026). Chemical class annotation of significant features was performed using SIRIUS


345 v6.5.4 (Dührkop et al., 2019). Full parameter settings are provided in Table S8. The metabolite


346 structure and chemical class predictions were performed using CSI:FingerID (Hoffmann et al., 2022)


347 and CANOPUS (Dührkop et al., 2021). For each feature, pathway, superclass and class, annotations in


348 the Natural Products Classifier (NPC) ontology were retained when the confidence score was ≥ 0.8.


349 Significant features were assigned identification-confidence levels according to the Metabolomics


350 Standards Initiative (MSI) guidelines (Schymanski et al., 2014), with MSI 2 obtained if cos score ≥


13


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


351 0.7. For each pathway, the number of significantly increased and decreased features was counted


352 across four comparison groups (aphid/beetle × flower heads/leaves) and visualised in a bar plot,


353 regardless of the chemotype. Pie charts were plotted to show the distribution of significant features


354 across chemical pathways for each chemotype.


355 To compare florivore-induced responses across the three platforms (floral VOCs, stored terpenes and


356 metabolic fingerprints), bar plots were generated for each dataset showing the number of significant


357 features that were specific to one chemotype or shared by two or more chemotypes, irrespective of


358 insect species. For the same datasets, Venn diagrams (package _VennDiagram_ ; Chen and Boutros,


359 2011) were plotted to depict the overlap of significant features between aphid and beetle treatments,


360 irrespective of chemotype. For stored terpenes and metabolic fingerprints, additional Venn diagrams


361 showing the overlap between flower heads and leaves, irrespective of chemotype and insect species,


362 were generated.


363


364 **3. Results**


365 **3.1 Effects of Chemotype and Florivory on Floral VOC Emissions**


366 In total, 136 putatively annotated floral VOCs were detected, which grouped according to chemotype


367 in a standard PCA (Figure S1). The thujone-related chemotypes (ABThu, BThu) clustered distinctly


368 from the Aacet, Keto and Myrox chemotypes, which share artemisia alcohol in their profiles. Aacet


369 and Keto additionally share artemisia ketone, while Aacet and Myrox share artemisyl acetate. The


370 overall VOC profiles did not separate by florivory treatment. However, the heatmap showed visible


371 differences in the abundance of several VOCs between treatments within each chemotype (Figure S2).


372 Analyses of all samples combined (i.e. considering all chemotypes) and chemotype-specific analyses


373 identified a total of 30 VOCs that were significantly increased or decreased by florivory ( _p_ &lt; 0.05, FC


374 - 2 or &lt; 0.5; Figure 2A; for detailed results of all moderated t-tests and FCs see Table S9).


375 Responding volatiles were mostly specific to both the chemotype and the florivore species and only a


376 few were shared among chemotypes or between the two insect species (Figure 2A, B). Aphid and


377 beetle infestation led to a similar number of responsive VOCs overall, but this number varied among


378 chemotypes depending on the insect species. Towards aphids, ABThu showed the strongest response


14


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


379 (8 VOCs), whereas Myrox and BThu showed the weakest (1 VOC each). Towards beetles, Aacet


380 showed the strongest response (5 VOCs), followed by Myrox and ABThu (4 each), whereas Keto and


381 BThu had only one significant VOC change each.


382


383 **3.2 Effects of Chemotype and Florivory on Stored Terpenes in Flower Heads and Leaves**


384 A total of 159 stored terpenes were detected in the flower heads and 149 in the leaves. All leaf


385 terpenes were also present in the flower heads. The total stored terpene content was significantly


386 higher in flower heads (median 427.6 a.u.) than in leaves (median 288.4 a.u.; _p_ = 0.0013). Multilevel


387 PCA revealed only a slight separation of flower head versus leaf profiles along PC1 (Figure 3A).


388 Within organs, the stored terpenes showed treatment-related variations in their abundance within each


389 chemotype (Figure S3). Analyses of all samples (i.e. considering all chemotypes) and chemotype

390 specific analyses identified a total of 20 stored terpenes that were significantly increased or decreased


391 by florivory in flower heads and 12 in leaves (Figure 3B, C; for detailed results of all moderated t-tests


392 and FCs see Table S10). In 8 of the 20 chemotype × insect × organ combinations, no terpene


393 significantly responded to florivory. While mostly minor terpenes responded to florivory, the


394 dominant terpene artemisia ketone decreased significantly in the flower heads of the Keto chemotype


395 following aphid infestation. By contrast, in the ABThu chemotype, where artemisia ketone is not


396 prominent, its abundance in flower heads increased significantly following both aphid and beetle


397 infestation. Responding terpenes were largely specific to the organ, the chemotype and florivore


398 species and only a few were significantly changed in both organs, multiple chemotypes, or by both


399 insect species (Figure 3B-D). In flower heads, both insects affected a similar number of terpenes. In


400 leaves, however, beetles led to a higher number of responding terpenes than aphids. The number of


401 responsive terpenes also varied among chemotypes when organ and insect species were not


402 considered, with the strongest response in Aacet (11), followed by ABThu (6), Myrox (4) and Keto (3)


403 and the weakest in BThu (2).


404


405 **3.3 Effects of Chemotype and Florivory on Metabolic Fingerprints in Flower Heads and Leaves**


15


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


406 A total of 7163 features were detected in the flower heads and 7134 in the leaves, of which 7122 were


407 shared. However, the abundance of many features differed substantially between the two organs


408 (Figure S4). Several features differed also in abundance between chemotypes. Overall, multilevel PCA


409 showed a clear separation between the metabolic profile of flower heads and leaves along PC1, which


410 explained 55% of the variance (Figure 4A). PC2 explained only 4% of the variance and reflected a


411 separation between individual plants, with all clones of a plant always clustering together. This


412 separation was not related to the chemotype, treatment or plant age and its biological basis remains


413 unclear. However, the supervised multilevel PLS-DA, performed separately for flower heads and


414 leaves, revealed some separation of metabolic profiles among florivory treatments. In flower heads,


415 the profiles of control samples were separated from those of insect-treated samples, whereas beetle

416 and aphid-treated samples slightly overlapped (Figure S5A). The control group was classified best by


417 the first component alone, whereas the second component was needed to separate aphid- and beetle

418 treated samples. With two components, class error rates ranged from 0.31 to 0.39, indicating


419 detectable but not strong treatment effects. In contrast to flower heads, the control group was not


420 clearly separated in the multilevel PLS-DA for the leaves and the second component mainly improved


421 the classification of the beetle-treated samples (Figure S5B). With two components, class error rates


422 ranged from 0.41 to 0.46, indicating only weak separation.


423 Considering all chemotypes, 93 features were significantly increased or decreased by florivory across


424 flower heads and leaves. There was no overlap between the features being changed in the two organs


425 and only minimal overlap between features affected by the two insect species ( _p_ &lt; 0.05, FC &gt; 2 or &lt;


426 0.5; Figure S6). In contrast, analyses performed separately for each chemotype identified a total of


427 1355 significantly modulated features, illustrating the need to consider intraspecific chemodiversity


428 (for detailed results of all moderated t-tests and FCs see Table S11). Of these, 710 were significantly


429 increased or decreased in flower heads and 756 in leaves, with 111 features being shared. Responding


430 features were largely specific not only to the organ and chemotype, but also to the florivore species.


431 Only a few significant features overlapped between insect species and chemotypes in both flower


432 heads and leaves (Figure 4B, C).


16


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


433 Across all chemotype × insect × organ combinations, the number of significant features and the


434 proportion of increased _versus_ decreased features varied considerably (Figure 4D). The number of


435 responding features varied significantly among chemotypes depending on the insect species (Table


436 1A). Under aphid feeding, Aacet showed more than twice as many significantly modified features


437 (estimated mean: 125; Table S12) than BThu (52; _p_ &lt; 0.001) and Myrox (49; _p_ &lt; 0.001; Table S13),


438 with ABThu (92) and Keto (91) having intermediate responses (significantly higher than in BThu and


439 Myrox, all _p_ &lt; 0.01). In contrast, under beetle feeding, ABThu (137) and Keto (130) responded most,


440 with both having more features significantly modified than Aacet (79; _p_ &lt; 0.001 and _p_ &lt; 0.01,


441 respectively), BThu (57) and Myrox (68; all _p_ &lt; 0.001), the latter three not differing in responsiveness


442 from each other (Tables S12, S13). Also, organ-specific effects depended on the insect species (Table


443 1). Beetles modulated a significantly higher number of features in leaves (98) than in flower heads


444 (81; _p_ &lt; 0.05), whereas aphids elicited similar responses in both organs (82 in flower heads and 72 in


445 leaves; _p_ - 0.05; Table S14). However, the higher number of responding features in leaves under


446 beetle feeding appeared to be restricted to ABThu (Figure 4D). The organ effect in beetles should


447 therefore be interpreted with caution, as it may be driven by this chemotype. The proportion of


448 increased _versus_ decreased features depended on the interaction between chemotype and insect species


449 and differed between organs (Table 1B). Only in the Aacet chemotype, beetle feeding caused mostly


450 increased features (0.61), whereas aphid feeding was mainly associated with decreased features (0.59;


451 odds ratio = 0.44, _p_ &lt; 0.001; Table S15). Across chemotypes and florivore type, increased and


452 decreased features were balanced in flower heads (0.50 increased), whereas leaves showed a higher


453 proportion of decreased features (0.56; Table S16), especially under aphid feeding (Figure 4D).


454 Overall, more chemotype × insect × organ combinations showed mainly decreased than increased


455 features (Figure 4D), although this trend was not tested statistically.


456 Of the 1399 features that responded significantly to florivory (1355 from the analyses performed


457 separately for each chemotype and 44 additional features from the chemotype-independent analyses;


458 Table S17, Figure 5), 15 could be annotated at the compound level (MSI 2), of which most belonged


459 to the shikimate and phenylpropanoid pathways, including caffeic acid derivatives, flavonoids and


460 other phenolic compounds, as well as an amino acid and a lipid (Table S17). The remaining features


17


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


461 were classified at the level of biosynthetic pathway, superclass or class (Table S17). Pathway


462 classification revealed that the shikimate and phenylpropanoid pathways responded most strongly to


463 florivory (Figure 5A). In these pathways, aphid feeding affected more features in flower heads than in


464 leaves, whereas beetle feeding showed the opposite pattern. Increased and decreased features were


465 balanced in all four groups. Amino acids and peptides responded with similar numbers of features


466 across insects and organs, but differed in direction: flower heads showed mainly increased features in


467 response to both insect species, whereas leaves showed mainly decreased features. All other


468 responding pathway classes (alkaloids, carbohydrates, fatty acids, polyketides and terpenoids) were


469 mostly comparable between insects and organs in both the number of responding features and the


470 proportion of increased and decreased features. Despite modest differences between chemotypes,


471 florivory-responsive features showed broadly similar proportional distributions across chemical


472 pathways when organ and insect species were not considered (Figure 5B).


473


474 **3.4 Overall Comparison of Florivore-Induced Responses across Analytical Platforms**


475 Of the 159 stored terpenes detected in the flower heads, 96 were also emitted (VOCs) _._ Of these 96


476 compounds, 21 responded to florivory in the VOCs and 10 in the stored fraction, but only two


477 unknown terpenes (RI1080 and RI1296) responded in both fractions. Across all three analytical


478 platforms, the overlap of metabolites responding to the two insect species was low (Figure 6), meaning


479 that most responding features were specific to one insect species. Moreover, only few features


480 responded in flower heads and leaves simultaneously, in both stored terpenes and metabolic


481 fingerprints, indicating an organ-specific response. Features that responded significantly in only one


482 chemotype clearly outnumbered those that were significant in two or more chemotypes, highlighting


483 chemotype-specific responsiveness.


484


485 **4. Discussion**


486 Our knowledge of induced plant responses to insects with different feeding modes is largely based on


487 leaves. In contrast, floral responses to different types of florivory and their systemic effects on leaf


488 chemistry, have received less attention. Moreover, little is known about how intraspecific diversity in


18


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


489 chemotypes affects these responses. Here, we investigated how feeding by florivores differing in


490 feeding mode shapes the metabolic responses of flower heads and leaves in different _T. vulgare_


491 chemotypes. Constitutive defences (total terpene content) were much higher in flower heads than in


492 leaves, while overall terpene profiles of flower heads and leaves remained largely stable after


493 infestation. Nevertheless, both local and systemic responses occurred in individual metabolites in an


494 organ-specific manner. The responding features were mainly related to the shikimate and


495 phenylpropanoid pathways, or were minor terpenes. The number and identity of responding features


496 also differed among chemotypes and depended on the insect species, indicating that the chemotype


497 modulates how strongly plants respond to different flower infestation.


498


499 **4.1 Stable Overall Profiles with Selective, Organ-Specific Responses to Florivory**


500 The overall volatile and stored metabolic profiles of flower heads were not altered by florivory. This is


501 in line with our expectations: flower heads already exhibit high constitutive defences and may


502 therefore require little induction upon attack (Kessler and Halitschke, 2009; McCall and Irwin, 2006).


503 Consequently, the overall profile of leaves, which could respond via systemic signalling, was also


504 largely unaffected. Maintaining a relatively stable floral chemical profile is likely beneficial, as


505 extensive changes in floral chemistry may interfere with pollinator attraction. Recent evidence


506 suggests that floral scent emission can remain remarkably stable following florivory across different


507 plant species: in _Centrosema pubescens_ (Fabaceae) and _Tocoyena formosa_ (Rubiaceae), florivory did


508 not alter the overall scent profile and only a few volatiles responded (Tunes et al., 2026). In _T. vulgare,_


509 individual metabolites changed in abundance in flower heads, but additionally in leaves. The vascular


510 connection between flowers and leaves may explain such systemic responses. Furthermore, wound

511 induced signals such as electrical signals and calcium waves, as well as volatiles released from


512 damaged tissues, can coordinate responses in distant tissues and thereby mediate within-plant


513 communication (Heil and Karban, 2010; Lee and Seo, 2022; Orians, 2005). Across all chemotype ×


514 insect × organ combinations, more of the responding features decreased rather than increased upon


515 florivory, suggesting a selective increase of a few compounds and an adjusted metabolism. Moreover,


516 these responses were largely organ-specific, while only a comparatively small number responded in


19


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


517 both organs. Similar patterns have been observed after leaf herbivory, where some compounds were


518 induced in both flowers and leaves, while others were induced in only one organ in _Solanum_


519 _peruvianum_ (Solanaceae) (Kessler and Halitschke, 2009). This suggests that a common wound signal


520 can trigger organ-specific metabolic responses. It may also reflect the different ecological functions of


521 reproductive and vegetative tissues, as floral chemistry is involved in both pollinator attraction and


522 defence against florivores, whereas leaf chemistry primarily mediates interactions with herbivores and


523 their natural enemies (Kessler and Chautá, 2020; Mithöfer and Boland, 2012).


524


525 **4.2 Stored and Emitted Terpenes are Only Partly Overlapping**


526 Stored and emitted terpenes were only partly matching in _T. vulgare_ . Both fractions shared the major


527 terpenes and the leaf and flower chemotypes were reflected in the floral VOCs. Upon infestation, the


528 chemotype-defining terpenes were only weakly altered, while responses were mainly restricted to


529 minor compounds. However, the volatile and stored fractions differed in their responsiveness, with


530 only two terpenes responding in both pools. The different responsiveness may be due to the fact that


531 VOCs were sampled one day earlier than the flower heads and leaves for analysis of stored terpenes


532 due to logistical reasons. A close connection between the volatile and stored fractions has been


533 demonstrated in several _Chrysanthemum_ species (Asteraceae), where floral terpene emission rates


534 largely correlated with stored concentrations in undamaged plants (Zhang et al., 2021). However, to


535 our knowledge, stored and emitted terpenes have not been compared in flowers after florivory. In


536 leaves, emissions increased pronouncedly in _Melaleuca alternifolia_ (Myrtaceae) after feeding by


537 _Paropsisterna tigrina_ beetles and the VOC composition reflected both the constitutive terpene profile


538 and compound volatility, indicating that stored terpenes are the main emission source (Bustos-Segura


539 and Foley, 2018). In contrast, in _T. vulgare_ leaves, constitutively stored and emitted terpene profiles


540 only partly overlapped, as some stored terpenes are poorly volatile (Clancy et al., 2016). Responses to


541 herbivory were likewise decoupled between pools in _T. vulgare_ shoots, where phloem-feeding aphids


542 increased monoterpenoid emissions without affecting stored fractions (Newrzella et al., 2026).


543 Similarly, herbivory of _Spodoptera exigua_ increased total volatile emissions of _Gossypium hirsutum_


544 (Malvaceae) leaves but hardly changed stored mono- and sesquiterpenes and some compounds, such


20


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


545 as α-pinene, were induced in the headspace but not in leaf tissue (Mamin et al., 2025). Chemotype

546 characteristic terpenes were induced only weakly, as in our data. The terpenes are usually stored in


547 glandular trichomes, while the emission of VOCs highly depends on the morphology of the trichomes


548 and the type and physicochemical properties of the compounds, but the transport processes of VOCs


549 are still not fully understood (Tissier et al., 2017). Thus, VOC and stored terpene profiles may not


550 necessarily overlap, neither under constitutive nor under induced conditions.


551


552 **4.3 Florivore- and Chemotype-Specific Responses to Florivory**


553 Beetles and aphids, applied to separate clones in the laboratory, altered the flower head metabolome of


554 _T. vulgare_ to a similar extent in the present study. In contrast, when aphids and beetles were applied to


555 separate stems of the same plant in the field, aphids had a weaker effect than the beetles (Gaar et al.,


556 2026). Systemic signalling within the plant and variable environmental conditions may have led to


557 these different outcomes. Here, we found that _T. vulgare_ leaf response was insect species-specific,


558 with beetles inducing slightly more systemic changes than aphids. Similarly, chewing caterpillars of


559 _Spodoptera littoralis_ systemically induced terpenes in undamaged leaves of _G. hirsutum_, whereas


560 phloem-feeding aphids _Aphis gossypii_ did not systemically affect specialised metabolites (Eisenring et


561 al., 2018). Our results show that the nature of the responsive metabolites was also insect species

562 specific, which corroborates previous observations (Gaar et al., 2026). In _T. vulgare_, aphid infestation


563 predominantly decreased metabolite levels, while beetles tended to cause an increase (Gaar et al.,


564 2026). This contrasting pattern was only observed in the Aacet chemotype in this study, showing the


565 need for further investigation on floral responses to florivores with different feeding modes. Chewing


566 and phloem-feeding insects can elicit distinct chemical responses. In leaves and other vegetative


567 organs, aphids and chewing herbivores induced different volatile blends in _Solanum tuberosum_


568 (Gosset et al., 2009) and differentially regulated glucosinolates in young _Brassica rapa_ (Sotelo et al.,


569 2014). Feeding-type-specific effects also extend to flowers when herbivores attack leaves, as shown in


570 two Brassicaceae species. In _Sinapis alba_, damage by the aphid _Lipaphis erysimi_ on vegetative plant


571 parts reduced floral VOC emission, while the chewing caterpillar _Plutella xylostella_ had no effect


572 (Pareja et al., 2012). Such insect-specific metabolic responses may reflect the insects’ contrasting


21


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


573 feeding modes and the signalling pathways they trigger (Koornneef and Pieterse, 2008; Pieterse et al.,


574 2012), as in _G. hirsutum_, where caterpillars increased jasmonoyl-L-isoleucine and abscisic acid levels,


575 whereas aphids decreased salicylic acid levels (Eisenring et al., 2018).


576 Plant responses to florivory were not only florivore-specific, but also chemotype-specific, as indicated


577 above for the distinct responses in the Aacet chemotype. Moreover, only a few responding terpenes


578 and untargeted features were shared between chemotypes and the number of responding features


579 varied among chemotypes depending on the florivore. These results are also congruent with previous


580 observations (Gaar et al., 2026). Chemotype-dependent induction has also been observed in other


581 organs of _T. vulgare_ . Root-chewing wireworms ( _Agriotes_ spp.) induced variable responses of stored


582 root sesquiterpenoids between _T. vulgare_ chemotypes, while aphids ( _M. tanacetaria_ ) increased shoot


583 monoterpenoid emissions in only one chemotype (Newrzella et al., 2026). In _G. hirsutum_, leaf


584 emissions of aldoximes and nitriles, as well as stored concentrations of ( _E_ )-β-ocimene, varied between


585 chemotypes after caterpillar feeding (Mamin et al., 2025) and terpene emissions following artificial


586 damage differed among chemotypes of _M. alternifolia_ (Bustos-Segura and Foley, 2018). Chemotype

587 specific responses of terpenes to florivory may be linked to differences in constitutive terpene profiles


588 and the underlying biosynthetic pathways, possibly through chemotype-specific regulation of terpene


589 synthase genes, as proposed for _T. vulgare_ roots and shoots (Newrzella et al., 2026). However,


590 chemotype-specific responses were also observed for other, non-terpenoid features, indicating that


591 these responses extend beyond terpene metabolism. Indeed, terpene chemotypes of _T. vulgare_ also


592 differ in their broader constitutive metabolome (Dussarrat et al., 2023), while alkaloids and fatty acids


593 allowed chemotype prediction with over 70% accuracy (Müller et al., 2026). Chemotype-specific


594 volatile responses may also act as within-plant signals and thereby modulate induced responses in


595 other metabolic pathways, including non-volatile features (Frost et al., 2007; Heil and Karban, 2010).


596


597 **4.4 Shared Metabolic Pathways behind Specific Responses to Florivory**


598 Although the identity of the responding metabolic features was insect species- and chemotype

599 specific, most annotated features were associated with the shikimate and phenylpropanoid pathways,


600 as also found in _T. vulgare_ flower heads in an earlier study (Gaar et al., 2026). These pathways


22


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


601 produce numerous defence-related metabolites, including caffeic acid derivatives, flavonoids and other


602 phenolic compounds, which can act as feeding deterrents, antioxidants or toxins and are frequently


603 induced upon herbivory (Mostafa et al., 2022; War et al., 2012). Many responding features were also


604 annotated as terpenoids and alkaloids in the present study. The responsiveness of terpenoids is


605 consistent with their central role in plant-insect interactions in _T. vulgare_ . Alkaloids are also well

606 known defence compounds (Mostafa et al., 2022; War et al., 2012), that can accumulate following


607 herbivore attack sometimes together with phenolic defences (Fernandes et al., 2011). As alkaloids also


608 differ between _T. vulgare_ chemotypes (Müller et al., 2026), they may contribute to the chemotype

609 specific responses observed here. A few amino acids and peptides responded to florivory in


610 contrasting patterns, with mostly increased in flower heads and decreases in leaves. Although based on


611 few compounds, this tendency may hint at a shift in nitrogen towards the flower heads, which are


612 important nitrogen sinks that can receive nitrogen remobilised from vegetative tissues (Tegeder and


613 Masclaux‐Daubresse, 2018).


614


615 **5. Conclusion**


616 In conclusion, the overall metabolic profiles of flower heads and leaves remained largely stable


617 following florivory. Nevertheless, feeding by both florivore species triggered local responses in flower


618 heads and systemic responses in leaves. The magnitude and composition of these responses depended


619 on the florivore’s feeding mode, plant chemotype and organ. This pattern was consistently observed


620 across floral VOC emissions and stored terpenes and untargeted metabolic fingerprints of flower heads


621 and leaves. Despite the differences in individual metabolic responses, the same major compound


622 classes were affected by both florivores, particularly features associated with the shikimate and


623 phenylpropanoid pathways, as well as terpenes, across both organs and all chemotypes. Taken


624 together, these findings show that plant chemistry is shaped by the interplay of florivore identity and


625 intraspecific chemodiversity. Future work could build on this by testing whether such metabolic shifts


626 have ecological consequences, for example for pollinator attraction, under natural conditions, where


627 plants face diverse insect communities and variable environmental conditions.


23


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


628 **Acknowledgements**


629 We acknowledge support for the publication costs by the Open Access Publication Fund of Bielefeld


630 University. We thank Annika Schindéle for her help with plant maintenance.


631


632 **Conflicts of Interest**


633 The authors declare no conflicts of interest.


634


635 **Data Availability Statement**


636 Chemical data, metadata and code will be made available upon acceptance of this manuscript.


637


638 **References**


639 Adams, R.P., 2007. Identification of Essential Oil Components by Gas Chromatography/Mass
640 Spectrometry. Allured Publishing Corporation, Illinois, USA.
641 Benjamini, Y., Hochberg, Y., 1995. Controlling the False Discovery Rate: A Practical and Powerful
642 Approach to Multiple Testing. Journal of the Royal Statistical Society Series B: Statistical
643 Methodology 57, 289–300. https://doi.org/10.1111/j.2517-6161.1995.tb02031.x
644 Boyer, M.D.H., Soper Gorden, N.L., Barber, N.A., Adler, L.S., 2016. Floral damage induces
645 resistance to florivory in Impatiens capensis. Arthropod-Plant Interactions 10, 121–131.
646 https://doi.org/10.1007/s11829-015-9411-y
647 Brown, P.D., Tokuhisa, J.G., Reichelt, M., Gershenzon, J., 2003. Variation of glucosinolate
648 accumulation among different organs and developmental stages of _Arabidopsis thaliana_ .
649 Phytochemistry 62, 471–481. https://doi.org/10.1016/S0031-9422(02)00549-6
650 Bruinsma, M., Lucas-Barbosa, D., Ten Broeke, C.J.M., Van Dam, N.M., Van Beek, T.A., Dicke, M.,
651 Van Loon, J.J.A., 2014. Folivory Affects Composition of Nectar, Floral Odor and Modifies
652 Pollinator Behavior. Journal of Chemical Ecology 40, 39–49. https://doi.org/10.1007/s10886-013653 0369-x
654 Bustos-Segura, C., Foley, W.J., 2018. Foliar Terpene Chemotypes and Herbivory Determine Variation
655 in Plant Volatile Emissions. Journal of Chemical Ecology 44, 51–61.
656 https://doi.org/10.1007/s10886-017-0919-8
657 Castells, E., Mulder, P.P.J., Pérez-Trujillo, M., 2014. Diversity of pyrrolizidine alkaloids in native and
658 invasive _Senecio pterophorus_ (Asteraceae): Implications for toxicity. Phytochemistry 108, 137–
659 146. https://doi.org/10.1016/j.phytochem.2014.09.006
660 Chen, H., Boutros, P.C., 2011. VennDiagram: a package for the generation of highly-customizable
661 Venn and Euler diagrams in R. BMC Bioinformatics 12, 35. https://doi.org/10.1186/1471-2105-12662 35

663 Chrétien, L.T.S., Khalil, A., Gershenzon, J., Lucas‐Barbosa, D., Dicke, M., Giron, D., 2022. Plant
664 metabolism and defence strategies in the flowering stage: Time‐dependent responses of leaves and
665 flowers under attack. Plant, Cell &amp; Environment 45, 2841–2855. https://doi.org/10.1111/pce.14363
666 Clancy, M.V., Haberer, G., Jud, W., Niederbacher, B., Niederbacher, S., Senft, M., Zytynska, S.E.,
667 Weisser, W.W., Schnitzler, J.-P., 2020. Under fire-simultaneous volatilome and transcriptome
668 analysis unravels fine-scale responses of tansy chemotypes to dual herbivore attack. BMC Plant
669 Biology 20, 551. https://doi.org/10.1186/s12870-020-02745-1
670 Clancy, M.V., Zytynska, S.E., Moritz, F., Witting, M., Schmitt‐Kopplin, P., Weisser, W.W.,
671 Schnitzler, J., 2018. Metabotype variation in a field population of tansy plants influences aphid host
672 selection. Plant, Cell &amp; Environment 41, 2791–2805. https://doi.org/10.1111/pce.13407


24


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


673 Clancy, M.V., Zytynska, S.E., Senft, M., Weisser, W.W., Schnitzler, J.-P., 2016. Chemotypic variation
674 in terpenes emitted from storage pools influences early aphid colonisation on tansy. Scientific
675 Reports 6, 38087. https://doi.org/10.1038/srep38087
676 Desurmont, G.A., Laplanche, D., Schiestl, F.P., Turlings, T.C.J., 2015. Floral volatiles interfere with
677 plant attraction of parasitoids: ontogeny-dependent infochemical dynamics in _Brassica rapa_ . BMC
678 Ecology 15, 17. https://doi.org/10.1186/s12898-015-0047-7
679 Dührkop, K., Fleischauer, M., Ludwig, M., Aksenov, A.A., Melnik, A.V., Meusel, M., Dorrestein,
680 P.C., Rousu, J., Böcker, S., 2019. SIRIUS 4: a rapid tool for turning tandem mass spectra into
681 metabolite structure information. Nature Methods 16, 299–302. https://doi.org/10.1038/s41592682 019-0344-8
683 Dührkop, K., Nothias, L.-F., Fleischauer, M., Reher, R., Ludwig, M., Hoffmann, M.A., Petras, D.,
684 Gerwick, W.H., Rousu, J., Dorrestein, P.C., Böcker, S., 2021. Systematic classification of unknown
685 metabolites using high-resolution fragmentation mass spectra. Nature Biotechnology 39, 462–471.
686 https://doi.org/10.1038/s41587-020-0740-8
687 Dussarrat, T., Schweiger, R., Ziaja, D., Nguyen, T.T.N., Krause, L., Jakobs, R., Eilers, E.J., Müller,
688 C., 2023. Influences of chemotype and parental genotype on metabolic fingerprints of tansy plants
689 uncovered by predictive metabolomics. Scientific Reports 13, 11645.
690 https://doi.org/10.1038/s41598-023-38790-7
691 Eckert, S., Eilers, E.J., Jakobs, R., Anaia, R.A., Aragam, K.S., Bloss, T., Popp, M., Sasidharan, R.,
692 Schnitzler, J.-P., Stein, F., Steppuhn, A., Unsicker, S.B., van Dam, N.M., Yepes, S., Ziaja, D.,
693 Müller, C., 2023. Inter-laboratory comparison of plant volatile analyses in the light of intra-specific
694 chemodiversity. Metabolomics 19, 62. https://doi.org/10.1007/s11306-023-02026-6
695 Eilers, E.J., Kleine, S., Eckert, S., Waldherr, S., Müller, C., 2021. Flower Production, Headspace
696 Volatiles, Pollen Nutrients, and Florivory in Tanacetum vulgare Chemotypes. Frontiers in Plant
697 Science 11, 611877. https://doi.org/10.3389/fpls.2020.611877
698 Eisenring, M., Glauser, G., Meissle, M., Romeis, J., 2018. Differential Impact of Herbivores from
699 Three Feeding Guilds on Systemic Secondary Metabolite Induction, Phytohormone Levels and
700 Plant-Mediated Herbivore Interactions. Journal of Chemical Ecology 44, 1178–1189.
701 https://doi.org/10.1007/s10886-018-1015-4
702 Erb, M., Meldau, S., Howe, G.A., 2012. Role of phytohormones in insect-specific plant reactions.
703 Trends in Plant Science 17, 250–259. https://doi.org/10.1016/j.tplants.2012.01.003
704 Fernandes, F.L., Picanço, M.C., Gontijo, P.C., De Sena Fernandes, M.E., Pereira, E.J.G., Semeão,
705 A.A., 2011. Induced responses of _Coffea arabica_ to attack of _Coccus viridis_ stimulate locomotion
706 of the herbivore. Entomologia Experimentalis et Applicata 139, 120–127.
707 https://doi.org/10.1111/j.1570-7458.2011.01113.x
708 Fox, J., Weisberg, S., 2019. An R companion to applied regression, Third edition. ed. SAGE, Los
709 Angeles London New Delhi Singapore Washington, DC Melbourne.
710 Frost, C.J., Appel, H.M., Carlson, J.E., De Moraes, C.M., Mescher, M.C., Schultz, J.C., 2007. Within‐
711 plant signalling via volatiles overcomes vascular constraints on systemic signalling and primes
712 responses against herbivores. Ecology Letters 10, 490–498. https://doi.org/10.1111/j.1461713 0248.2007.01043.x
714 Gaar, S., Müller, C., Dussarrat, T., 2026. Impacts of different types of florivores on flower
715 metabolomes in the field. bioRxiv Preprint https://doi.org/10.64898/2026.04.30.721624
716 Gols, R., 2014. Direct and indirect chemical defences against insects in a multitrophic framework.
717 Plant, Cell &amp; Environment 37, 1741–1752. https://doi.org/10.1111/pce.12318
718 Gosset, V., Harmel, N., Gobel, C., Francis, F., Haubruge, E., Wathelet, J.-P., Du Jardin, P., Feussner,
719 I., Fauconnier, M.-L., 2009. Attacks by a piercing-sucking insect ( _Myzus persicae_ Sultzer) or a
720 chewing insect ( _Leptinotarsa decemlineata_ Say) on potato plants ( _Solanum tuberosum_ L.) induce
721 differential changes in volatile compound release and oxylipin synthesis. Journal of Experimental
722 Botany 60, 1231–1240. https://doi.org/10.1093/jxb/erp015
723 Grof-Tisza, P., Kruizenga, N., Tervahauta, A.I., Blande, J.D., 2022. Volatile-Mediated Induced and
724 Passively Acquired Resistance in Sagebrush ( _Artemisia tridentata_ ). Journal of Chemical Ecology
725 48, 730–745. https://doi.org/10.1007/s10886-022-01378-y
726 Heil, M., Karban, R., 2010. Explaining evolution of plant communication by airborne signals. Trends
727 in Ecology &amp; Evolution 25, 137–144. https://doi.org/10.1016/j.tree.2009.09.010


25


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


728 Hoffmann, M.A., Nothias, L.-F., Ludwig, M., Fleischauer, M., Gentry, E.C., Witting, M., Dorrestein,
729 P.C., Dührkop, K., Böcker, S., 2022. High-confidence structural annotation of metabolites absent
730 from spectral libraries. Nature Biotechnology 40, 411–421. https://doi.org/10.1038/s41587-021731 01045-9
732 Jakobs, R., Müller, C., 2018. Effects of intraspecific and intra-individual differences in plant quality
733 on preference and performance of monophagous aphid species. Oecologia 186, 173–184.
734 https://doi.org/10.1007/s00442-017-3998-x
735 Jakobs, R., Schweiger, R., Müller, C., 2019. Aphid infestation leads to plant part‐specific changes in
736 phloem sap chemistry, which may indicate niche construction. New Phytologist 221, 503–514.
737 https://doi.org/10.1111/nph.15335
738 Keskitalo, M., Pehu, E., Simon, J.E., 2001. Variation in volatile compounds from tansy ( _Tanacetum_
739 _vulgare_ L.) related to genetic and morphological differences of genotypes. Biochemical
740 Systematics and Ecology 29, 267–285. https://doi.org/10.1016/S0305-1978(00)00056-9
741 Kessler, A., Chautá, A., 2020. The ecological consequences of herbivore-induced plant responses on
742 plant–pollinator interactions. Emerging Topics in Life Sciences 4, 33–43.
743 https://doi.org/10.1042/ETLS20190121
744 Kessler, A., Halitschke, R., 2009. Testing the potential for conflicting selection on floral chemical
745 traits by pollinators and herbivores: predictions and case study. Functional Ecology 23, 901–912.
746 https://doi.org/10.1111/j.1365-2435.2009.01639.x
747 Kessler, D., Diezel, C., Clark, D.G., Colquhoun, T.A., Baldwin, I.T., 2013. Petunia flowers solve the
748 defence/apparency dilemma of pollinator attraction by deploying complex floral blends. Ecology
749 Letters 16, 299–306. https://doi.org/10.1111/ele.12038
750 Kleine, S., Müller, C., 2011. Intraspecific plant chemical diversity and its relation to herbivory.
751 Oecologia 166, 175–186. https://doi.org/10.1007/s00442-010-1827-6
752 Kolde, R., 2025. pheatmap: Pretty Heatmaps.
753 Koornneef, A., Pieterse, C.M.J., 2008. Cross Talk in Defense Signaling. Plant Physiology 146, 839–
754 844. https://doi.org/10.1104/pp.107.112029
755 Krassowski, M., Arts, M., Lagger, C., Max, 2022. krassowski/complex-upset: v1.3.5.
756 https://doi.org/10.5281/ZENODO.3700590
757 Lee, K., Seo, P.J., 2022. Wound-Induced Systemic Responses and Their Coordination by Electrical
758 Signals. Frontiers in Plant Science 13. https://doi.org/10.3389/fpls.2022.880680
759 Lenth, R.V., Piaskowski, J., Banfai, B., Bolker, B., Buerkner, P., Giné-Vázquez, I., Hervé, M., Jung,
760 M., Love, J., Miguez, F., Riebl, H., Singmann, H., Derumigny, A., Schmidt, P., 2026. emmeans:
761 Estimated Marginal Means, aka Least-Squares Means.
762 Lucas‐Barbosa, D., Sun, P., Hakman, A., Van Beek, T.A., Van Loon, J.J.A., Dicke, M., 2016. Visual
763 and odour cues: plant responses to pollination and herbivory affect the behaviour of flower visitors.
764 Functional Ecology 30, 431–441. https://doi.org/10.1111/1365-2435.12509
765 Mamin, M., Clancy, M.V., Flückiger, G., Quijano-Medina, T., Pérez-Niño, B., Abdala-Roberts, L.,
766 Turlings, T.C.J., Bustos-Segura, C., 2025. Induction by caterpillars of stored and emitted volatiles
767 in terpene chemotypes from populations of wild cotton ( _Gossypium hirsutum_ ). BMC Plant Biology
768 25, 127. https://doi.org/10.1186/s12870-025-06088-7
769 McCall, A.C., Irwin, R.E., 2006. Florivory: the intersection of pollination and herbivory. Ecology
770 Letters 9, 1351–1365. https://doi.org/10.1111/j.1461-0248.2006.00975.x
771 Mithöfer, A., Boland, W., 2012. Plant Defense Against Herbivores: Chemical Aspects. Annual
772 Review of Plant Biology 63, 431–450. https://doi.org/10.1146/annurev-arplant-042110-103854
773 Mondello, 2015. Mass spectra of flavors and fragrances of natural and synthetic compounds FFNSC 3.
774 Library.
775 Moore, B.D., Andrew, R.L., Külheim, C., Foley, W.J., 2014. Explaining intraspecific diversity in plant
776 secondary metabolites in an ecological context. New Phytologist 201, 733–750.
777 https://doi.org/10.1111/nph.12526
778 Mostafa, S., Wang, Y., Zeng, W., Jin, B., 2022. Plant Responses to Herbivory, Wounding, and
779 Infection. IJMS 23, 7031. https://doi.org/10.3390/ijms23137031
780 Müller, C., Dussarrat, T., Dam, N.M. van, 2026. What is a plant chemotype, anyway? Trends in Plant
781 Science 31, 1292–1302. https://doi.org/10.1016/j.tplants.2026.05.003
782 Newrzella, H., Heinen, R., Villasante, C.M., Zimmer, I., Weber, B., Kary, P., Gerl, G., Neuhaus, A.,
783 Sigalas, A., Ojeda‐Prieto, L., Winkler, J.B., Weisser, W.W., Schnitzler, J. ‐P., 2026. Organ‐specific


26


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


784 terpenoid responses in _Tanacetum vulgare_ are chemotype‐dependent. Plant Biology 28, 1437–
785 1453. https://doi.org/10.1111/plb.70208
786 NIST, 2014. National Institute of Standard and Technology. Library.
787 Orians, C., 2005. Herbivores, Vascular Pathways, and Systemic Induction: Facts and Artifacts. Journal
788 of Chemical Ecology 31, 2231–2242. https://doi.org/10.1007/s10886-005-7099-7
789 Pang, Z., Lu, Y., Zhou, G., Hui, F., Xu, L., Viau, C., Spigelman, A.F., MacDonald, P.E., Wishart,
790 D.S., Li, S., Xia, J., 2024. MetaboAnalyst 6.0: towards a unified platform for metabolomics data
791 processing, analysis and interpretation. Nucleic Acids Research 52, W398–W406.
792 https://doi.org/10.1093/nar/gkae253
793 Pareja, M., Qvarfordt, E., Webster, B., Mayon, P., Pickett, J., Birkett, M., Glinwood, R., 2012.
794 Herbivory by a Phloem-Feeding Insect Inhibits Floral Volatile Production. PLoS One 7, e31971.
795 https://doi.org/10.1371/journal.pone.0031971
796 Pieterse, C.M.J., Van Der Does, D., Zamioudis, C., Leon-Reyes, A., Van Wees, S.C.M., 2012.
797 Hormonal Modulation of Plant Immunity. Annual Review of Cell and Developmental Biology 28,
798 489–521. https://doi.org/10.1146/annurev-cellbio-092910-154055
799 R Core Team, 2026. R: A Language and Environment for Statistical Computing.
800 Raguso, R.A., 2016. More lessons from linalool: insights gained from a ubiquitous floral volatile.
801 Current Opinion in Plant Biology 32, 31–36. https://doi.org/10.1016/j.pbi.2016.05.007
802 Ritchie, M.E., Phipson, B., Wu, D., Hu, Y., Law, C.W., Shi, W., Smyth, G.K., 2015. limma powers
803 differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids
804 Research 43, e47. https://doi.org/10.1093/nar/gkv007
805 Rohart, F., Gautier, B., Singh, A., Cao, K.-A.L., 2017. mixOmics: An R package for ‘omics feature
806 selection and multiple data integration. PLoS Computational Biology 13, e1005752.
807 https://doi.org/10.1371/journal.pcbi.1005752
808 Rusman, Q., Hooiveld‐Knoppers, S., Dijksterhuis, M., Bloem, J., Reichelt, M., Dicke, M., Poelman,
809 E.H., 2022. Flowers prepare thyselves: leaf and root herbivores induce specific changes in floral
810 phytochemistry with consequences for plant interactions with florivores. New Phytologist 233,
811 2548–2560. https://doi.org/10.1111/nph.17931
812 Rusman, Q., Lucas-Barbosa, D., Poelman, E.H., Dicke, M., 2019. Ecology of Plastic Flowers. Trends
813 in Plant Science 24, 725–740. https://doi.org/10.1016/j.tplants.2019.04.007
814 Sasidharan, R., Brokate, L., Eilers, E.J., Müller, C., 2023. Chemodiversity in flowers of _Tanacetum_
815 _vulgare_ has consequences on a florivorous beetle. Plant Biology 25, 1071–1082.
816 https://doi.org/10.1111/plb.13576
817 Schiestl, F.P., 2015. Ecology and evolution of floral volatile-mediated information transfer in plants.
818 New Phytologist 206, 571–577. https://doi.org/10.1111/nph.13243
819 Schiestl, F.P., Kirk, H., Bigler, L., Cozzolino, S., Desurmont, G.A., 2014. Herbivory and floral
820 signaling: phenotypic plasticity and tradeoffs between reproduction and indirect defense. New
821 Phytologist 203, 257–266. https://doi.org/10.1111/nph.12783
822 Schmitz, G., 1998. The phytophagous insect fauna of _Tanacetum vulgare_ L. (Asteraceae) in Central
823 Europe. Beiträge zur Entomologie 48, 219–235. https://doi.org/10.21248/contrib.entomol.48.1.219824 235
825 Schuman, M.C., Baldwin, I.T., 2016. The Layers of Plant Responses to Insect Herbivores. Annual
826 Review of Entomology 61, 373–394. https://doi.org/10.1146/annurev-ento-010715-023851
827 Schweiger, R., Castells, E., Da Sois, L., Martínez-Vilalta, J., Müller, C., 2021. Highly Species828 Specific Foliar Metabolomes of Diverse Woody Species and Relationships with the Leaf
829 Economics Spectrum. Cells 10, 644. https://doi.org/10.3390/cells10030644
830 Schweiger, R., Heise, A.-M., Persicke, M., Müller, C., 2014. Interactions between the jasmonic and
831 salicylic acid pathway modulate the plant metabolome and affect herbivores of different feeding
832 types. Plant, Cell &amp; Environment 37, 1574–1585. https://doi.org/10.1111/pce.12257
833 Schymanski, E.L., Jeon, J., Gulde, R., Fenner, K., Ruff, M., Singer, H.P., Hollender, J., 2014.
834 Identifying Small Molecules via High Resolution Mass Spectrometry: Communicating Confidence.
835 Environmental Science and Technology 48, 2097–2098. https://doi.org/10.1021/es5002105
836 Smallegange, R.C., van Loon, J.J.A., Blatt, S.E., Harvey, J.A., Agerbirk, N., Dicke, M., 2007. Flower
837 vs. Leaf Feeding by _Pieris brassicae_ : Glucosinolate-Rich Flower Tissues are Preferred and Sustain
838 Higher Growth Rate. Journal of Chemical Ecology 33, 1831–1844. https://doi.org/10.1007/s10886839 007-9350-x


27


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


840 Soper Gorden, N.L., Adler, L.S., 2016. Florivory shapes both leaf and floral interactions. Ecosphere 7,
841 e01326. https://doi.org/10.1002/ecs2.1326
842 Sotelo, P., Pérez, E., Najar-Rodriguez, A., Walter, A., Dorn, S., 2014. Brassica Plant Responses to
843 Mild Herbivore Stress Elicited by Two Specialist Insects from Different Feeding Guilds. Journal of
844 Chemical Ecology 40, 136–149. https://doi.org/10.1007/s10886-014-0386-4
845 Tegeder, M., Masclaux‐Daubresse, C., 2018. Source and sink mechanisms of nitrogen transport and
846 use. New Phytologist 217, 35–53. https://doi.org/10.1111/nph.14876
847 Theis, N., Lerdau, M., Raguso, R.A., 2007. The Challenge of Attracting Pollinators While Evading
848 Floral Herbivores: Patterns of Fragrance Emission in _Cirsium arvense_ and _Cirsium repandum_
849 (Asteraceae). International Journal of Plant Sciences 168, 587–601. https://doi.org/10.1086/513481
850 Tissier, A., Morgan, J.A., Dudareva, N., 2017. Plant Volatiles: Going ‘In’ but not ‘Out’ of Trichome
851 Cavities. Trends in Plant Science 22, 930–938. https://doi.org/10.1016/j.tplants.2017.09.001
852 Tunes, P., Canaveze, Y., Machado, S.R., Dötterl, S., Guimarães, E., 2026. Resilience of floral scent
853 emission after florivory. Plant Biol J 28, 1491–1506. https://doi.org/10.1111/plb.70209
854 van den Dool, H., Kratz, P.D., 1963. A generalization of the retention index system including linear
855 temperature programmed gas—liquid partition chromatography. Journal of Chromatography A 11,
856 463–471. https://doi.org/10.1016/S0021-9673(01)80947-X
857 van Leur, H., Vet, L.E.M., van der Putten, W.H., van Dam, N.M., 2008. Barbarea vulgaris
858 Glucosinolate Phenotypes Differentially Affect Performance and Preference of Two Different
859 Species of Lepidopteran Herbivores. Journal of Chemical Ecology 34, 121–131.
860 https://doi.org/10.1007/s10886-007-9424-9
861 Venables, W.N., Ripley, B.D., 2002. Modern applied statistics with S, 4th ed. ed, Statistics and
862 computing. Springer, New York.
863 Walling, L.L., 2000. The Myriad Plant Responses to Herbivores. Journal of Plant Growth Regulation
864 19, 195–216. https://doi.org/10.1007/s003440000026
865 War, A.R., Paulraj, M.G., Ahmad, T., Buhroo, A.A., Hussain, B., Ignacimuthu, S., Sharma, H.C.,
866 2012. Mechanisms of plant defense against insect herbivores. Plant Signaling &amp; Behavior 7, 1306–
867 1320. https://doi.org/10.4161/psb.21663
868 Wetzel, W.C., Whitehead, S.R., 2020. The many dimensions of phytochemical diversity: linking
869 theory to practice. Ecology Letters 23, 16–32. https://doi.org/10.1111/ele.13422
870 Wickham, H., Averick, M., Bryan, J., Chang, W., McGowan, L., François, R., Grolemund, G., Hayes,
871 A., Henry, L., Hester, J., Kuhn, M., Pedersen, T., Miller, E., Bache, S., Müller, K., Ooms, J.,
872 Robinson, D., Seidel, D., Spinu, V., Takahashi, K., Vaughan, D., Wilke, C., Woo, K., Yutani, H.,
873 2019. Welcome to the Tidyverse. JOSS 4, 1686. https://doi.org/10.21105/joss.01686
874 Wolf, V.C., Berger, U., Gassmann, A., Müller, C., 2011. High chemical diversity of a plant species is
875 accompanied by increased chemical defence in invasive populations. Biological Invasions 13,
876 2091–2102. https://doi.org/10.1007/s10530-011-0028-5
877 Zangerl, A.R., Berenbaum, M.R., 2009. Effects of florivory on floral volatile emissions and
878 pollination success in the wild parsnip. Arthropod-Plant Interactions 3, 181–191.
879 https://doi.org/10.1007/s11829-009-9071-x
880 Zhang, W., Jiang, Y., Chen, S., Chen, Fadi, Chen, Feng, 2021. Concentration-dependent emission of
881 floral scent terpenoids from diverse cultivars of _Chrysanthemum morifolium_ and their wild
882 relatives. Plant Science 309, 110959. https://doi.org/10.1016/j.plantsci.2021.110959
883 Ziaja, D., Müller, C., 2025. Intraspecific and intra‐individual chemodiversity and phenotypic
884 integration of terpenes across plant parts and development stages in an aromatic plant. Plant
885 Biology 27, 637–650. https://doi.org/10.1111/plb.13763
886 Ziaja, D., Müller, C., 2023. Intraspecific chemodiversity provides plant individual- and
887 neighbourhood-mediated associational resistance towards aphids. Frontiers in Plant Science 14,
888 1145918. https://doi.org/10.3389/fpls.2023.1145918
889


890 **Supporting Information**


891 Additional supporting information can be found online in the Supporting Information section


28


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


892 **Table**


893 **TABLE 1:** Results of likelihood ratio tests (LR χ²) from the negative binomial generalised linear


894 model for the number of responding features (A) and from the binomial generalised linear model for


895 the proportion of increased features (number of increased out of all significant features, B), as affected


896 by chemotype, insect species, organ and interactions.

|Factor|LRχ²|df|p-value|
|---|---|---|---|
|**A: Number of responding features**|**A: Number of responding features**|||
|Chemotype|92.989|4|**&lt; 0.001**|
|Insect|5.004|1|**0.025**|
|Organ|0.305|1|0.581|
|Chemotype × Insect|26.333|4|**&lt; 0.001**|
|Organ × Insect|5.882|1|**0.015**|
|**B: Proportion of increased features**|**B: Proportion of increased features**|||
|Chemotype|7.494|4|0.112|
|Insect|10.636|1|**0.001**|
|Organ|6.639|1|**0.010**|
|Chemotype × Insect|9.981|4|**0.041**|
|Organ × Insect|1.211|1|0.271|



897 LR χ² = likelihood ratio chi-square statistic; df = degrees of freedom; Significant effects ( _p_ &lt; 0.05) are


898 shown in bold


29


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


899


900 **FIGURE 1.** Experimental setup for testing chemotype-specific flower and leaf responses of


901 _Tanacetum vulgare_ to infestation by florivorous insects. Plants of five chemotypes &#91;Aacet, artemisyl


902 acetate/artemisia ketone/artemisia alcohol mix chemotype: _n_ = 6; ABThu, α-/β-thujone mix


903 chemotype: _n_ = 7; BThu, β-thujone chemotype: _n_ = 9; Keto, artemisia ketone chemotype: _n_ = 4;


904 Myrox, ( _Z_ )-myroxide/santolina-triene/artemisyl acetate mix chemotype: _n_ = 4&#93; were each cloned three


905 times and each clone received one of three treatments (aphids, beetles, control). Depending on the


906 number of flower heads per bouquet, 5-15 flower heads were enclosed in cups and either two aphids


907 ( _Macrosiphoniella tanacetaria_ ) added per flower head, one beetle ( _Olibrus_ spp.) per three flower


908 heads or no insects (control). After three days, volatile organic compounds were collected on PDMS


909 tubes (analysis via TD-GC-MS). On day four, flower heads and the second youngest unfolded leaf


910 were harvested and analysed for terpenes (GC–MS) and metabolic fingerprints (UHPLC–QTOF–


911 MS/MS).


30


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


912


913 **FIGURE 2.** Changes of floral volatile organic compounds (VOCs) of _Tanacetum vulgare_ of different


914 chemotypes in response to florivory. Chemotypes are abbreviated as follows: Aacet, artemisyl


915 acetate/artemisia ketone/artemisia alcohol mix chemotype; ABThu, α-/β-thujone mix chemotype;


916 BThu, β-thujone chemotype; Keto, artemisia ketone chemotype; and Myrox, ( _Z_ )-myroxide/santolina

917 triene/artemisyl acetate mix chemotype. (A) Heatmap of the group mean values of normalised


918 abundances of floral VOCs that differed significantly between treatments in moderated t-tests and fold


919 change analyses (aphid or beetle versus control, _p_ &lt; 0.05; fold change &lt; 0.5 for decreases or &gt; 2 for


920 increases), analysed across all samples and chemotype-specifically. Rows correspond to treatment


921 groups within each chemotype, columns correspond to the significantly responding VOCs, clustered


922 based on Pearson correlation and colours of different intensities represent normalised abundances. (B)


923 Upset plot showing the number of significantly responding floral VOCs identified by moderated t-tests


924 and fold changes, conducted for each chemotype individually. Numbers in parentheses on the left


925 indicate the total number of significant VOCs within each group, while bars on the top represent


926 intersections between groups, illustrating shared and unique volatile responses to insect infestation. In


927 total, 136 floral VOCs were detected across all samples.


31


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


928


929 **FIGURE 3.** Stored terpene profiles of _Tanacetum vulgare_ flower heads and leaves in response to
930 florivory. (A) Multilevel PCA score plot showing the separation of samples between the two organs in
931 a paired dataset comprising plant clones. Analysis was performed on normalised GC-MS data (159
932 terpenes). Ellipses indicate the 95 % confidence intervals for each group. (B, C) Heatmaps of the
933 group mean values of normalised abundances of terpenes that differed significantly between
934 treatments in moderated t-tests and fold change analyses (aphid or beetle _versus_ control, _p_ &lt; 0.05; fold
935 change &lt; 0.5 for decreases or &gt; 2 for increases) in (B) flower heads and (C) leaves. Analyses were
936 performed using the pooled data set and chemotype-specifically. Rows correspond to treatment groups
937 within each chemotype, columns correspond to the terpenes, clustered based on Pearson correlation
938 and colours of different intensities represent normalised abundances. (D) Upset plot showing the
939 numbers of terpenes significantly responding to florivory, identified by moderated t-tests and fold940 change analyses for each organ-chemotype combination, aphid or beetle _versus_ control. Numbers in
941 parentheses on the left indicate the total number of significantly responding terpenes within each
942 group, while bars on the top represent intersections between groups, illustrating shared and unique
943 terpene responses to insect infestation. Only groups with at least one significantly responding terpene
944 are shown; groups without significant responses were omitted from the plot. Abbreviations used in
945 (B)-(D): Aacet, artemisyl acetate/artemisia ketone/artemisia alcohol mix chemotype; ABThu, α-/β946 thujone mix chemotype; BThu, β-thujone chemotype; Keto, artemisia ketone chemotype; and Myrox,
947 ( _Z_ )-myroxide/santolina-triene/artemisyl acetate mix chemotype; A, aphid; B, beetle; F, flower head; L,
948 leaf.


32


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


949


950 **FIGURE 4.** Metabolic fingerprints of _Tanacetum vulgare_ flower heads and leaves in response to
951 florivory. (A) Multilevel PCA score plot showing the separation of samples between the two organs in
952 a paired dataset comprising plant clones. Analysis was based on normalised data of LC-MS metabolic
953 fingerprints (7176 features). Ellipses indicate the 95 % confidence intervals for each group. Significant
954 florivory-responsive features in (B-D) were identified by moderated t-tests and fold-change analyses
955 conducted separately for each chemotype (aphid or beetle _versus_ control, _p_ &lt; 0.05 and fold change &lt;
956 0.5 for decreases or &gt; 2 for increases). (B, C) Upset plots showing the numbers of features
957 significantly responding to florivory in (B) flowers or (C) leaves. Numbers in parentheses on the left
958 indicate the total number of significant features within each group, whereas top bars show
959 intersections between groups, illustrating shared and unique metabolic responses to insect infestation.
960 Only intersections of ≥ 3 features are shown. (D) Bar plot showing the percentage of significantly
961 responding features with increased abundance relative to the control following florivory across
962 chemotypes. Percentages are based on all significant features per chemotype, organ and treatment
963 (total numbers are shown in white within the bars); the remaining percentage represents significantly
964 decreased features. Numbers in parentheses below each chemotype indicate unique features per
965 chemotype. The dashed horizontal line indicates 50 %, with values above or below indicating a
966 predominance of increased or decreased features, respectively. Aacet, artemisyl acetate/artemisia
967 ketone/artemisia alcohol mix chemotype; ABThu, α-/β-thujone mix chemotype; BThu, β-thujone
968 chemotype; Keto, artemisia ketone chemotype; and Myrox, ( _Z_ )-myroxide/santolina-triene/artemisyl
969 acetate mix chemotype; A, aphid; B, beetle; F, flower head; L, leaf.


33


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


970


971 **FIGURE 5.** Pathway classification of significant features induced in _Tanacetum vulgare_ flower heads


972 and leaves by aphid and beetle florivory. Significant features were identified by moderated t-tests and


973 fold-change analyses conducted using the pooled data set as well as chemotype-specifically (aphid or


974 beetle _versus_ control, _p_ &lt; 0.05 and fold change &lt; 0.5 for decreases or &gt; 2 for increases). (A) Bar plots


975 showing the number of significant florivory-responsive features, grouped by chemical pathway. The


976 numbers below the pathway labels on the left indicate the total number of unique significant features


977 assigned to each pathway (1399 features in total). For each pathway, the sum of features across the


978 four groups may exceed the total number of significant features assigned to that pathway, because a


979 single feature can be significant in more than one group. ND = not determined. (B) Pie charts showing


980 the distribution of florivory-responsive features across chemical pathways for each chemotype,


981 regardless of organ and insect species. Numbers in parentheses indicate the total number of responding


982 features within each chemotype. Chemotypes are abbreviated as follows: Aacet, artemisyl


983 acetate/artemisia ketone/artemisia alcohol mix chemotype; ABThu, α-/β-thujone mix chemotype;


984 BThu, β-thujone chemotype; Keto, artemisia ketone chemotype; and Myrox, ( _Z_ )-myroxide/santolina

985 triene/artemisyl acetate mix chemotype.


34


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.13.766439; this version posted November 16, 2026. The copyright holder for this&#93;(https://doi.org/10.64898/2026.11.13.766439)

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
&#91;perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.&#93;(http://creativecommons.org/licenses/by-nc-nd/4.0/)


986


987 **FIGURE 6.** Overall comparison of florivore-induced metabolic responses across three analytical


988 platforms: floral VOCs (TD-GC-MS), stored terpenes (GC-MS) and untargeted metabolomic


989 fingerprints (UHPLC-QTOF-MS/MS). The figure summarises numbers of significant metabolites in


990 _Tanacetum vulgare_ floral volatiles _,_ flower heads and leaves following aphid or beetle florivory.


991 Significant metabolites were identified separately for each chemotype using moderated t-tests and


992 fold-change analyses (aphid or beetle _versus_ control, _p_ &lt; 0.05 and fold change &lt; 0.5 for decreases or &gt;


993 2 for increases). Venn diagrams show metabolites responding to aphid or beetle florivory or


994 differences among flower heads and leaves, with shared metabolites presented in the intersections.


35


