Ultra-Violet Induced Biochemical Changes in an Invasive Weed and their Implications for Plant-Biocontrol Agent Interactions

TL;DR

This study explores how UV radiation affects the biochemistry of an invasive shrub and its impact on a biocontrol beetle's performance.

Contribution

The study investigates UV-induced biochemical changes in an invasive plant and their implications for biocontrol agent effectiveness.

Findings

Heather plants exposed to different UV levels showed significant changes in secondary metabolites from the shikimate-phenylpropanoid pathway.

Bioassays showed no impact of these biochemical changes on the biocontrol beetle's prepupal weight or larval survival.

UV-induced changes may not directly affect the beetle's performance despite altering plant biochemistry.

Abstract

Introducing insect biocontrol agents sourced from a plant’s native range is an effective, sustainable management strategy for invasive plants. However, not all biocontrol programmes achieve the desired outcome because control agents either fail to establish or are ineffective. Heather beetle Lochmaea suturalis (Coleoptera: Chrysomelidae), introduced from the United Kingdom (UK) to New Zealand (NZ) to control the invasive shrub Calluna vulgaris (heather), was difficult to establish and achieved poor population growth rates and expansion relative to its conspecifics in its native UK range. Poor performance in biocontrol is often attributed to various abiotic or biotic factors but seldom considers alterations to a target plants biochemical phenotype. A recent study revealed, heather has a significantly different biochemical profile in NZ compared with the UK, between which there is considerable difference in ultra-violet (UV) radiation. UV is known to drive plant biochemical change, including defensive secondary metabolites and we hypothesized that this factor could enhance heathers’ defensive capability leading to poor biocontrol agent performance. Testing this hypothesis involved exposing heather plants to 20% and 95% UV attenuating screens and using metabolomics to measure plant secondary metabolite responses. Our results demonstrate significant alterations to many compounds derived from the shikimate-phenylpropanoid pathway. However, a bioassay revealed no impact on prepupal weight or larval survival of the biocontrol agent L. suturalis. We discuss and explore possible reasons for this outcome, the magnitude and impact of UV-induced biochemical changes on plant-insect interactions and the potential of metabolomics to support weed biocontrol.

The online version contains supplementary material available at 10.1007/s10886-026-01699-2.

Figures

b). All temperature data were subjected to a paired t-test (Minitab v 21.1.0.) to assess potential daytime temperature differences between treatments. Temperature variance between the 20% and 95% attn. treatments was checked by assessing the recordings obtained between 8:00 am and 4:00 pm on the 10th, 20th (summer solstice) and 30th Dec 2020. A minor difference of ∼ 1.17 °C was evident during the afternoon peak, but mean values throughout the three periods for 20% attn. (20.3 °C) and 95% attn. (20.5 °C) revealed no significant difference (*t*(292)= −0.399, *p* = 0.69) (see Supplemental Fig. 2)

Pathway enrichment analysis of *C. vulgaris* plants showing the pathways impacted when exposed to the 95% attenuation of UV treatment. Pathway impact value (x axis) and −log10(p) value of significance (y axis) indicate phenylpropanoid biosynthesis and the flavonoid biosynthesis pathways with impact factors of 0.078 and 0.015 and significance values of 10.94 and 10.17 -Log10(p), respectively. The size and colour of the dots represent the impact value, and the pathway *p* value respectively i.e. increasing *p* values change from yellow through to red

Intensity (au) and fold change (FC) values for the 12 reduced phenylpropanoid and quinic acid derived compounds in *Calluna vulgaris* (heather) plants when exposed to 95% UV attenuation as compared to those exposed to 20% attenuation. p* < 0.001, Mann-Whitney *U*-test. X = mean, ― = median. A FC of 1 = 0 change, FC of 1.50 would = 50% reduction etc. All metabolites in the top row plus chlorogenic acid and quinic acid in the second row have been recorded in the literature, to respond to variation in ultra-violet lightFig. 4Intensity (au) and fold change (FC) values for the 4 amplified phenylpro

Mean prepupal weight (mg) (a) and larval survival (b) of heather beetle (*Lochmaea suturalis*) feeding on 20% and 95% UV attenuated heather (*Calluna vulgaris*) foliage. No significant difference was observed between weights, *p* > 0.05 or survival, *p* > 0.05. Error bars = 1 SE

Full text

Introduction

Invasive plants can seriously threaten terrestrial habitats that they invade, with potential to cause considerable production and economic losses in managed ecosystems such as agriculture, horticulture or forestry. Similarly, ecological damage to environmentally sensitive ecosystems such as rangelands and conservation areas can impact essential ecosystem services (Rai et al. 2022) and cumulatively result in considerable losses to biodiversity at the species, community or ecosystem level (Roy et al. 2023).

For many of these ecosystems, the introduction of weed biocontrol agents sourced from the plant’s native range (classical biocontrol) offers a sustainable long-term alternative for managing invasive plants (Hayes et al. 2013; Schwarzländer et al. 2018). However, not all biocontrol programmes achieve the desired outcome due to agents failing to establish (Hayes et al. 2013) or being ineffective if they do establish (McClay and Balciunas 2005; Raghu and Dhileepan 2005). Biocontrol agent population establishment is known to be influenced by abiotic (Grevstad 1999; Harms et al. 2020) and in some cases biotic factors (Syrett 1983; Briese 1986; Schulz et al. 2019). Literature exploring abiotically induced changes to target plant biochemical phenotypes which might impose limitations on biocontrol agent establishment or effectiveness is scarce (but see Falla et al. 2023; Barrett et al. 2024). Wheeler and Schaffner (2013) also review the importance of understanding the interplay between target plant phytochemistry and biocontrol agent responses, performance and safety. Despite this, only a small number of publications exist investigating target plant constitutive phytochemistry in host range testing and agent performance (Vrieling and de Boer 1999; Wheeler 2005; Wheeler et al. 2007, 2014; Rapo et al. 2019).

Plants perceive light and use it as essential environmental cues. Photosynthetically active radiation (PAR), defined as the 400–700 nm waveband is essential for plant growth. They also detect a range of other quantitative (intensity per unit area) wavelengths which vary with time of day, season and/or latitude. These include the blue (450–495 nm), red (645–665 nm) and far-red (720–740 nm) wavebands. Qualitative wavelengths, that is, the balance of PAR, ultra-violet A (321–400 nm) and ultra-violet B (291–320 nm) similarly, act as important cues (Roberts and Paul 2006; Caldwell et al. 2007; Ballaré et al. 2012). Visible light, including red and blue wavelengths are sensed by plant photoreceptors e.g., phytochromes and cryptochromes (Wang et al. 2018), while ultra-violet B (UV-B) is sensed by the UV RESISTANCE LOCUS8 (UVR8) protein (Jenkins 2014). These elicit plant responses that modify growth, biochemical composition and morphology and regulate plant primary and secondary metabolism (Roberts and Paul 2006; Yang et al. 2018). Primary metabolites are linked to basic survival functions such as photosynthesis, growth or cell division, whereas secondary metabolites function in regulating primary metabolites and growth, but also importantly, they mediate ecological interactions with other organisms and the environment (Erb and Kliebenstein 2020).

Ultra-violet (UV) radiation influences the shikimic acid - phenylpropanoid metabolic pathway (Bassman 2004; Singh et al. 2021, 2023), resulting in changes to many phenylpropanoid derived metabolites such as flavonoids, flavonol glycosides and hydroxycinnamic acids (HCAs) (Kuhlmann and Müller 2010; Barnes et al. 2017). These compounds play an important role in photoprotection of plants, i.e. protecting against cellular DNA damage (Gill et al. 2015) while maintaining optimum photosynthetic capacity of the mesophyll (Wargent et al. 2015; Barnes et al. 2017). Additionally, UV-radiation can also affect other phenylpropanoid groups including phenolic acids, coumarins, chlorogenic acids (CGAs), quinic acids, stilbenes, terpenoids and tannins, many of which are involved in plant defence against insect herbivores (Bassman 2004; Zhang and Bjorn 2009; Escobar-Bravo et al. 2017; Kumar et al. 2020). This has been amply demonstrated in the literature where exposure of plants to high UV results in impaired insect herbivore performance (Izaguirre et al. 2007; Kuhlmann and Müller 2010; Qi et al. 2018), or when exposure to reduced UV improves herbivore performance or increases feeding rates (Rousseaux et al. 2001; Ballaré et al. 2012; Dinh et al. 2013; Ballaré 2014).

Heather, Calluna vulgaris (L.) Hull, is a highly invasive shrub, well established on the North Islands Central Plateau (CP) of New Zealand (NZ). Previous work has demonstrated the plasticity of secondary metabolites in C. vulgaris at different altitudes with reference to varying light and soils (Monschein et al. 2010), as well as seasonality (Jalal et al. 1982) in Europe, and in response to multiple abiotic factors in the CP plants (Effah et al. 2020a, b). We also have good evidence that C. vulgaris foliage in NZ has significantly elevated levels of both primary and secondary metabolites when compared to levels found in C. vulgaris from Scotland in the United Kingdom (UK) (Barrett et al. 2024). In the sub-alpine CP environment, peak summertime UV index (UVI) measures as high as 14.0 whereas the same index throughout the UK is ∼ 7.0 (Liley and McKenzie 2006). Thus, C. vulgaris is exposed to considerably higher levels of UV on the CP which may elevate the level of phenylpropanoid metabolites in these plants relative to those of the UK.

In 1996, heather beetle, Lochmaea suturalis (Thomson, 1866) (Coleoptera: Chrysomelidae), sourced from the UK was introduced into New Zealand (NZ) as a biocontrol agent for C. vulgaris in the CP region. The beetle was difficult to establish and achieved poor population growth rates and expansion relative to those recorded in its UK range (Pakeman et al. 2002; Peterson et al. 2007, 2024; Rosenburgh and Marrs 2010; Fowler et al. 2015). Now however, two and a half decades after the first releases, beetle populations are well established and causing widespread damage to C. vulgaris. We posit that reduced assimilability due to altered phytochemistry in C. vulgaris in the CP region would conceivably impose strong selective pressure on L. suturalis and induce rapid eco-evolutionary adaptation in this control agent (McEvoy et al. 2012; Szűcs et al. 2012, 2019) and result in greater population densities.

Thus, to help explain the poor early performance of this control agent, we sought to explore if elevated UV-radiation influences the secondary metabolome of C. vulgaris. Conversely, we might expect CP plants exposed to reduced levels of ultra-violet (with the application of UV attenuation screen technology) to display reduced levels of secondary phenylpropanoids and we hypothesize that artificially reducing exposure to UV-radiation will lead to a reduction in these metabolites in C. vulgaris. To test this hypothesis, we exposed potted C. vulgaris plants to 20% [‘UV-transparent’] and 95% [‘UV-opaque’] attenuating polyethylene screens over two growing seasons. Using high performance liquid chromatography – mass spectrometry (HPLC-MS) and non-targeted metabolomic techniques we measured the magnitude of change to the plant metabolome for each UV treatment. To determine any effect of altered metabolite levels on insect herbivore performance, foliage from plants in each treatment was used in a bioassay to test if larval survival and prepupal live weights of L. suturalis were affected.

Materials and Methods

Plants and Experimental Conditions

In November 2020, heather plants, entirely free of L. suturalis, were collected from a field population at Waiouru on the North Islands Central Plateau (Lat. −39.455413, Lon. 175.677705) and planted into large (30 cm dia.) pots while keeping all roots and associated soil intact. On 7th Nov 2020, eighteen plants were placed into each of two tunnel house treatments at the plant growth unit of Massey University, Palmerston North, with sufficient spacing to avoid shading from neighbouring plants. The treatments exposed plants to two different intensities of UV using UV-modifying polyethylene screens i.e. 95% attenuation (attn.) of UV [‘UV-opaque’ – Lumisol 018] and 20% attn. of UV [‘UV-transparent’ – Lumisol 019] (BPI Visqueen™, UK). On 15th Jan 2021, photosynthetically active radiation (PAR) and total UV irradiance for each screen were confirmed using a scanning UV-vis monochromatic spectroradiometer. During cloudless sunny conditions, 3 scans were recorded under each screen.

Average total UV irradiance (290–320 (UV-B) and 321–400 (UV-A) wavelengths combined) at 20% attn. and 95% attn. was 4178.8 and 336.7 uW⋅cm^− 2^⋅nm^− 1^, respectively. Therefore, total UV transmittance by the 95% attn. screen was considerably less (≤ 8.1% of total UV) than that of the 20% attn. screen. Average photosynthetically active radiation (PAR) (the 401–750 wavelengths), at 20% and 95% attn. was 783.6 and 802.7 uW⋅cm^− 2^⋅nm^− 1^ respectively, thus each were within 2.5% of PAR transmittance (see Supplemental Fig. 1a). During April (Southern hemisphere autumn) ambient UV significantly declines, thus on 1 st April 2021, all plants were placed outside into ambient light, where UV levels continued to decrease during winter to ∼ 1.0–2.0 uW⋅cm^− 2^⋅nm^− 1^ (McKenzie and Liley 2010) and then increase again in spring. All plants were placed back into their respective treatments on 1 st Oct 2021. Thus, plants were exposed to two summer seasons of these UV attenuation treatments.

Temperature was logged continuously (Tinytag™ Plus 2, TGP-4020; Gemini Data Loggers (UK) Ltd.) for each of the attenuation and ambient outside treatments. Overall mean temperatures varied seasonally with the highest weekly mean of 25.45 °C which included the highest daily peak (42.2 °C) in the second week of January and the lowest mean of 6.6 °C during the last week of June 2021. The lowest daily peak of −1.8 °C occurred during the second week of July 2021 (see Supplemental Fig. 1b). All temperature data were subjected to a paired t-test (Minitab v 21.1.0.) to assess potential daytime temperature differences between treatments. Temperature variance between the 20% and 95% attn. treatments was checked by assessing the recordings obtained between 8:00 am and 4:00 pm on the 10th, 20th (summer solstice) and 30th Dec 2020. A minor difference of ∼ 1.17 °C was evident during the afternoon peak, but mean values throughout the three periods for 20% attn. (20.3 °C) and 95% attn. (20.5 °C) revealed no significant difference (t(292)= −0.399, p = 0.69) (see Supplemental Fig. 2).

Humidity in each treatment was logged continuously using a Tinytag ^tm^ Plus 2, TGP-4500 (Gemini Data Loggers (UK) Ltd.) logger. Data were analysed using Tinytag™ Explorer software and MS Excel. Humidity throughout the experimental period averaged 70.75%. Water was supplied individually to each plant every second day, using drip feed irrigation at the rate of ∼ 1 L day^− 1^ i.e., four times 250 ml pot^−1^day^− 1^.

Metabolomic Analysis

Foliage samples for metabolomic analysis were collected on 14th Dec 2021, from the 18 plants in each treatment. Samples consisted of ∼ 30 fresh current seasons sprigs approximately 2 cm in length, which were cut and immediately frozen on dry ice before being transferred to −80 °C for storage. All samples were freeze dried then stored for 2 weeks at −20 °C prior to grinding to 150–50 μm particle size before extraction. For extraction, 50 ± 0.5 mg of ground sample were weighed into 2 mL microcentrifuge tubes, each containing a 2.5 mm glass bead. To each tube, 800 µL of pre-chilled chloroform: methanol (CHCl_3_:CH_3_OH; 1:1 v/v) was added and then homogenised for 5 min at 25 Hz sec^−1^ using a Retsch MM400 Mill/TissueLyser and stored for 1 h at −20 °C. Then, 400 ml of HPLC grade H_2_O was added to each tube, similarly homogenised, then centrifuged (Sigma 1–14 K microcentrifuge) at 4 °C and 11,000 RPM for 15 min to create a biphasic layer. From each sample, 2 × 200 µl aliquots of the upper layer were pipetted into 2 mL microcentrifuge tubes for C18 and HILIC chromatography analysis. A final 200 µl aliquot was added to a 150 ml tube, ultimately pooling every sample into a homogenous mix, then sub-aliquots (200 µl) of this mix were transferred into microcentrifuge tubes to use as quality control checks (QCs). All samples were then dried under a continuous flow of N_2_ (3.5 L min^−1^) at 38 °C for 60 min using a BT LabSystems sample concentrator, then immediately stored at −80 °C until reconstitution.

Reconstitution solvents which included an internal standard (cat no. MSK-QC-KIT) (Cambridge Isotope Laboratories Inc.) at the concentration of 10 µl ml^− 1^ were prepared for C18 analysis in acetonitrile: water (CH_3_CN: H_2_O; 1:9, v/v) and for HILIC in acetonitrile: water (CH_3_CN: H_2_O; 1:1, v/v). Immediately prior to HPLC-MS analysis, all samples plus seven QCs were reconstituted by adding 200 µl of solvent, vortexed until dissolved and transferred by pipette to a 200 µl glass insert in an amber autosampler vial, capped and kept chilled until loading. The sequence was: five vials of reconstitution solvent only (blanks), one amino acid standard (A9906; Sigma-Aldrich, NZ), two QCs, 36 samples with a QC every 9th slot to finish with a final QC and an amino acid.

Chromatography and tandem mass-spectrometry analysis of polar and semi-polar compounds were achieved using a Thermo LC–MS system (Thermo Fisher Scientific, Waltham, MA, USA) which consisted of a Dionex UltiMate 3000 UHPLC system coupled with a high-resolution Q Exactive Focus Quadrupole-Orbitrap mass spectrometer utilising heated electrospray ionisation run in both positive and negative modes. For semi-polar compounds, samples were cooled in the auto-sampler at 4 °C and a 5 µL aliquot was injected into a 1.9 μm Thermo Hypersil GOLD™ C18 column (UHPLC, 100 mm × 2.1 mm, Thermo Fisher Scientific, USA) at 25 °C with a gradient elution programme and a flow rate of 350 µL min^− 1^. For polar compounds, samples were cooled in the auto-sampler at 4 °C and a 5 µL aliquot was injected into a 5 μm ZIC^®^-pHILIC column (100 mm × 2.1 mm, Merck Darmstadt, Germany) at 25 °C with a gradient elution programme and a flow rate of 150 µL min^− 1^. All chromatography and mass spectral parameters are detailed in Supplemental Information (Doc. 1).

Thermo derived.raw files for each stream i.e., C18 pos and C18 neg and HILIC pos and HILIC neg modes, were converted to mzML format using MSConvertGUI (Adusumilli and Mallick 2017), uploaded into MZmine (Pluskal et al. 2020) to determine the appropriate baseline noise threshold and then into XCMS online https://xcmsonline.scripps.edu/ for feature detection, alignment and exploratory data analysis (Domingo-Almenara and Siuzdak 2020). Feature detection parameters for C18 data were: m/z deviation 10 ppm, min and max peak width 5 and 20 respectively, mzdiff 0.001, s/n threshold 20, prefilter intensity 1e4 and noise filter 2.5e4. For HILIC data the same parameters were 10 ppm, 10 and 60, 0.001, 20, 1e4 and 8e3. After downloading the output, raw mass spectrometry data were organised into data matrices comprising m/z, retention time and the corresponding ion intensity measurements for each feature making it suitable for statistical analysis.

Reduction of background variability in the data matrix for each stream was performed using a QC vs. Blank t-test thus allowing subtraction of features with p values > 0.05, and secondly, t-values corresponding to any features high in the blanks. These data matrices were each uploaded into MetaboAnalyst 6.0 (MA ver 6.0) (Pang et al. 2021) and data integrity checked to confirm the number of samples, number of peaks, missing values, and the number of treatment groups. No missing values were detected in any of the data sets. For filtration of variables showing low repeatability, the threshold to remove those with high percent relative standard deviation (RSD) was set at 30% to that of the QCs and the data normalised by auto-scaling (mean-centred and divided by the standard deviation of each variable) and Gaussian distribution confirmed so that feature mass intensities are comparable.

For each stream, using MA ver 6.0, we explored these data matrices, and compared the 20% and 95% UV attenuated treatments by subjecting them to orthogonal partial least squares discriminant analysis (OPLS-DA); the separation between treatments was statistically confirmed using 20 permutations to provide acceptable R^2^, Q^2^, and p values. Features for annotation were identified by applying a paired t-test to these data matrices and retaining all the features below the false discovery threshold value of FDR < 0.05 (Benjamini and Hochberg 1995).

Annotations were conducted and confidence levels confirmed (Sumner et al. 2007) for each metabolite by interrogating the original.raw files using Xcalibur Freestyle. Formula matches were confirmed, and mass accuracy parameters were set within +/– 10.0 ppm. For all C18 features, confidence levels were set according to Sumner et al. (2007) with level 2 (parent ion plus at least one fragment) and level 3 (the parent ion only) being confirmed using the MassBank.eu (https://massbank.eu/) spectral database. Where the same annotated and confirmed compound (ion) appeared in both positive and negative modes, the one with the highest signal intensity (au) was included in the final data table. A literature search provided identification of those annotated compounds that were (1) considered as defensive metabolites with potential to influence the bioassay trial and (2) have demonstrated responses to experimentally manipulated UV exposures. For each annotated compound, fold change values were calculated and non-parametric Mann-Whitney U-tests used to investigate ranked distributional differences between treatments for each.

Using the intensity data of all the annotated compounds listed in Table 1 and the pathways enrichment analysis platform in MA ver 6.0, we identified the metabolic pathways that are significantly impacted (i.e. either upregulated or down regulated) by exposure to 95% UV. The upland cotton (Gossypium hirsutum) pathway library in MA ver.6 was used as a reference.Table 1. Annotated metabolites from C18 Pos and C18 Neg streams. KEGG = that database’s entry reference No.; Strm = column mode; R/A indicates if compound is reduced R, (↓) or amplified A, (↑) in the Attenuation treatments; CL = confidence level of annotation whereby level 2 is the parent ion and at least one co-eluted fragment and 3 is the parent ion only; m/z = mass to charge ratio; rt. = retention time. Antiherbivore defence (Herb.) References. (a) Sing et al. 2021, (b) Dixit et al. 2017; (c) Hussain et al. 2018; (d) Kundu and Vadassery 2019; (e) Ramaroson et al. 2022. Ultra-violet (UV) References. (f) Ruhland et al. 2005, (g) Chowdhray et al. 2021, (h) Luis et al. 2007; (i) Zhang and Bjorn 2009, (j) Baker et al. 2020; k. Lavola et al. 1997; l. Holub et al. 2019Class/Sub ClassNameKEGGStrmR/ACLm/zrt.FragmentsHerb. Ref.UV.Ref.Hydroxycinnamic acidCaffeic acidC01481C18+↓2181.04953.16163.0381, 145.0278a.f.Hydroxycinnamic acidp-Coumaric acidC00811C18+↓2165.05466.05119.0509b.f.Hydroxycinnamic acidRosmarinic acidC01850C18-↓3359.07526.04g. h. i.Hydroxycinnamic acidSinapoyl malateC18-↓3339.07217.85j.Cinnamyl alcohol4-HydroxycoumarinC20414C18+↓3163.03893.16Coumarin Glycoside4-Methylumbelliferyl glucuronideC11584C18+↓3353.08653.17CoumarinCoumarinC05851C18+↓3147.04407.60c.Stilbene glycoside3,4’,5-Trihydroxystilbene-3-beta-D-glucosideC18-↓3389.12469.78i.Flavonoid glycosideKaempferol-3-Glucoside-2’’-p-coumaroylC18+↑3595.14441.87Flavonoid glycosideMyricetin-3-XylosideC18-↑3449.07339.04FlavonoidMyricetinC10107C18+↑3319.04469.03Quinic acidCaffeoylquinic acid (3-O-caffeoyalquinic acid)C00852C18+↓2355.10218.25163.0370, 181.0470, 164.0410, 135.0430Quinic acid1,5-Dicaffeoylquinic acid (Cynarine)C18-↓2515.12008.37351.0718Quinic acidChlorogenic acid (5-O-caffeoylquinic acid)C17147C18+↓2355.10213.16163.0394, 181.0466, 164.0422, 165.0430d. e.k.Quinic acidD-(-)-Quinic acid (Quinic Acid)C00296C18-↓3191.05568.24Quinic acid3-Feruloylquinic acidC18+↑2369.11777.59177.0528, 178.0604l.Terpene lactone(-) - BilobalidaeC07605C18-↓3325.09327.89Amino acid conjugateN-Fructosyl TyrosineC18+↓3344.13387.61

Larval Bioassay

Four days after the foliar samples were collected (18th Dec 2021) and using the same eighteen plants from each 20% and 95% UV attenuated treatments, a larval performance bioassay was conducted in the lab to assess if UV attenuation of C. vulgaris plant foliage impacted larval growth and mortality of the heather beetle L. suturalis. Here, our investigations are limited to the larval/pre-pupal life stages. Adults and larvae don’t co-occur in natural field populations, so adults would require foliage of different phenological stages to measure parameters such as fecundity and overwintering lipid deposition. Thus, thirty-six (200 mm L x 130 mm W x 210 mm H) mesh ventilated plastic cages were set up for 18 replicates each of 20% and 95% attenuated heather foliage. Foliage stems were mounted into Oasis ^tm^ polystyrene plugs that protruded through the cage floor to contact water below and one individual C. vulgaris plant provided foliage for the entirety of each replicate. All 36 cages were held under constant 22 °C and illuminated with 5000 Lm LED grow lights at 16:8 h light: dark. Ten lab reared, unfed, hatchling larvae were placed onto the foliage in each cage and fresh C. vulgaris sprigs were added every three days while leaving the earlier foliage in place to avoid disturbing feeding larvae.

Larvae were collected for weighing at the pre-pupal stage whereupon they drop to the cage floor and seek the humidity provided by the oasis plug. Pre-pupa were weighed (mg) using a Sartorius Quintix balance. The first collection occurred on the 3rd of Jan 2022 with the final one 10 days later. All larvae observed dead on the cage floor during the trial period were recorded and removed. Mean prepupal weights were calculated for each replicate within each treatment. Survival for each replicate was calculated as the number of weighed prepupa subtracted from the original ten placed in each cage. Overall percent mortality for each treatment was also calculated. Both mean prepupal weight and survival data were subjected to non-parametric Mann-Whitney U-test to compare the effects between treatments. We also tested if the bioassay prepupal weights differed from prepupa weights obtained from a natural field population. Using the non-parametric Kruskal-Wallis test we compared, prepupa weights from the two UV treatments and pre-pupa weight data collected from a field population on the CP close to where the experimental heather plants were originally sourced.

Results

UV-Radiation Induces Changes in the Secondary Metabolism of Heather

Our analyses show a clear separation between the metabolite profiles collected from C. vulgaris plants exposed for two summers to 20% (control) and the 95% UV-attenuation treatment. OPLS-DA (Fig. 1) revealed significant separation between the groups for the C18 positive stream, R^2^ = 0.99, Q^2^ = 0.74, p < 0.05 and for the C18 negative stream, R^2^ = 0.99, Q^2^ = 0.82, p < 0.05. No significant separation was observed for either of the HILIC positive or negative data sets (results not shown). A C18 column is considered the most suitable for analysing secondary metabolites due to its versatility in separating a wide range of semi-polar compounds. HILIC columns, however, are best suited to separate polar compounds and analyse primary metabolites (Xiao et al. 2012) which shows that under the tested conditions, secondary but not primary metabolism was influenced most by changes in UV-radiation.

Fig. 1. Separation of Calluna vulgaris (heather) plant metabolite profiles after exposure over two summers of 20% (control) and the 95% UV attenuation. OPLS-DA results (p < 0.05) reveal complete separation for both (a) C18 positive and (b) C18 negative ionisation streams. Each cluster represents the eighteen plants (dots) in each attenuation treatment with 95% confidence ellipses

UV-Treatments Influence the Phenylpropanoid Pathway and its Associated Defensive Compounds

This analysis identified the phenylpropanoid and flavonoid biosynthesis pathways with impact factors of 0.078 and 0.015 and significance values of 10.94 and 10.17 -Log10(p), respectively, as being the two most impacted when exposed to 95% attenuation. Impact on the flavonoid biosynthesis pathway is upregulated, represented by the flavonoids and on the phenylpropanoid pathway downregulated, represented by hydroxycinnamic acids, coumarins and caffeoylquinic acids (see Table 1). Other pathways such as the ubiquinone/terpenoid and flavone/flavonol biosynthesis pathways from these data indicate minor yet significant impacts, both with values of < 0.001 (Fig. 2).

Fig. 2. Pathway enrichment analysis of C. vulgaris plants showing the pathways impacted when exposed to the 95% attenuation of UV treatment. Pathway impact value (x axis) and −log10(p) value of significance (y axis) indicate phenylpropanoid biosynthesis and the flavonoid biosynthesis pathways with impact factors of 0.078 and 0.015 and significance values of 10.94 and 10.17 -Log10(p), respectively. The size and colour of the dots represent the impact value, and the pathway p value respectively i.e. increasing p values change from yellow through to red

From the C18 positive and negative streams combined, a total of eighteen compounds were annotated. Sixteen are derived from the Shikimate-phenylpropanoid pathway (see Table 1). Hydroxycinnamic acids, flavonoid glycosides, coumarins, caffeoylquinic acids and one each of a stilbene glycoside, cinnamyl alcohol and a flavonoid comprise this group. Two of the caffeoylquinic acids, 3-O and 5-O, each with the parent ion m/z 355.1021 were separated by the presence of the distinctive fragment m/z 135.043 belonging to 3-O-Caffeoylquinic acid (Yang et al. 2022). Two further compounds, a terpene lactone and an amino acid conjugate make up the remainder.

Six of these compounds were confirmed at level 2 confidence based on parent and fragment ion matches with library spectra and the remainder at level 3 (i.e., parent ion only so identified to class/sub class with the named compound being the most probable) (Table 1). All annotated compound confirmation data is provided in Supplemental Information (Doc 2).

Mass spectral relative signal intensity of ions is expressed as arbitrary units (au). These intensities and the fold change (FC) ratio for each compound resulting from the 95% attenuation treatment as compared to the 20% control are provided in Fig. 3 (for compounds reduced) and Fig. 4. (for compounds amplified). All relevant non-parametric statistics are presented in S. Table 1. Note: FC of 1 = 0 change, FC of 1.50 = 50% amplification or reduction, FC of 2.5 = 150% amplification/reduction etc.Fig. 3. Intensity (au) and fold change (FC) values for the 12 reduced phenylpropanoid and quinic acid derived compounds in Calluna vulgaris (heather) plants when exposed to 95% UV attenuation as compared to those exposed to 20% attenuation. ***p < 0.001, Mann-Whitney U-test. X = mean, ― = median. A FC of 1 = 0 change, FC of 1.50 would = 50% reduction etc. All metabolites in the top row plus chlorogenic acid and quinic acid in the second row have been recorded in the literature, to respond to variation in ultra-violet lightFig. 4Intensity (au) and fold change (FC) values for the 4 amplified phenylpropanoid and quinic acid derived compounds in Calluna vulgaris (heather) plants when exposed to 95% UV attenuation as compared to those exposed to 20% attenuation. ***p < 0.001, Mann-Whitney U-test. X = mean, ― = median

The reviewed literature (annotated in Table 1) revealed that caffeic acid, rosmarinic acid, sinapoyl malate, 3,4’,5-Trihydroxystilbene-3-beta-D-glucoside, chlorogenic acid and quinic acid have all been recorded as being influenced by UV in other plant species. Similarly, caffeic acid, p-coumaric acid, coumarin and chlorogenic acid, were identified in other studies as having anti-herbivore activity (Table 1).

Larval Survival and Prepupal Weight are not Affected when Feeding on Foliage from Different Treatments

Mean prepupal weights and larval survival for L. suturalis feeding on foliage exposed to 20% or 95% attenuation of UV (n = 18 for both treatments) were very similar and revealed no significant difference between treatments (Fig. 5). Prepupal weights were 153.6 ± 5.75 mg for 20% attn. and 155.7 ± 11.1 mg for 95% attn., Mann-Whitney, U = 145.00, p = 0.580; larval survival (out of 10) was 8.8 ± 1.1 for 20% attn. and 8.5 ± 1.5 for 95% attn., U = 148.00, p = 0.642. Overall, the mortality rates were 12.2% and 15% for the 20% and 95% attenuation treatments respectively.

Fig. 5. Mean prepupal weight (mg) (a) and larval survival (b) of heather beetle (Lochmaea suturalis) feeding on 20% and 95% UV attenuated heather (Calluna vulgaris) foliage. No significant difference was observed between weights, p > 0.05 or survival, p > 0.05. Error bars = 1 SE

Discussion

Plant Metabolite Response to UV-Radiation

Our results show that changes in UV-radiation influence the shikimate-phenylpropanoid secondary metabolism of C. vulgaris, primarily through the phenylpropanoid and flavonoid biosynthetic pathways both of which are dependent on p-coumaric acid as a key precursor but have displayed different responses. Under 95% attenuation of UV, hydroxycinnamic acids, coumarins, a stilbene glycoside and a number of caffeoylquinic acids, all catalysed by p-coumaroyl-CoA, shikimic acid and quinic acids via the phenylpropanoid biosynthesis pathway (Alcázar Magaña et al. 2021) are reduced. Conversely malonyl-CoA plus p-coumaroyl-CoA catalyse the biosynthesis of naringenin in the flavonoid pathway (Singh et al. 2021) leading to, under 95% attenuation, myricetin, myricetin-3-xyloside and kaempferol-3-glucoside-2’’-p-coumaroyl remaining amplified (Table 1). This suggests these particular flavonoids are possibly upregulated due to another stressor such as soil nutrient availability, thus maintaining their amplified state. The soils in which these trials were run are extremely low in both nitrogen (N) and phosphate (P) and there is evidence that many flavonoids are amplified and function as antioxidants involved in reducing reactive oxygen species induced by stress including low soil nutrients (Lillo et al. 2008; Patil et al. 2024; Stewart et al. 2001). Support for this outcome is also provided from our own field trials at a CP site where the compound kaempferol-3-glucoside-2’’-p-coumaroyl and other flavonoid glycosides were significantly reduced in C. vulgaris foliage after the application of N and P fertilizers (Barrett et al. In preparation).

The literature provides several reports (Table 1), showing responses of various plant species to UV (particularly the UV-B spectrum) which support our findings. For hydroxycinnamic acids, work with Deschampsia antarctica (É.Desv) using screen technology to manipulate UV-B showed decreased caffeic and p-coumaric acid concentrations when exposed to reduced levels of UV-B. This is linked to the photoprotective function of these metabolites and their trade off with the growth parameters of total leaf area and leaf mass ratio in this plant (Ruhland et al. 2005). Rosmarinic acid, sinapoyl malate and the stilbene glycoside, resveratrol (3,4’,5-Trihydroxystilbene-3-beta-D-glucoside), are all understood to be UV absorbing metabolites and respond to UV by accumulating in the upper epidermis of leaves which reduces the transmittance of damaging UV to the mesophyll (Baker et al. 2020; Luis et al. 2007; Zhang and Björn 2009). Quinic acid derived compounds, chlorogenic acid and 3-feruloylquinic acid are also documented as responding to and providing photo protection to Betula pendula (Roth), (birch) and Hordeum vulgare (L.) (barley) respectively (Lavola et al. 1997; Holub et al. 2019).

From our earlier studies (Barrett et al. 2024), chlorogenic acid, caffeic acid and 4-hydroxycoumarin are compounds that had significantly amplified in field-collected C. vulgaris samples from New Zealand CP sites where peak summertime UV index is ∼ 14, compared to those samples from Scotland (SC) in the UK with an index of ∼ 6–7. Fold change levels of these three compounds were higher in the CP samples at 1.13, 1.22 and 1.2-fold for chlorogenic acid, caffeic acid and 4-Hydroxycoumarin, respectively and we extrapolate that these compounds are elevated when exposed to high levels of UV in New Zealand’s CP environment (Barrett et al. 2024). Additionally, earlier work by Effah et al. (2020b) also indicated the sensitivity of C. vulgaris to UV light, when exposure to different intensities of UV under the same attenuating screens induced changes in a range of sesquiterpene volatile metabolites. These results suggest that amplification of many phenylpropanoid compounds in C. vulgaris on the CP are likely being driven by elevated UV light.

Our findings align with and provide further evidence that secondary metabolites can be significantly altered in invasive plants that have established in a new environment (Wolf et al. 2011; Skubel et al. 2020; Medina van Berkum et al. 2023). Changes in the light environment are known to alter plant physiological processes and the resulting biochemical phenotypes, which consequently can affect insect herbivore performance (Rousseaux et al. 2001; Kuhlmann and Müller 2010; Ballaré et al. 2012; Dinh et al. 2013; Ballaré 2014; Qi et al. 2018; Fallah et al. 2023). Where light parameters differ significantly between sites, regions or ranges, we suggest investigation of potential change to the biochemical phenotype of a host plant may deserve exploration.

Insect Responses to UV-Related Metabolite Changes

We found no significant difference on the assessed life history parameters of the biocontrol agent L. suturalis after feeding on plants from each of the UV attenuation treatments. To validate this result, we also compared live weights from pre-pupae collected from a field population on the CP close to where the experimental heather plants were sourced originally. There was no significant difference (Kruskal-Wallis H(2) = 0.77, p = 0.680), in mean live weights of pre-pupae from that field population and the mean weights of pre-pupae obtained from either of the UV attenuated trials, confirming the lack of effect of the UV treated plants (see also S Fig. 3).

Three compounds recorded in this study that are known from the literature to be both altered by UV and involved in anti-herbivore defence are caffeic acid, chlorogenic acid and p-coumaric acid (Dixit et al. 2017; Singh et al. 2021). The literature provides little evidence that these three compounds impair performance in Coleopteran spp. but chlorogenic acid is a feeding deterrent for Spodoptera exigua (Hübner, 1808) feeding on honeysuckle Lonicera maackii (Rupr.) Maxim. Caffeic and chlorogenic acids are reported to inhibit larval growth and development of the caterpillar Helicoverpa zea (Boddie, 1850) when feeding on cotton Gossypium hirsutum (L.) and caterpillar growth of Sesamia nanogriodes (Lefebvre, 1827) is impaired by p-coumaric acid when feeding on corn Zea mays (L.). However, unlike the specialist L. suturalis, these species are broad generalist (S. exigua and H. zea) or oligophagous (S. nanogriodes).

In a review of the specialist - generalist paradigm, Ali and Agrawal (2012) point out that while specialists are much better adapted to tolerate plant defences than generalists, at higher concentrations of many defensive compounds, few specialists are entirely immune to them. The specialist lepidopteran tobacco hornworm Manduca sexta (Linnaeus, 1763) showed no significant difference in pupal mass after feeding on a control diet containing no nicotine and one with 0.1% wet weight nicotine, but a 5-fold increase (0.5%) in nicotine resulted in a significant reduction of pupal mass (Harvey et al. 2007). They also postulate that for FC increases at low to moderate levels, specialists can benefit from such compounds but at higher FC levels they will suffer deleterious effects. As an example, Richards et al. (2012) using the specialist buckeye caterpillar Junonia coenia (Hübner, 1822) demonstrated that diets with low concentrations of iridoid glycosides (ranging from 1 to 6-fold differences), caterpillar survival progressively increased but at concentrations equivalent to 8 to 10-fold, survival decreased significantly. It’s therefore possible that the magnitude of change in our UV trials may have been insufficient to impact growth and survival in L. suturalis larvae.

Assuming any or all of these compounds i.e. caffeic acid, chlorogenic acid or p-coumaric acid may indeed impair L. suturalis larvae at higher concentrations, this raises the question of the magnitude of change induced by the 95% UV treatment. In our trials, the FC difference for the above three compounds were relatively low, i.e. 1.73, 1.54 and 2.45-fold, respectively. A FC reduction of these magnitudes would potentially provide an impetus to growth in a generalist insect herbivore but may be of little consequence to a specialist that has evolved the ability to tolerate variation in the concentration of defensive compounds and even use them as feeding stimulants, detoxify them pre- or post-ingestion, or sequester them as defences against predators (Ali and Agrawal 2012; Jeckel et al. 2022; Kshatriya and Gershenzon 2024). Thus, the relatively small FC reduction in compounds of the 95% attn. treatment, appears to have little influence on the assimilability of C. vulgaris foliage for L. suturalis.

We acknowledge that the foliage presented to the larvae were cut stems. Severed stems can produce wound responses, including amplification of jasmonic and abscisic acids post severance (Da Costa et al. 2013), but may not be producing an appropriate herbivore damage response. Thus, the larvae may be exposed to constitutive compounds only, and not the full suite of induced defensive compounds and their potential synergistic effects (Barbehenn and Kochmanski 2013) as could be expected from a living plant. It’s also possible that the mature heather plants collected from the CP field site, where they have had continuous exposure and are well acclimated to high natural levels of UV, had limited potential to respond to reduced UV during the attenuation period. Thus, exposing plants to controlled UV-intensities from seedling emergence and for longer periods may induce greater magnitudes of metabolite change and if used as live plants, provide more definitive results when testing herbivore responses.

Could Elevated Phenylpropanoids be Impairing L. suturalis on the CP?

Using data from our earlier work, we established the fold change differences for thirty-five phenylpropanoid derived compounds in C. vulgaris between the CP (NZ) and Scotland (UK) (Barrett et al. 2024). Several flavonoid, coumarin and hydroxycinnamic acid compounds i.e. myricetin, kaempferol-3-glucoside-2’’-p-coumaroyl, 4-Hydroxycoumarin, caffeic acid, o-coumaric acid (an isomer of p-coumaric acid) and chlorogenic acid, co-occur in both NZ vs. UK data (Barrett et al. 2024) and in the attenuated UV trial results presented here. The fold changes for these compounds between NZ and UK ranged from 1.13 to 1.82 and for this UV trial were 1.54 to 2.45. These ranges are within those given above in Harvey et al. (2007) and Richards et al. (2012).

L. suturalis as a specialist herbivore is ubiquitous wherever C. vulgaris occurs in the UK and Europe and is a species with regular population outbreaks, causing complete destruction of all plants within the affected site over a wide range of habitats and degrees of latitude (Pakeman et al. 2002; Rosenburgh and Marrs 2010). Again, from our earlier work, we established for fifteen phenylpropanoid derived compounds, the fold change range of variation between four geographically and geologically different sites in a part of C. vulgaris’ native range in Scotland, UK. These are all well-known flavonoids, coumarins and hydroxycinnamic acids with magnitude of inter site variation ranging from 1.57 to 7.29-fold. It seems that L. suturalis in its native range, is well adapted to a greater magnitude of phenylpropanoid variance, that it is not encountering on the CP or from our UV attenuation trial.We postulate then, that at the magnitude of difference measured in our earlier work and this trial, there will be little effect of amplified phenylpropanoids on larval performance in the CP L. suturalis population. It may be, that the higher levels of phenylpropanoids recorded in CP C. vulgaris in Barrett et al. (2024), provide the plant with enhanced photoprotection, but don’t necessarily enhance its anti-herbivore defensive capability against this specialist herbivore.

Nearly three decades of research have been accomplished since the release of L. suturalis against the invasive shrub C. vulgaris on the CP in New Zealand, which now provides a broad understanding of this biocontrol programme (Barrett et al. 2021). Publications include the effect of parasitoids and predators of L. suturalis eggs and larvae (Peterson et al. 2004), adult beetle body size and winter survival (Fowler et al. 2015) and comparing the efficacy of biocontrol vs. herbicides (Peterson et al. 2020). The application of fertiliser to improve soil nutrient status and increase host plant foliar nitrogen with a consequent improvement in L. suturalis larval and adult performance parameters and establishment rates is the most recent (Peterson et al. 2024). That work links poor L suturalis performance and establishment with low foliar nitrogen of C. vulgaris foliage on the CP. We are now using metabolomics to provide insight into how increases in foliar nitrogen of C. vulgaris interacts with both primary nitrogen containing and secondary metabolites, to promote the demonstrated increase in L. suturalis performance and establishment.

The work covered in this report using metabolomic techniques, attempts to address an unanswered but fundamental question relating to plant biochemical defenses and their effects on this biocontrol agents’ larval stages in the CP environment. We acknowledge there are other life stages that could potentially be affected by plant defences such as adult fecundity and overwintering fat body deposition. L. suturalis adults being univoltine however, generally don’t co-occur with larvae on C. vulgaris during late spring and summer. Adults predominate during early spring leading up to oviposition and in autumn through flowering as next generation pre-overwintering adults, thus different phenological stages of C. vulgaris are required to test relevant adult life history parameters.

In Conclusion

Our trials using attenuating UV screen technology significantly altered the intensity of several phenylpropanoid metabolites in heather (C. vulgaris) with several metabolites being in common with similar trials reported for a range of plant species in the literature. Heather exposed to 95% attenuation of UV showed reduced intensities of metabolites. The corollary of this is that elevated UV (20% attenuation or ambient in natural field conditions) amplifies many phenylpropanoids which indicates an upregulation of the shikimate – phenylpropanoid pathway. This is what was expected and is further supported by our results. Overall, we accept our hypothesis that exposure to artificially reduced ultra-violet light reduces phenylpropanoid metabolites of heather.

Results from the bioassays to determine if UV induced changes to the levels of defensive secondary metabolites result in differences in mean prepupal weight and larval survival of heather beetle L. suturalis larvae showed no significant effects. Given the adaptations of specialist insects to tolerate plant defences, and that there is considerable variability in the range of concentrations of metabolites both between plants and between sites, in both the native and invaded ranges, the magnitude of change to C. vulgaris plants exposed to the two UV attenuating treatments appears to be insufficient to have any effect on these life history parameters. Notwithstanding that, we consider the use of metabolomics to have great potential to elucidate insect performance parameters as a function of plant metabolite variation and to provide valuable insights into the biochemistry of plant-biocontrol agent interactions.

Where light or other abiotic parameters such as soil nutrients differ significantly between regions or ranges, we suggest investigation of potential change to the biochemical phenotype of a target host plant may deserve exploration. Such metabolomic assessments might be included early in a programme where biological control of a plant using insect herbivores is being considered or indeed retrospectively where a programme has proven to be ineffective. For investigation of UV induced phytochemical changes we encourage exposing plants to the experimental UV-intensities from seedling emergence and for extended periods of time to promote maximum potential for change and to use live plants when testing herbivore performance responses to those changes.

Finally, there are a multitude of questions which could be addressed using metabolomics (or more correctly, ecometabolomics) to assist with understanding fundamental ecological processes relating to biochemical phenotypic change in invasive plants. The application of metabolomics to address biochemical defensive effects on plant-insect interactions relating to biocontrol of weeds is in its infancy. We hope, however, the investigations reported here will prompt further use of this exciting technology in future weed biocontrol programmes and/or retrospectively to assess programmes that failed to meet expectations.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1 (DOCX 15.7 MB)

Supplementary Material 2 (DOCX 222 KB)