Symbiosis plays a critical role in plant biology. Temperate grasses often associate with several symbiotic fungi simultaneously, including Epichloë endophytes and arbuscular mycorrhizal (AM) fungi, in shoots and roots, respectively. These symbionts often modulate plant-herbivore interactions by influencing nutritional traits (i.e., AM fungi-mediated nutrient uptake) and/or the secondary chemistry (i.e., endophytic alkaloids) of their host plant. Moreover, such grasses also accumulate large amounts of silicon (Si) from the soil, which can be deposited in tissues to act as a physical antiherbivore defense.

Recent evidence suggests that both endophytes and AM fungi independently facilitate Si uptake. However, the consequences of their interactions for piercing-sucking insects (i.e., aphids), or whether Si supply, endophytes and AM fungi interact in this regard, are currently unknown. While Si deposition may be less effective against aphids than other herbivores (i.e., chewing caterpillars), Si supply can also alter plant secondary metabolite defenses, which could affect sucking insects.

In a factorial greenhouse experiment we evaluated whether these components, acting alone or in combination, altered 1) foliar primary chemistry, 2) Si and symbiont-chemical (endophytic alkaloids) defenses, as well as 3) performance of the bird-cherry oat aphid (Rhopalosiphum padi) feeding on tall fescue (Festuca arundinacea).

Endophytes decreased all aphid performance parameters, including population growth and repoduction by 40%, but their impact was reversed by the presence of AM fungi, leading to a 52% increase in aphid performance compared to plants solely hosting endophytes. This improvement in performance was associated to reduced loline alkaloid levels and higher shoot nitrogen in AM-endophytic plants. Endophytes and AM fungi exhibited antagonism, with endophytes reducing AM colonization by 34% and AM presence decreasing endophyte loline alkaloids by 44%. While both fungi jointly increased Si accumulation by 39% under Si-supplied conditions, Si had no noticeable effects on aphids. Moreover, although Si supply had no identifiable effects on AM colonisation, it reduced endophyte peramine alkaloids by 24%.

Synthesis: Our findings indicate that symbiotic fungal partnerships and silicon provision may benefit plants but could weaken anti-herbivore defenses when combined. Revealing the complex interactions among diverse fungal symbionts and showcasing their effects on different anti-herbivore defenses (chemical and physical) and herbivore performance for the first time.

Plants, Insects and Experimental procedure

Two hundred tall fescue plants (Festuca arundinacea cv. INIA Fortuna) either Epichloë-free (Nil; N = 100) or infected with the animal safe AR584 novel Epichloë coenophiala (formerly Neotyphodium coenophiala) strain (N = 100) were grown individually from seed in one-litre pots. Initially, these seeds were inoculated with the AR584 endophyte strain by AgResearch NZ, in collaboration with INIA under a work cooperation agreement. Following the endophyte incorporations, two distinct seed lines were maintained at the Margot Forde Forage Germplasm Centre (Palmerston North, NZ). Pots contained gamma-irradiated (50 kGy) 1:1 topsoil-sand mix that was homogenised with a soil mixer. Sand was used to reduce bioavailable phosphorus to <16 mg P kg^-1 and silicon to < 11 mg Si kg^-1 (Table S1 Appendix S1).

The top ¼ of each pot received one of two AM fungi treatments: No AM (-AM) or Commercial AM (+AM). +AM was achieved by inoculating soil with Start-up Ultra© (Microbe Smart Plty. Ltd., South Australia) which contains spores from four isolates of Rhizophagus irregularis (formerly Glomus intraradices). Before inoculation, spores were extracted and separated from the inert substrate (calcined diatomaceous earth) using wet sieving and applied at a rate equivalent to the recommended rate of 250 g of inoculum per 2.5 L with hydrating water. To generate -AM treatment the same steps were performed, except that the inoculant (extracted spores) was sterilised by autoclaving twice (121°C) to ensure spores were non-viable. To standardise the microbial community within each pot, the soil-sand mix in all treatments received a microbial filtrate (300 ml/10 L) a week after irradiation. This filtrate was created by using the extraneous extraction solution (without spores) from both the commercial AM fungi after wet sieving and the soil before irradiation following the procedure described in Frew et al., (2018). Only the top ¼ of soil in each pot was inoculated with the -AM or the +AM solutions.

Plants were grown in a single, naturally lit glasshouse chamber at 22/18°C (day/night) and 60% relative humidity at the Hawksbury Institute for the Environment in Richmond, NSW, Australia. Pots were randomly shifted weekly to avoid position bias, and manually irrigated three times a week with 50 ml of either a solution with (+Si) or without (-Si) silicon (Si). Briefly, the +Si solution contained potassium silicate (K₂SiO₃; Agsil32, PQ Australia) at a concentration of 2 mM, whereas the -Si solution contained KCl to balance the addition of potassium in the +Si treatment. Further, using HCl, both solutions were adjusted to pH 7 following procedures described in Hall et al., (2020). This resulted in a total of eight treatments that consisted of combinations of two different endophyte treatments (Nil or AR584), two AM fungi treatments (-AM or +AM), and two Si supply treatments (-Si or +Si). Each treatment combination contained 25 pots/replicates (200 plants in total; Fig. 1 Symbiont treatments).

After eight weeks, 10 plants per treatment combination were inoculated and caged with five adult (apterous) bird cherry-oat aphid Rhopalosiphum padi (Linnaeus, 1758). Transparent cylindrical Perspex cages with meshed air vents were fitted to the pots of all plants similar as to those used by figure 1S of Cibils-Stewart et al (2021). Cultures of R. padi were established from a single parthenogenetic female obtained from a laboratory culture at Agriculture Victoria Research (Horsham, VIC, Australia) and reared on caged barley (Hordeum vulgare cultivar ‘Hindmarsh’). Aphids reared on barley before the experiment prevent prior exposure to the plant species, standardize the setup, control the aphid population, and reduce variability in behaviour and responses. On a weekly basis, ten teneral adult females were transferred to new caged barley for 24 h. Adults were then removed, leaving only same-age nymphs on plants. This procedure ensured same-aged aphids for experiment initiation.

Aphid-inoculated plants were compared with 10 caged aphid-free plants, selected at random, allowing us to determine the impact of aphid infestation on plant parameters. Using these plants, treatment effects on aphid population performance parameters were determined (section 2.2.1; Fig. 1 Impact on aphid population). The remaining five plants per treatment (Table S2) were inoculated with three apterous adults R. padi each (120 aphids total). Each individual aphid was confined to a clip cage that followed Cibils-Stewart et al (2015) design, featuring a 0.5-cm thick foam rectangle (6.2 x 3.6 cm outside, 5.1 x 2.5 cm inside) with adhesive on tops and bottoms. Pre-applied adhesive secured no-see-um mesh on one side for leaf cages, preventing aphid escape while ensuring ventilation for aphids. The remaining adhesive was attached to the leaf surface. Following Rowe et al., (2020) procedures, each cage was placed on the youngest leaf of three different tillers per plant. Using these plants, treatment effects on individual aphid performance parameters were determined (section 2.2.2) (see Table S2 for treatment combinations; Fig. 1 Impact on individual aphid performance).

To assess the effect of the symbionts and Si supply (alone and in combination) on plant physiological performance, net photosynthesis (Anet), stomatal conductance (gs), and water-use efficiency (WUE) were measured in aphid-free plants with an infrared gas analyser (IRGA, LI6400XT, Li-Cor, Lincoln, NE, USA) following Vandegeer et al., (2020) procedures. Additionally, using a Minolta chlorophyll SPAD 502 meter chlorophyll content (i.e. greenness) was measured in the same leaves, and utilized as a proxy of plant vigour (Druille et al., 2013).

Immediately after harvest the symbiotic status of plants was detected (section 2.3); for this a 1 g section of the root was preserved in ethanol to determine AM status (section 2.3.2). Roots were carefully washed, and a 1 g (wet weight) subsample was taken from the same area from all replicates per treatment combination. Subsamples were stored in tissue embedding cassettes in 70% ethanol. Remaining shoots and roots were snap frozen in liquid nitrogen, freeze-dried, weighed (MS-TA Analytical balances; Mettler Toledo), and ball-milled to fine powder (Mixer Mills MM 400; Retsch). Samples were stored at -20°C until further chemical analysis (section 2.4). The removed root sample was not accounted for in the overall root weight (Fig. 1).

Aphid parameters

Aphid population

Three weeks after aphid inoculation, aphid populations were categorized (apterous, adults and nymphs) and counted. The finite rate of population change was then calculated for the entire 21-day trial (λ₂₁) as a ratio of change in aphid densities at the start (N₀) and end (N₂₁) of assay, where λ₂₁ = N₂₁/N₀.

Individual aphid performance

After 24 h, the original aphid and all but the youngest nymph (founder nymphs) was removed from cage. Nymphal development (Pr; pre-reproductive period), adult fecundity (F; number of born nymphs daily), and longevity (L) of the founder nymphs were recorded daily for the entire lifespan of each aphid. Newly born nymphs were removed daily from clip cages after daily adult fecundity recordings. Using the above parameters, the intrinsic rate of increase (r_m) and the generation time (GT) were calculated as r_m = 0.74 (ln F)/L) and GT=4Pr/3, respectively (Wyatt & White, 1977).

Symbiont detection

Endophyte in shoots

The fresh cut end of one tiller of each plant was pressed onto a nitrocellulose membrane (tissue-print immunoblotting) to confirm absence (Nil) or presence (AR584) of Epichloë in planta following di Menna, et al., (2012) procedures; this was further confirmed with histological staining following Cibils-Stewart et al., (2020) procedures. No colonization was detected in the endophyte-free (Nil) plants.

AM fungi in roots

Mycorrhizal colonization was scored  following Frew et al (2018) procedures with minor modifications. Briefly, samples within cassettes were rinsed with cold water and cleared with 10% KOH at room temperature for five consecutive days. Samples were then water-rinsed and stained with 5% ink-vinegar in a 60°C water bath for 10 mins (Vierheilig et al., 2005). Finally, samples were rinsed until water ran clear and were submerged in lacto glycerol (de-staining solution) overnight. Ten 1 cm segments per sample were mounted on glass slides with glycerine under a cover slip and a minimum of 25 intersects were scored for presence of AM fungi using the intersect method (McGonigle et al., 1990). Only hyphae for which there was a visible connection to AM fungal structures (arbuscules, vesicles, spores) were counted. No colonization was detected in the -AM plants.

Chemical analyses

Shoot silicon, carbon and nitrogen

Eighty milligrams of ground leaf material were analysed to measure Si concentration using an X-ray fluorescence spectrometer (Epsilon 3 ×; PANalytical, EA Almelo, The Netherlands) following the method of Reidinger et al., (2012) and using certified plant material of known Si concentrations (see Hiltpold et al., 2016 for full details). A subsample of 6-7 mg of ground tissue of aphid-free plants used for Si analysis was further utilized to measure carbon and nitrogen concentrations with an elemental analyser (FLASH EA 1112 Series CHN analyser, ThermoFinnigan, Waltham, MA, USA).

Epichloë alkaloid concentrations

Shoot peramine production was analysed according to Berry et al., (2019) using a Thermo LTQxl linear ion-trap mass spectrometer equipped with an Accela 1250 HPLC in endophytic-plants only. Chromatography was achieved using a SeQuant ZIC-HILIC column (150 × 2.1mm, 5 µ, Merck KGaA, Darmstadt, Germany). Samples were extracted with 500 µL of 50% methanol (containing 1.7 µg/mL homoperamine as an internal standard) for 1 h in the dark. Samples (20 mg) were then centrifuged (5000 g, 5 min) and the supernatant transferred to 2 mL amber HPLC vials via a 0.45 µm syringe filter (PVDF). Peak integration was conducted using LCQuan 2.7 (Thermo Fisher Scientific Inc., San Jose, CA, USA); alkaloid concentrations were determined from peak areas and calculated standard curves. Production of the three loline derivatives N-acetylloline (NAL), N-acetylnorloline (NANL), and N-formylloline (NFL) (added to afford the total lolines) weas quantified following Bastías et al. (2018).

 Statistical analysis

R was utilized for all statistical analyses (R Core Team, 2015) and all figures were produced using the ‘ggplot2’ package (Wickham, 2016). Assumptions of normality for residuals for all models were verified according to the inspection of quantile-quantile plots, and variables log-transformed in case of lack of normality (aphid λ).

Shoot and root biomass of freeze-dried tissue, along with foliar Si concentration were analysed with a four-way analysis of variance (ANOVA) using Si supply, AM fungi, endophyte and herbivory as fixed effects. AM fungi colonisation, alternatively, was analysed with a three-way ANOVA using Si supply, endophyte, and herbivory as fixed effects; only +AM plants were utilized for this analysis.

Aphid parameters were analysed using a three-way ANOVA, with Si supply, AM fungi and endophyte as fixed effects using aphid-inoculated plants. Likewise, plant physiological and nutritional traits were analysed using a three-way ANOVA, with Si supply, AM fungi and endophyte as fixed effects, aphid-free plants only. For all models described above, differences between treatment means were determined by Tukey’s HSD test using the ‘emmeans’ package (Lenth et al., 2018). A multivariate analysis of variance (manova) using Si supply, AM fungi and herbivory as fixed effect factors was utilized to determine differences in overall alkaloid profiles (sum of all alkaloids produced by the AR584-strain) and to assess treatment effects on individual alkaloids at the strain-specific level. Finally linear regressions were used to determine the relationship between aphid parameters and specific alkaloid expressions.