Plants deploy diverse anti-herbivore defences which reduce feeding and performance of herbivores. Temperate grasses use silicon (Si) accumulation and Epichloë-endophytes for physical and chemical (i.e. endophytic-alkaloids) defence against insect herbivores. Recent studies suggest that Epichloë-endophytes increase Si accumulation in their host grass. It is unknown, however, how this affects Si-deposition on the leaf surface, their impacts on insect herbivore feeding efficiency and their immunity to potential infection/parasitism.

To address this knowledge gap, we grew tall fescue (Festuca arundinacea) hydroponically with and without Si, in the absence or presence of the novel AR584 Epichloë-strain. We exposed plants to Helicoverpa armigera (Lepidoptera: Noctuidae) in both in-situ (intact leaves) and ex-situ (excised leaves) feeding trials and determined the effects of Si and endophyte defences on herbivore feeding efficiency, growth rates and immunity against potential infection/parasitism.

Endophytic plants supplied with Si showed 110% and 143% increases in leaf silica density and leaf Si concentrations, respectively, when exposed to herbivory, compared to non-endophytic plants that were herbivore-free. Despite the endophyte-mediated increases in Si concentrations, H. armigera was only affected by Si supply; growth rates decreased by 87% and most feeding efficiency indices decreased by at least 30%. Si supply also increased mandibular wear by 16%, which was negatively correlated with H. armigera growth rates. Cellular and humoral immunity of H. armigera were negatively affected by both Si and endophytes. Endophytic-loline alkaloid concentrations were unaffected by Si supply or herbivory, whereas herbivory increased peramine concentrations by 290%. 

To our knowledge, this is the first report of Si defences and Epichloë-endophyte-derived alkaloids compromising insect immunity via reduced melanisation response. Using tall fescue and H. armigera, our study suggests that deploying both physical (i.e. Si accumulation) and chemical (i.e. endophytic-alkaloids) defences acting against multiple insect herbivore traits, including feeding efficiency, growth and immunity, may be a successful defence strategy in temperate grasses. This multi-faceted defence may be particularly difficult for insect herbivores to overcome.

The statistical software R (R Core Team, 2015) was utilized for all statistical analyses. Assumptions of normality for residuals were verified for all models according to the inspection of quantile-quantile plots. All figures were produced using the ‘ggplot2’ package (Wickham, 2016).

To address hypothesis one (H1) that states that increased Si-defences (e.g. increase silicified phytoliths in the leaf surface) have no impact on endophyte-specific alkaloid production, foliar Si concentration and resulting silica cell density (response variables) of +Si plants only were analysed using a two-way ANOVA (type III sum-of-squares) with endophyte and herbivory treatments as fixed effect factors. To address H2 states that Si supply has no impact on endophyte-specific alkaloid production, using only endophytic plants, a multivariate analysis of variance (permutational MANOVA) with Si and herbivory treatments as fixed effects was utilized to determine differences in overall alkaloid profiles (response variables) with the ‘vegan’ package (Oksanen et al., 2019). The MANOVA function further provided results for univariate ANOVAs for treatment effects on individual alkaloids.

Moreover, to address H3 and H4 that states that only Si supply negatively affects feeding efficiency indices (H3), whereas both Si and endophyte have (H4) negative effects on insect immunity, all herbivore parameters including, relative growth rate, mandibular wear, food utilization indices, and larval cellular and humoral immunity (response variables), were explored using a two-way ANOVA (type III) with endophyte and Si treatment as fixed factors. Additionally, a Pearson rank correlation coefficient test was used to analyse the relationship between relative growth rate and mandibular wear (Knowles, Siegmund, & Zhang, 1991). Finally, although not part of the hypothesis, for plant growth and primary chemistry, a three-way ANOVA (type III) with endophyte, Si supply and herbivory treatments as fixed effect factors was performed.

For the data analysed, we employed Type III ANOVAs and reported all interaction terms in the tables; we did not conduct model simplification analyses to assess the impact of non-significant interaction terms on the overall model because removing non-significant interactions can lead to an increase in the apparent sample size and may affect the power to detect main effects. Note that some individuals escaped our in-situ trial and thus, the data for this is unbalanced.