Effects of leaf herbivory and autumn seasonality on plant secondary metabolites: A meta-analysis

Abstract

Plant secondary metabolites (PSMs) are produced by plants to overcome environmental challenges, both biotic and abiotic. We were interested in characterizing how autumn seasonality in temperate and subtropical climates affects overall PSM production in comparison to herbivory. Herbivory is commonly measured between spring to summer when plants have high resource availability and prioritize growth and reproduction. However, autumn seasonality also challenges plants as they cope with limited resources and prepare survival for winter. This suggests a potential gap in our understanding of how herbivory affects PSM production in autumn compared to spring/summer. Using meta-analysis, we recorded overall production of 22 different PSM subgroups from 58 published papers to calculate effect sizes from herbivory studies (absence to presence) and temperate to subtropical seasonal studies (summer to autumn), while considering other variables (e.g., plant type, increase in time since herbivory, temperature, and precipitation). We also compared production of five phenolic PSM subgroups - hydroxybenzoic acids, flavan-3-ols, flavonols, hydrolysable tannins, and condensed tannins. We wanted to detect a shared response across all PSMs and found that herbivory increased overall PSM production in herbaceous plants. Herbivory was also found to have a positive effect on individual PSM subgroups, such as flavonol production, while autumn seasonality was found to have a positive effect on flavan-3-ol and condensed tannin production. We discuss how these responses might stem from plants producing some PSMs constitutively, whereas others are induced only after herbivory, and how plants produce metabolites with higher costs only during seasons when other resources for growth and reproduction are less available, while other phenolic PSM subgroups serve more than one function for plants and such functions can be season dependent. The outcome of our meta-analysis is that autumn seasonality changes some PSM production differently from herbivory, and we see value in further investigating seasonality-herbivory interactions with plant chemical defense.

Keywords: autumn seasonality; flavonoids; herbivory; phenolics; plant chemical defense; plant secondary metabolites.

FIGURE 1

Meta‐regression results for plant secondary metabolism differences between plant type (a – herbaceous, b – woody) and study type (autumn seasonality vs. herbivory). Narrow bars denote prediction intervals, while thick bars denote confidence intervals (95%). k is the number of effect sizes included in the analysis. Circle size indicates weight in analysis (inverse of standard error). Whiskers denote 95% confidence intervals; estimates are statistically significant if the confidence intervals do not cross the dashed vertical line. Statistical results are shown in Table S9.

FIGURE 2

Meta‐regression results for plant secondary metabolism differences between hours after herbivory. Narrow bars denote prediction intervals, while thick bars denote confidence intervals (95%). k is the number of effect sizes included in the analysis. Circle size indicates weight in analysis (inverse of standard error). Whiskers denote 95% confidence intervals; estimates are statistically significant if the confidence intervals do not cross the dashed vertical line. Statistical results are shown in Table S10.

FIGURE 3

Meta‐regression results for individual phenolic plant secondary metabolism subgroup differences between herbivory and early autumn seasonality measured in papers. Letters correspond to the following: (a) hydroxybenzoic acids, (b) flavan‐3‐ols, (c) flavonols, (d) hydrolysable tannins, (e) condensed tannins. Narrow bars denote prediction intervals, while thick bars denote confidence intervals (95%). k is the number of effect sizes included in the analysis. Circle size indicates weight in analysis (inverse of standard error). Whiskers denote 95% confidence intervals; estimates are statistically significant if the confidence intervals do not cross the dashed vertical line. Statistical results are shown in Table S11.

FIGURE 4

Meta‐analytic results for plant secondary metabolism differences with changes in precipitation. Circles indicate effect sizes of five largest metabolic subgroups in dataset and circle size indicates weight in analysis (inverse of standard error). Dashed lines show marginally significant intercept and slope estimates for flavan‐3‐ol and hydroxybenzoic acid meta‐regression models. Black line indicates the slope for all 22 metabolic subgroups found in dataset (Table S2). Statistical results are shown in Table S12.

FIGURE 5

Meta‐analytic results for plant secondary metabolism differences with changes in temperature. Circles indicate effect sizes of five largest metabolic subgroups in dataset and circle size indicates weight in analysis (inverse of standard error). Red line shows significant intercept and slope estimates for flavan‐3‐ol meta‐regression models. Black line indicates the slope for all 22 metabolic subgroups found in dataset (Table S2), while red slope indicates flavan‐3‐ol production, which decreased as temperature lowered from summer to fall (p = .016). Statistical results are shown in Table S12.