The Price of the Induced Defense Against Pests: A Meta-Analysis

Abstract

Plants and phytophagous arthropods have co-evolved for millions of years. During this long coexistence, plants have developed defense mechanisms including constitutive and inducible defenses. In an effort to survive upon herbivore attack, plants suffer a resource reallocation to facilitate the prioritization of defense toward growth. These rearrangements usually end up with a penalty in plant growth, development or reproduction directly linked to crop losses. Achieving the balance to maximize crop yield requires a fine tune regulation specific for each host-arthropod combination, which remains to be fully elucidated. The purpose of this work is to evaluate the effects of induced plant defenses produced upon pest feeding on plant fitness and surrogate parameters. The majority of the studies are focused on specific plant-pest interactions based on artificial herbivory damage or simulated defoliation on specific plant hosts. In this meta-analysis, the relevance of the variables mediating plant-pest interactions has been studied. The importance of plant and pest species, the infestation conditions (plant age, length/magnitude of infestation) and the parameters measured to estimate fitness (carbohydrate content, growth, photosynthesis and reproduction) in the final cost have been analyzed through a meta-analysis of 209 effects sizes from 46 different studies. Herbivore infestation reduced growth, photosynthesis and reproduction but not carbohydrate content. When focusing on the analyses of the variables modulating plant-pest interactions, new conclusions arise. Differences on the effect on plant growth and photosynthesis were observed among different feeding guilds or plant hosts, suggesting that these variables are key players in the final effects. Regarding the ontogenetic stage of a plant, negative effects were reported only in infestations during the vegetative stage of the plant, while no effect was observed during the reproductive stage. In addition, a direct relation was found between the durability and magnitude of the infestation, and the final negative effect on plant fitness. Among the parameters used to estimate the cost, growth and photosynthesis revealed more differences among subgroups than reproduction parameters. Altogether, this information on defense-growth trade-offs should be of great help for the scientific community to design pest management strategies reducing costs.

Keywords: fitness; growth; induced defenses; photosynthesis; phytophagous; plant; reproduction; trade-off.

Figure 1

A diagram depicting the concept of growth defense tradeoffs, the parameters to evaluate the fitness and the main variables involved. Resource allocation is related to different processes by arrows. Solid arrows refer to natural processes occurring in plants, while arrows with dashed lines refer to events in which resource allocation is altered by herbivore infestation.

Figure 2

Effect sizes on general parameters related to fitness. Sample sizes are provided in brackets. Symbols specify mean values of Hedge's g with their 95%CI. Negative values indicate a higher negative effect in fitness on attacked plants than control plants. Different letters indicate significant differences between subgroups (p-value < 0.05). Statistical analysis were performed using Kruskal Wallis tests for non-normally distributed data followed by post-hoc Dunn's tests. Rosenberg's fail-safe numbers are reported in italics. An asterisk indicates a significant fail-safe number.

Figure 3

Growth effect sizes classified by subgroups. Fitness was analyzed based in different parameters related to growth. Subgroups included feeding guild (A), type of plant host (B), plant ontogenetic stage (C), infestation length (D), and magnitude of infestation (E). Sample sizes are provided in brackets. Symbols specify mean values of Hedge's g with their 95%CI. Negative values indicate a higher negative effect in fitness on attacked plants than control plants. Different letters indicate significant differences between subgroups (p-value < 0.05). When comparing two groups, statistical analyses were performed using the parametric Student's t-test. If more than two groups were compared, statistical analysis were performed using One-way ANOVA for normally distributed data followed by Bonferroni test, and Kruskal Wallis tests for non-normally distributed data followed by post-hoc Dunn's tests. Rosenthal's fail-safe numbers are reported in italics. An asterisk indicates a significant fail-safe number.

Figure 4

Photosynthesis effect sizes classified by subgroups. Fitness was analyzed based in different parameters related to photosynthesis. Subgroups included feeding guild (A), type of plant host (B), plant ontogenetic stage (C), infestation length (D) and magnitude of infestation (E). Sample sizes are provided in brackets. Symbols specify mean values of Hedge's g with their 95%CI. Negative values indicate a higher negative effect in fitness on attacked plants than control plants. Different letters indicate significant differences between subgroups (p-value < 0.05). When comparing two groups, statistical analyses were performed using the non-parametric Mann-Whitney U-test. If more than two groups were compared, statistical analysis were performed using One-way ANOVA for normally distributed data followed by Bonferroni test, and Kruskal Wallis tests for non-normally distributed data followed by post-hoc Dunn's tests. Rosenberg's fail-safe numbers are reported in italics. An asterisk indicates a significant fail-safe number.

Figure 5

Reproduction effect sizes classified by subgroups. Fitness was analyzed based in different parameters related to reproduction. Subgroups included feeding guild (A), type of plant host (B), plant ontogenetic stage (C), infestation length (D) and magnitude of infestation (E). Sample sizes are provided in brackets. Symbols specify mean values of Hedge's g with their 95% CI. Negative values indicate a higher negative effect in fitness on attacked plants than control plants. No statistical differences are present between subgroups (p-value > 0.05). When comparing two groups, statistical analyses were performed using the parametric Student's t-test with Welch's correction for unequal variances. If more than two groups were compared, statistical analysis were performed using One-way ANOVA for normally distributed data. Rosenberg's fail-safe numbers are reported in italics. An asterisk indicates a significant fail-safe number.