Variability in plant nutrients but not defenses reduces the performance of phytophagous insects: A meta-analysis
Much of the theoretical and empirical research on plant-herbivore interactions has focused on the influence of mean plant defensive and nutritional phenotypes on herbivore performance. However, plant traits often vary in space and time, at scales ranging from individuals to populations, and the consequences of this variability for herbivores are often overlooked. One implication of plant variability is increased or decreased mean herbivore performance through averaging across nonlinear functions. This mathematical property (known as Jensen’s inequality) depends on the shape of a herbivore's performance function. If performance were a concave-down function of plant phenotype, then herbivores experiencing variability in that phenotype would have reduced performance relative to herbivores experiencing constant plants. The opposite is true for convex-down functions. Three previously published non-quantitative reviews of the shapes of herbivore performance functions have come to contrasting conclusions about the effects of plant variability on herbivores. To determine if there is a general pattern in the effects of plant variability on herbivores, we conducted a meta-analysis using herbivore performance data from 99 studies of 66 species of phytophagous insects. We tested if the effects of plant variability differed between plant defensive and nutrient traits and with herbivore traits like host breadth and mobility.
Herbivore performance functions were concave-down significantly more often than they were other shapes, suggesting that variability in plant traits frequently depresses herbivore performance. For herbivore growth or reproduction, 53% of species had mainly concave-down performance functions, whereas only 11% of species had mainly convex-down functions. This negative effect of variability was driven by nutritive plant traits, for which 71% of species had mostly concave-down growth or reproduction functions. For defensive plant traits, there was no consistent negative or positive effect of variaility on herbivore performance. These patterns were consistent across the range of herbivore host breadth and mobility. Although most plant-herbivore theory focuses on means, this work suggests that mere variability in plant traits may be key to understanding herbivory. Based on the shapes of herbivore performance curves, our results indicate that variability in plant nutrition will typically reduce average herbivore performance relative to diets that are less variable. This suggests that selection for behavioral or physiological mechanisms to cope with variability will be consistently strong for nutrient traits. Failure to consider variability in plant nutrients may limit our ability to scale from spatiotemporally restricted studies to understand plant-herbivore dynamics on larger scales.