Mean plant toxicity modulates the effects of plant defense variability

Plant trait variation is thought to suppress herbivore performance, but experiments are usually done with a single mean level of the trait. We manipulated the mean and variation of glucosinolate toxin concentration on Arabidopsis thaliana leaves and fed them to Trichoplusia ni in three field and greenhouse experiments. Plants painted with a greater mean glucosinolate concentration had higher fitness and resistance to herbivores; however, at high mean concentrations, variation reduced the defensive effect, while at lower mean concentrations, variation enhanced it. This reversal is consistent with models that include herbivore food selectivity, but our simulations revealed that the benefit of food selectivity to herbivores was minuscule. Instead, nonlinear averaging and physiological tracking effects likely drove patterns in plant fitness and resistance to herbivores. We suggest that high defense variation in plants may be a widespread convergent defensive phenotype, but for well defended..., , , # Mean plant toxicity modulates the effects of plant defense variability

Vincent S. Pan, Kadeem J. Gilbert, William C. Wetzel

[https://doi.org/10.5061/dryad.cz8w9gjbm](https://doi.org/10.5061/dryad.cz8w9gjbm)

**02-Dec-2024 update: VSP made a mistake in inputting the name of the last author 'William Wetzel' as 'Kadeem Wetzel'. The mistake has been corrected.**

## Description of the data and file structure

#### Data Files:

* field_variability_experiment.csv
* Field data on *Arabidopsis thaliana* fitness (experiment 1). Each row is a unique plant
* greenhouse_variability_data.csv
* Greenhouse data on *Trichoplusia ni* performance (experiment 2). Each row is a unique caterpillar.
* dose_response_data.csv
* Greenhouse data on *Trichoplusia ni* dose response curves (experiment 3). Each row is a unique caterpillar

#### Code Files:

* eff_part_functions_bayes.R
* Custom functions for performing counterfactual simulations that partition the effects of variability
* figures.R
* C...