The evolution of defenses

My research has focused on the evolution of chemical defenses in multiple tissues. Herbivore interactions vary across the plant; I'm interested in how this heterogeneous selection shapes tissue specificity of defenses. To study this, I use Boechera stricta, a wild mustard. Like many mustards, B. stricta makes several types of glucosinolates, chemical defenses against herbivores. These defenses vary within the plant. Using a combination of lab, greenhouse, and field work, I've studied how glucosinolates vary within the plant and how tissue specificity responds to selection.

Study system
Boechera stricta is a short-lived perennial native to western North America. A close relative of Arabidopsis thaliana, it offers extensive genetic and genomic resources, including a GWAS population of 500 genotyped families. B. stricta has grown in relatively undisturbed habitats for 4,000 years, developing long-term local adaptation and interactions. 
B. stricta is attacked by a number of insect herbivores, which feed on leaves, stems, buds, flowers and fruits. Herbivory, particularly on reproductive tissues, can have a large effect on plant fitness. One of the main defenses against herbivores are glucosinolates, also known as mustard oils. B. stricta makes up to four different types of glucosinolates; which types are made has been shown to affect herbivore damage. I've found significant variation in both concentration and types of glucosinolates between different plant tissues; leaves and fruits vary in their defenses.

Constraint in the evolution of defenses
Interactions with herbivores vary within the plant. Is it possible for plants to evolve effective defenses against all enemies?

Using a quantitative genetics framework, I asked whether covariances between glucosinolate profile in each tissue constrained the response to selection throughout the plant. In order to test the effects of covariances, I performed a large-scale field experiment, using 100 genotypes collected from one population, grown in a common garden located within that population. I measured herbivory, fitness, and glucosinolate traits in the field.

I found antagonistic selection on glucosinolate profile between tissues; having a high proportion of branched chain-derived glucosinolates (BC-GS) in the leaves was favored, while low BC-GS was favored in the fruits. Gluosinolate profile is strongly correlated between leaves and fruits. The genetic correlation and antagonistic selection caused strong constraints on the evolution of defenses; glucosinolate profile was predicted to evolve in a maladaptive direction. 

These constraints were driven by the pleiotropic effects of a single gene, BCMA, which controls much of the genetic variation for production of BC-GS. In a restricted population that was fixed for the functional allele of BCMA​, we observed no significant correlations and no evidence of constraints.
​
​Constraints in glucosinolate evolution. Showing the proportion of BC-GS in cauline leaves (PropBC-C) and fruits (PropBC-F). The red arrow indicates the selection gradient. The blue arrow shows the predicted response to selection. The black arrow shows the predicted response to selection when covariances are computationally removed. The lower graph shows a restricted population with only the genotypes that have the functional allele of BCMA​.

Evolution of the G-matrix

Genetic covariances between PropBC in leaves and fruits for three genetic groups within the EAST subspecies (COL, NOR, and UTA), and the WEST subspecies. Differences in the genetic correlations between PropBC in leaves and fruits drive much of the divergence in G​ between subspecies.
Within its western range, B. stricta has two subspecies, EAST and WEST, which are differentiated both genetically and morphologically (Lee and Mitchell-Olds 2013). EAST has been further sub-divided into three genetic groupings (B. Wang, unpublished data). This population structure provides an opportunity to look at divergence in the G-matrix between recently diverged populations. 

Using multiple tests of matrix similarity, we found variation in the structure of G both between EAST and WEST, and among the groups within EAST. Much of the difference between the subspecies was driven by variation in PropBC, particularly the strength of the genetic correlation between leaves and fruits.

We predict that much of the divergence in G between subspecies is due to changes in the allele frequency of BCMA. Work is ongoing to further investigate the genetic basis of the genetic correlations between tissues.