*2.3. Principle Component Analysis*

For each trait and accession, we calculated a response ratio by dividing the value of the trait due to growth under FL by its value under U. Based on the log−2 transformed response ratios of all phenotypes to FL relative to U, a principal component analysis was constructed (Figure 3). Principal component 1 accounted for 87.9% of the total variance, and was characterized by large loadings for traits in growth and development (PLA, leaf size, leaf number). Principal component 2 accounted for another 9.3%, and was dominated by a large loading for NPQ. Given that principal component 1 accounted for most of the variation and was dominated by traits related to growth, the PCA reveals that variation in growth was the biggest determinant for the overall variation in our data. Generally, there was little clustering of accessions based on any of the two principal components (Figure 3), suggesting large genetic variation for both groups of traits among the 36 accessions studied here.

**Figure 3.** Principal component analysis of 36 Arabidopsis accessions, based on differences in phenotypical response ratio under fluctuating vs. uniform light (log−2 transformed FL/U ratio). PC1 accounts for 87.9% and PC2 accounts for 9.3% of the total variation.
