Genome size scaling through phenotype space.

BACKGROUND AND AIMS: Early observations that genome size was positively correlated with cell size formed the basis of hypothesized consequences of genome size variation at higher phenotypic scales. This scaling was supported by several studies showing a positive relationship between genome size and seed mass, and various metrics of growth and leaf morphology. However, many of these studies were undertaken with limited species sets, and often performed within a single genus. Here we seek to generalize the relationship between genome size and the phenotype by examining eight phenotypic traits using large cross-species comparisons involving diverse assemblages of angiosperm and gymnosperm species. These analyses are presented in order of increasing scale (roughly equating to the number of cells required to produce a particular phenotypic trait), following the order of: cell size (guard cell and epidermal), stomatal density, seed mass, leaf mass per unit area (LMA), wood density, photosynthetic rate and finally maximum plant height. SCOPE: The results show that genome size is a strong predictor of phenotypic traits at the cellular level (guard cell length and epidermal cell area had significant positive relationships with genome size). Stomatal density decreased with increasing genome size, but this did not lead to decreased photosynthetic rate. At higher phenotypic scales, the predictive power of genome size generally diminishes (genome size had weak predictive power for both LMA and seed mass), except in the interesting case of maximum plant height (tree species tend to have small genomes). There was no relationship with wood density. The general observation that species with larger genome size have larger seed mass was supported; however, species with small genome size can also have large seed masses. All of these analyses involved robust comparative methods that incorporate the phylogenetic relationships of species.
CONCLUSIONS: Genome size correlations are quite strong at the cellular level but decrease in predictive power with increasing phenotypic scale. Our hope is that these results may lead to new mechanistic hypotheses about why genome size scaling exists at the cellular level, and why nucleotypic consequences diminish at higher phenotypic scales.