Data from: How important are functional and developmental constraints on phenotypic evolution? An empirical test with the stomatal anatomy of flowering plants

Quantifying the relative contribution of functional and developmental constraints on phenotypic variation is a longstanding goal of macroevolution, but it is often difficult to distinguish different types of constraints. Alternatively, selection can limit phenotypic (co)variation if some trait combinations are generally maladaptive. The anatomy of leaves with stomata on both surfaces (amphistomatous) presents a unique opportunity to test the importance of functional and developmental constraints on phenotypyic evolution. The key insight is that stomata on each leaf surface encounter the same functional and developmental constraints, but potentially different selective pressures because of leaf asymmetry in light capture, gas exchange, and other features. Independent evolution of stomatal traits on each surface implies that functional and developmental constraints alone likely do not explain trait covariance. Packing limits on how many stomata can fit into a finite epidermis and cell-size-mediated developmental integration are hypothesized to constrain variation in stomatal anatomy. The simple geometry of the planar leaf surface and knowledge of stomatal development makes it possible to derive equations for phenotypic (co)variance caused by these constraints and compare them with data. We analyzed evolutionary covariance between stomatal density and length in amphistomatous leaves from 236 phylogenetically independent contrasts using a robust Bayesian model. Stomatal anatomy on each surface diverges partially independently, meaning that packing limits and developmental integration are not sufficient to explain phenotypic (co)variation. Hence, (co)variation in ecologically important traits like stomata arises in part because there is a limited range of evolutionary optima. We show how it is possible to evaluate the contribution of different constraints by deriving expected patterns of (co)variance and testing them using similar but separate tissues, organs, or sexes.