Developmental integration cannot explain major features of stomatal anatomical evolution in seed plants

Developmental integration can cause traits to covary over macroevolutionary time and in some cases prevent populations from reaching their adaptive optima. Developmental integration between stomatal size and density may contribute to two major features of stomatal anatomical evolution: inverse size-density scaling and bimodal stomatal ratio. If these patterns result from developmental integration, we predicted that in amphistomatous leaves 1) stomatal size and density should covary similarly on both abaxial and adaxial surfaces and 2) stomatal traits (size and density) on each surface should covary isometrically. We synthesized data on stomatal density and length from amphistomatous leaves of 711 terrestrial seed plant taxa mostly from the literature. We estimated the covariance in divergence between stomatal traits from 327 phylogenetically independent contrasts using a robust Bayesian model. Adaxial stomatal density, but not length, is evolutionarily labile and not strongly integrated with stomatal length or abaxial stomatal density. Hence, developmental integration alone cannot explain inverse size-density scaling nor bimodal stomatal ratio. Quasi-independent evolution of stomatal anatomical traits facilitates largely unfettered access to fitness optima. If stomatal anatomical traits are near their current fitness optimum, this implies that limits on trait (co)variance result from selective rather than developmental constraints. However, we cannot rule out that developmental integration is important in some lineages. Future research should identify the mechanistic basis of(dis)integration in stomatal development.