SummaryPhenotypic robustness is a central property of life, manifested in the ability of organisms to endure perturbing conditions throughout their development and often yield rather constant phenotypes. Fundamental questions on robustness (canalization) remain to be answered (see [1]). Do alleles that confer robustness against one perturbation also confer robustness to others? Is the robustness observed in multiple traits/taxa achieved through shared or specific mechanisms? Here, we describe an elementary model of trait development that yields phenotypic robustness without dedicated systems of developmental or transcriptional buffering. Robustness emerges when extremely low or high levels of gene activity lead to either depletion or saturation of the developmental system. We use this model to show that experimental results associating robustness to apparently redundant cis-regulatory sequences (from [2]) probably reflect a similar elementary system of saturation/depletion. We then analyze a large dataset of phenotypic responses of diverse traits of animals, plants and bacteria (from [3]) and show that the amount of response is mostly determined by the distance to the phenotypic extremes. Moreover, the most robust genotypes are often those that yield either extremely low or high phenotypes. Our results help reframing the concepts of canalization and plasticity, suggesting that phenotypic responses are mainly the result of variation in the very systems controlling each trait, rather than being attributable to either “plasticity genes” or “canalization genes”. Furthermore, they provide a hint on the causes of the genomic ubiquity of apparently redundant cis-regulatory sequences [4,5].
It takes extremes to be robust
