Skip to main content
Fig. 2 | BMC Ecology and Evolution

Fig. 2

From: Development and selective grain make plasticity 'take the lead' in adaptive evolution

Fig. 2

Phenotypic distributions arising from genetic, epigenetic or environmental perturbations are not independent. In a large random ensemble of GRNs (n = 106), systematic parametric variations were introduced into each of their elements. Each perturbation on an element generates a collection of phenotypes in a two-trait morphospace (a XPM map), characterised by a linear slope SXPM (see Fig. 1D–F). A For each GRN, we compare these slopes, two by two, searching for their correlations in the two-slope spaces (note that these are not two-trait morphospaces). Each dot is a GRN, and the yellow shaded region contains 90% of the GRNs. Correlations are significant (Pearson r > 0.3) for every combination of maps considered. B Histograms showing the probability distribution of maps with developmental insensitivity to the first (Sx ≤ 0.01; yellow) or second (Sy ≤ 0.01; red) type of inputs; and of correlated (parallel slopes); anti-correlated (perpendicular slopes) and non-correlated slopes (otherwise). Each of these cases correspond to the sector of a hypothetical circumference engulfing all points of (A), as exemplified in the coloured circumference, and the relative frequency represents the probability of each point to be located within each sector. C The complexities of the parameter-to-phenotype maps (i.e., how non-linear they are, see “Methods”); rather than between their linear slopes are also positive (Pearson r > 0.56). In (C), the colour represents slope similarity: similar slopes (black colour) are associated to simpler (i.e., more linear) maps. n = 30 replicates, GRN + multilinear model (see Additional file 1: Fig S2 for correlations under other models and Additional file 1: Fig. S5 for a null model on C)

Back to article page