In my last post, I mentioned that I was reading Cycles of Contingency: Developmental Systems and Evolution and while I will highlight its major themes in a later post, I wanted to again briefly focus on one of its essays that I found fascinating. Developmental biologist H. Frederik Nijhout’s essay “The Ontogeny of Phenotype” (based on a 1997 paper) primarily argues that developmental pathways cannot be isolated from their genetic background, i.e., context matters, and the integration of developmental networks are also not cast in stone. There is a potential and remarkable fluidity to developmental systems that I was not previously aware of.
In a computer model, Nijhout and Paulsen created a diffusion gradient (Fig. 1) with six “genes” controlling six different aspects of the gradient. The gene that produces the diffusion molecule (source), the molecule’s rate of diffusion and rate of decay, the time at which the gradient is “read” (T-end) and the threshold of activation, and the background rate of production of the molecule. Each gene has two alleles: a high and low value. They assigned the gradient a “phenotypic value” that can range from zero to one and is an abstract measure of the interactions among the six genes. (The exact nature of the phenotypic value is not crucial, I don’t think.)
With selection, the phenotypic value predicably falls from a high ~17 to a low 2 (Fig. 2A). There isn’t too much excitement here but that’s because the “phenotype” is hiding all of the dynamics going on behind-the-scenes…
As Fig. 2B shows, not all the genes respond the same to selection. Source and T-end react immediately: their allele frequencies are zero by generation 8 while the other four allele frequencies remain high. Diffusion subsequently falls to zero, followed by the other three in a progression of precipitous drops. While the phenotype fell deterministically toward zero, the six genes responded erratically.
Fig. 2C shows the correlation between each gene and the phenotype. The correlations follow from 1B: at first Source is most receptive to change by selection and most highly correlated with the phenotype, followed by the other five genes, which provides us an intriguing idea: there is no single “master” gene controlling this developmental system. The master gene, the one gene most highly correlated with phenotype and modifies the developmental system as it itself changes, shifts from one gene to another throughout the course of selection. Furthermore, variation with a single gene may not always have large effects on the phenotype – only sometimes (Fig. 2C). It’s the context – the genetic background – that matters.
This is a computer model and I wonder if any similar work has been done on living organisms, if the project would even be feasible. None of the articles on Google Scholar that cite this paper seem to have done so. I don’t see how though!- the dynamics of a “simple” diffusion gradient provide a much more complicated picture than I had previously thought. Why didn’t Sean B. Carroll talk about this in Endless Forms Most Beautiful?
Nijhout, H., & Paulsen, S. (1997). Developmental Models and Polygenic Characters The American Naturalist, 149 (2) DOI: 10.1086/285996
Nijhout, H. F. “The Ontogeny of Phenotypes.” Cycles of Contingency: Developmental Systems and Evolution. Comp. Susan Oyama, Paul E. Griffiths, and Russell D. Gray. Cambridge, MA: MIT, 2001. 129-40.