Evo-Devo: Innovation and Robustness in Evolution

When I first read the paper by Ciliberti et al. (2007) I was disappointed. On the surface, the paper seems to be addressing an important issue in evolutionary theory; namely, how can you get significant innovation in light of the fact that most biological systems resist change? On closer reading, however, it seemed more complicated than that. The authors were actually dealing with a phenomenum called "robustness." This is a popular description of a simple fact—the fact that many mutations are neutral so that there can be many variants of a protein that all carry out the same function. This has been known for decades.

The people who use the word "robustness" tend to elevate it to a level of significance that makes me nervous. Furthermore, they rarely use the term "random genetic drift" or "accident" in their papers, giving the impression that "robustness" is an adaptation that favors evolution.

There's much to criticize in the field of evolutionary developmental biology or evo-devo. Some of the "theories" are little more than wide-eyed speculation. I'm thinking particularly of The Plausibility of Life by Marc Kirschner and John Gehart.
Animal Chauvinism
That's one of the problems I have with this paper. The other problem is that it's a modeling paper. The authors create a model of evolution and demonstrate that their model produces systems that evolve. I have a problem with these models. While a mathematical model is useful to show that a mechanism can work, it does not prove that it does work.

Let me give a quick example to show you why I'm skeptical of claims by modelers. It is possible to model a Lamarckian process where species inherit acquired characteristics. The result will be evolution but that does not mean that the inheritance of acquired characteristics is a real mechanism of evolution. This point is not always made clearly in papers that describe mathematical models of evolution. To often, the fact that the model produces evolution is taken as evidence that the assumptions in the model are correct and it is an accurate representation of real biological evolution. This is the same problem with just-so stories [Just-So Stories].

Let's see how Ciliberti et al. (2007) set up their experiments in the introduction to the paper.

Biologists increasingly realize that genetic systems need to be robust to both genetic and nongenetic change (7–14). Robustness means that a system keeps performing its function in the face of perturbations. For example, many proteins can continue to catalyze chemical reactions, regulate transcription, communicate signals, and serve other roles despite mutations changing many amino acids; regulatory gene networks continue to function despite noisy expression of their constituent genes; embryos continue to develop normally even when faced with substantial environmental variation. Mutational robustness means that a system produces little phenotypic variation when subjected to genotypic variation caused by mutations. At first sight, such robustness might pose a problem for evolutionary innovation, because a robust system cannot produce much of the variation that can become the basis for evolutionary innovation.

The language sounds a little strange to me but I soon realized that there were many other authors who talked about "robustness" in this way. To me, the fact that there's neutral genetic variation in a population is just a natural consequence of chance mutation and random genetic drift. I don't see why biologists think that systems "need" to be robust and I don't see why the presence of neutral variation poses a "problem" for innovative change. It's perfectly acceptable to have beneficial mutations occurring on a background of neutral variation.

The "problem" seems to be more serious for evolutionary developmental (evo-devo) biologists than for others. It has given rise to much speculation about the evolution of evolvability. If you are interested in that sort of thing you should read the book The Plausibility of Life by Marc Kirschner and John Gelhart. (Warning, the contents may not be suitable for pluralists.)

The authors of the paper (Ciliberti et al.) claim that far from being a "problem" the existence of neutral variation is actually required for innovative evolution to occur.

As we shall see, there is some truth to this appearance, but it is in other respects flawed. Robustness and the ability to innovate cannot only coexist, but the first may be a precondition for the second.

This is pretty much where I stopped reading the first time. However, Michael White over on Adaptive Complexity has highlighted this paper in a posting put up yesterday [Evolution's Balancing Act]. This suggest that the paper resonates with some evolutionary biologists and piques my interest.

The paper describes a model of an evolutionary system. It happens to be gene regulatory networks but it could be just about anything. Ciliberti et al. (2007) show that if you have a single system with no variation then the possibility for innovative change is limited. On the other hand, if you have a robust system where there are many different variants—in different species—then there are more pathways to innovative change. Seems like a pretty trivial conclusion to me. It's the sort of thing Sewell Wright was talking about (Wright, 1932).

The course of evolution through the general field [adaptive landscape-LAM] is not controlled by direction of mutation and not directly by selection, except as conditions change, but by a trial and error mechanism consisting of a largely nonadaptive differentiation of local races (due to inbreeding and by occasional crossbreeding) and a determination of long time trend by intergroup selection.

The paper doesn't mention Wright, random genetic drift, or neutral mutations; although it does talk about neutral networks.

Instead, the paper seems to be fitting in with the evo-devo concepts of evolvability and facilitated variation. In other words the idea here seems to be getting very close to the concept that the variations in different species are selected because they increase the long term potential for innovative evolution. This is very different from what Wright was saying. He said—and I agree with him—that the variation is strictly accidental and just happens to provide potential for future evolution. The distinction is important for our understanding of evolution. Does evolution see into the future? Is there a hidden purpose?

Is "robustness" selected? I doubt that any of the authors would answer yes if the question was put directly but the paper certainly gives the impression that there's something positive going on. So does the description offered by Michael White when he says thing like,

Evolution carries out an incredibly tricky balancing act: the genetic program of a species has to be resistant to small changes, yet also susceptible to the adaptive remodeling of natural selection ....

So how does evolution maintain both stability and the potential for innovation?

This could be just metaphoric. The personification of "evolution" as acting to creat robustness may be excusable on that grounds. Nevertheless, a lot of this sort of language is creeping into the evo-devo literature and I wonder if it doesn't mean something more.

---

S. Ciliberti, s., Martin, O.C. and Wagner, A. (2007) Innovation and robustness in complex regulatory gene networks. Proc. Natl. Acad. Sci. (USA) 104:13591-13596. Abstract

Wright, S. (1932) The roles of mutation, selection, inbreeding, crossbreeding, and selection in evolution. Proc. VI Intl. Cong. Genet. 1:356-366.