Books: "Biased Embryos and Evolution" by Wallace Arthur

This is a post from June 28, 2005, reviewing one of my favourite new evolution books:

Ever since I read Gould's Ontogeny and Phylogeny in about 1992 or 1993., I knew I wanted to do research that has something to do with evolution, development and timing. Well, when I applied to grad school, I could choose between evolution OR development OR timing, but not any combination of two or more - the true evo-devo folks were just not available for me at that precise moment in history. I chose timing and than worked dilligently to infuse my work with as much evolution and development as I could. I see a lot of similarities between evo-devo and chronobiology.

While I do not ready every evo-devo paper like PZ has to, I have maintained my interest in the field and read a number of books on the topic, e.g., Embryos, Genes and Evolution by Raff and Kaufman, Shapes of Time by Ken McNamara and many others that touch on it in one way or another (mostly books against genocentrism). I took a graduate seminar based on Raff's The Shape of Life when it just came out in 1996. One I liked a lot was A Theory of the Evolution of Development by Wallace Arthur. I bought and skimmed a few times his more recent Origin of Animal Body Plans but have never read it through - I always wanted to, but real life interfered. Thus, I was excited to hear that he has published a new one, shorter and apparently less technical book, Biased Embryos and Evolution.

Biased Embryos and Evolution is a slim volume written in a very easy, flowing, almost chatty tone, quite accessible to lay people with an interest in evolution. While many authors promise to keep the technical details to the minimum, Arthur actually does manage to achieve that goal. He warns, in several places, that he is oversimplifying in order to make a point. This strategy makes it very easy to understand the topic, yet, if one looks up the complex detailed complete story elsewhere, the point he makes is still valid. Thus, I guess, it is not oversimplification, but just simplification.

A number of books have been written that attempt to downplay natural selection as a major mechanism of evolutionary change in favor of authors' pet alternatives. Arthur does not do that. While chastising the Darwinian Synthesis for its flaws, he does not in any way try to reduce the important of natural selection. While some authors treat the compendium of evolutionary mechanisms as a zero-sum game in which rise in importance of one mechanism necessarily diminishes the importance of others, Arthur attempts a much loftier goal - a synthesis that puts several mechanisms together in a non-additive manner. He builds a theory of evolution in which several mechanisms work together, not compete against each other. I think that he succeeds brilliantly.

It is a disservice to Arthur to try to summarize in a couple of paragraphs what he spent a whole book to develop, but I will try anyway. While Fisherian mathematized population genetics focuses sharply on the changes in populations, with genes being just imaginary units of inheritance, and Dawkinsian selfish-gene hypothesis focuses on genes as the only relevant players, Arthur reframes the whole problem in a new way. In order to do so, he gradually introduces the reader to each of the main factors in evolution. First, he explains the potency of the model of an adaptive landscape. Then, he shows how, everything else being equal, the size of standing variation affects evolution. Then he demonstrates how mutations of different sizes/effects, everything else being equal, affect evolution. Then he shows how developmental bias, i.e., the probabilities of particular changes in developmental trajectories being unequal, everything else being equal, affect evolution. Next, he demonstrates how the shape of the adaptive landscape, everything else being equal, affects evolution. Finally, he shows how contigency of history affects evolution.

An essential tool in understanding Arthur's model is the mental shift in thinking about organisms. Instead of thinking about adults only, one needs to think about organisms as life cycles. An egg, a larva, a pupa, a youngster, a mature adult, a senescent individual - all those phases in the life cycle are affected by mutations, by developmental bias, and by natural selection.

There are seven major messages of his book that he briefly summarizes in the end. The first one is "the return of the organism". Instead of focusing just on populations, or just on genes, Arthur re-focuses evolutionary theory at the most relevant level, that of an organism. Focusing on the organism does not place either genes or the populations into the background. Quite contrary, it serves as a connection between genes and the populations. As Robert Brandon famously stated, genes are invisible to selection. Yet, population genetics assumes that the genes are visible to selection. How? Via phenotypes. But that is an oversimplified notion that a mutation in one gene predictably leads always to the same change in the phenotype. This one-gene one-trait view is sometimes called "bean-bag genetics".

What Arthur does is place development in between - his second major point. Mutations result in changes in developmental trajectories. The changes in developmental trajectories lead to developmental reorganization. Some such changes are easy to make, others more difficult, yet others impossible. Some such changes are more likely to occur (through sheer statistics), others less likely (thus less likely to appear FIRST and influence the future biases). This view leads one to think of "all possible creatures"., i.e., to not assume that every phenotypic change is possible. Some creatures are much more possible than others. Six-legged horses are improbable, but five-legged horses are impossible. Population genetics assumes equal potency for each mutation and envision all possible creatures to be really possible. Incorporation of developmental bias eliminates many such creatures as actually impossible.

The third major point is the differentiation between internal and external selection. Almost all of the literature on adaptation tests hypotheses of external selection, i.e., adaptations to the vagaries of the external environment. Preciously few concern themselves with the adaptiveness of well-built animal. Arthur uses the term "coadapatation" to write about the adaptive value of having a balanced development leading to a healthy, strong, fast adult, no matter in which environment such animal finds itself in. This distinction between external and internal adaptatations and the ways the two interact has actually been one of the core ideas of evolutionary chronobiology.

The developmental reaction norm - the fourth major message of the book - concerns itself with the way genes and environment, via development, interact with each other. The same genotypes will develop different phenotypes in different environments. The reaction norm is also likely to be genetically based. A mutation, thus, may alter the way a particuler developmental pathway responds to the environmental cues. It is here that Arthur makes an important distinction between population and quantitative genetics. This was a jolt for me because I took a graduate class called "Population and Quantitative Genetics" in which one professor (an evo-devo guy, actually) would spend a couple of lectures utilizing a particular mathematical method to solve a particular problem concerning changes of frequency of one or two genes in a population. The following couple of lectures would then be taught by a quantitative geneticist (one of the authors of the premier textbook in the field), solving similar questions with similar math, but expanding it to many genes. As a result, there appeared a seamless continuum between population and quantitative genetics, not a night-and-day contrast that Arthur paints. There was not that much mention of developmental reaction norms and such. Did I miss something about quantitative genetics or did Arthur? And yes, I took that class AFTER I have read a bunch of Richard Lewontin and evo-devo, so I was taking it primarily in a "know thy enemy" mental mode.

The fifth major message is the topic of modularity, something that has been a topic of interest for quite a long time mainly in the philosophy of biology literature, but more recently also in the evo-devo literature. In essence, there are quasi-autonomous developmental modules. Development of limbs may be quite independent of the development of the spleen. Yet, the autonomy is never complete, as many genes are used in multiple developmental modules. Still, existence of semi-autonomy allows for various types of developmental reprogramming, including heterochrony (timing or rates of development of two or more parts of the organism shift relative to each other), heterotopy (a part develops in a different region of the body), heterometry (change in amount of some developmental module, i.e,. getting bigger or smaller), and heterotypy (production of a novel part).

A chapter on phylogeny reconstruction brings in the sixth major message - that of contingency. The evolution took a particular trajectory. Out of many "possible creatures" some actually evolved and others did not. The developmental reprogramming that led to the development of that creature shifts the world of possible creatures: some previously possible become impossible and vice versa. There is no going back once a path in the fork was taken.

The seventh major point is cooption, the only truly new idea Arthur discusses, one that came out of recent evo-devo research. You can see two excellent examples of cooption here and here. Cooption of old genes (and not just genes, but higher-level entities, e.g., hormones) for new roles relates to the ease of modifying embryos in different ways - some changes being easier (biased for) than others (biased against). The positive connotation of such developmental bias, and the use of the terms cooption and reprogramming, IMHO, are much better than the use of the term "developmental constraints" with its primarily negative meaning (that creationists can latch onto - selection as a weeding process incapable of producing novelty).

While I thoroughly enjoyed the book (I have never seen homeobox genes explained so clearly), I have a small beef. In a couple of places, Arthur mentions that he is talking only about mutations in developmental genes and not the housekeeping genes. But, if one seriously thinks of organisms as life-cycles, there is no such thing as a housekeeping gene - all genes are developmental, even if they only help you survive from one moment to another in your old age. Mutations in genes responsible for gene replication, transcription or translation will have effects on every process in the body throughout its life-cycle. Mutations in genes responsible for day-to-day physiology or behavior likewise affect the moment-to-moment developmental trajectory during at least one phase of life between zygote and death. Granting that ALL genes are developmental genes actually makes Arthur's model more powerful.

Many books have been written criticizing Neo-Darwinism and its concommitant genocentrism, without putting much new to replace it. Some books, like Gould's Magnum Opus have built an edifice of modern evolutionary theory for the 21st century that puts embryology in the center where it belongs. But I have yet to find a book that explains it as clearly and simply as Arthur's "Biased Embryos and Evolution". When Gould's Big Book came out just before he died, I read it and lamented that he has written Das Kapital for the intelligentzia, but nobody has yet written the equivalent of the Communist Manifesto for the masses (or: he has written the new Bible, but not a readable church pamphlet, or George Lakoff's slim "Elephant" in comparison to his serious "Moral Politics"). I think that "Biased Embryos" is as close to the new Evolutionary Manifesto as is possible. I strongly recommend it. Now, if someone would just write an evolutionary biology textbook based on this view, so we can start training the new generations of scientists correctly....