Almost everyone looking at the process of evolution that has created the vast biosphere is struck initially by how surprisingly efficient the whole process has been. But quantitative estimates that might tell us whether this intuition is valid are frustratingly elusive. No one has been able to model or to estimate or even to place a reasonable lower bound on the pace of evolutionary change in a biosphere with the stats of our own Gaia. The question we would like to be able to ask is whether blind mutation and natural selection constitute a sufficient mechanism to explain all that we see in biology, and the answer is, “no one knows”. I hasten to add that it is not just religious fundamentalists who are skeptical. My favorite example is an essay by Carl Woese, but I might have cited a dozen others.

One key to the question (how evolution manages to be as efficient as it is) is the realization that the process of evolution is subject to evolution. This is “evolution of evolvability” or, as I like to call it, Evolution Squared. The idea is that in the beginning, evolution may have depended on blind mutation and natural selection, but the process has become vastly more sophisticated and efficient since then, because as nature selects (directly) for increasing fitness, she also selects (indirectly) for those communities that are advancing in fitness more rapidly. I use the word “communities” advisedly, because evolution isn’t something that happens to an individual; the smallest unit that can evolve is a deme, meaning a local set of animals or plants, all of the same species, that interbreed with one another.

Evolution of evolution has led to many innovations that we see and document, the greatest of which is sexual sharing and mixing of genes. There is no doubt that the ability to adapt to the environment within an individual’s lifetime and transmmit that adaptation to offspring would be a tremendously useful innovation. This is Lamarckian inheritance, and if it were ever to arise, it would have been copiously rewarded by natural selection for ever increasing fitness. Is Lamarckian inheritance a reality?

Fifty years after Lamarck, Darwin believed that Lamarck’s mechanism played a role in evolution. But Darwin’s heirs in the 20th Century decided that Lamarckian inheritance was implausible. If, for example, a muscle is conditioned and strengthed by constant exercise, how could the information about that muscle ever be communicated to the germ cells, the sperm or egg cells in the gonads that would be the progenitors of the next generation? Then, in the 1920s, Lysenko’s wild claims about Lamarckian inheritance pulled all credibility out from under the idea, and the scientific community firmly rejected the possibility.

Then, toward the end of the twentieth century, a strange thing happened. A new kind of semi-permanent inheritance was discovered, and it was fully Lamarckian in its implementation. This is epigenetic inheritance, the inheritance not of different versions of genes, but of patterns of gene expression. The choice of which genes are turned on or off is erased from the DNA and reprogrammed with each new embryo. But through the reprogramming, a selective memory remains; an afterimage of what was found to be useful in the previous lifetime is transmitted to the next generation.

Epigenetic memory lasts a few generations, but it is not as permanent as changes in the DNA sequence (= genetic inheritance). Could it be that genetic changes are not completely random but, like epigenetic changes, they are subject to Lamarckian influence? The prevailing skepticism of this idea is rooted in theory, and our understanding of biochemistry. Remarkably, there has been no thorough experimental exploration, not even a well-designed single trial looking for evidence of Lamarckian inheritance.

But we now know that information about gene expression does get fed back to the germline in the form of epigenetic markers. From here, it does not seem so implausible that the epigenetic markers may be translated into more permanent changes in the genome. In this paper from Washington State biologists last year, the last link in the chain is closed. The authors expose rats to a toxic fungicide, and confirm the previously-observed epigenetic changes in the rats, changes that are transmitted to their offspring. They then go on to breed the rats for three more generations, and note that there are extra copies of hundreds of genes, some of which are useful in the detox of the fungicide. These genetic changes appeared in the third generation after exposure, but they were absent in the first generation. They can’t be written off as mutagenic effects in the fungicide, because they were three generations removed from exposure.

This report does not claim creation of new genes or even new alleles, but it does include permanent changes to the germline DNA. The emerging view is that gene expression is more important in determining an organism’s structure and function (and fitness) than the precise form of the alleles themselves. 98% of our DNA is not genes but introns, the segments of DNA between genes that collectively determine the timing and circumstance of gene expression. A curious finding stressed by the authors of this study is that there is zero overlap between the areas of the genome that were epigenetically modified in Generation One after exposure and the areas of the genome that later produced extra copies in Generation Three. This suggests that the mechanism for this first example of Lamarckian genetic inheritance remains a complete mystery.

Far-reaching implications

I now believe that the remaining pieces of a fully Lamarckian evolutionary mechanism will fall into place. Books on evolution will have to be rewritten starting from Chapter 1. Everything that was learned about evolution in the 20th Century will be subject to reinterpretation, and much of it will be deemed irrelevant or naive. If Lamarckian inheritance pans out, it will turn the science of evolution on its head, and give it a good shake.