Physics for the 21st Century logo

Section 6: Evolution

Biological evolution remains one of the most contentious fields to the general public, and a dramatic example of emergent phenomena. We have been discussing the remarkable complexity of biology, and it is now natural to ask: How did this incredible complexity emerge on our planet? Perhaps a quote from French Nobel Laureate biologist Jacques Monod can put things into perspective: "Darwin's theory of evolution was the most important theory ever formulated because of its tremendous philosophical, ideological, and political implications." Today, over 150 years after the publication of "On the Origin of the Species," evolution remains hotly debated around the world, but not by most scientists. Even amongst the educated lay audience, except for some cranks, few have doubt about Newton's laws of motion or Einstein's theories of special and general relativity, but about half of the American public don't agree with Darwin's theory of evolution. Surely, physics should be able to clear this up to everybody's satisfaction.

The variation in Galapagos finches inspired Charles Darwin's thinking on evolution, but may evolve too fast for his theory.

Figure 22: The variation in Galapagos finches inspired Charles Darwin's thinking on evolution, but may evolve too fast for his theory.

Source: © Public Domain from The Zoology of the Voyage of H.M.S. Beagle, by John Gould. Edited and superintended by Charles Darwin. London: Smith Elder and Co., 1838. More info

Or maybe not. The problem is that simple theories of Darwinian evolution via random mutations and natural selection give rise to very slow change. Under laboratory conditions, mutations appear at the low rate of one mutated base pair per billion base pairs per generation. Given this low observed rate of mutations, it becomes somewhat problematic to envision evolution via natural selection moving forward to complex organisms such as humans. This became clear as evolution theories tried to move from Darwin's vague and descriptive anecdotes to a firmer mathematical foundation.

Recent work on the Galapagos Islands by the Princeton University biologists Peter and Rosemary Grant revealed something far more startling than the slow evolution of beak sizes. The Grants caught and banded thousands of finches and traced their elaborate lineage, enabling them to document the changes that individual species make in reaction to the environment. During prolonged drought, for instance, beaks may become longer and sharper, to reach the tiniest of seeds. Here is the problem: We are talking about thousands of birds, not millions. We are talking about beaks that change over periods of years, not thousands of years. How can evolution proceed so quickly?

Fitness landscapes and evolution

In our protein section, we discussed the concept of a free energy landscape. This indicates that proteins do not sit quietly in a single free energy minimum, but instead bounce around on a rough landscape of multiple local minima of different biological functional forms. But this idea of a complex energy landscape did not originate from proteins or spin glasses. It actually came from an American mathematical biologist named Sewall Wright who was trying to understand quantitatively how Darwinian evolution could give rise to higher complexity—exactly the problem that has vexed so many people.

Natural selection can be viewed as movement on a fitness landscape.

Figure 23: Natural selection can be viewed as movement on a fitness landscape.

Source: © Wikimedia Commons, Public Domain. Author: Wilke, 18 July 2004. More info

We can put the problem into simple mathematical form. Darwinian evolution is typically believed to be due to the random mutation of genes, which occurs at some very small rate of approximately 10-9 mutations/base pair-generation under laboratory conditions. At this rate, a given base pair would undergo a random mutation every billion generations or so. We also believe that the selection pressure—a quantitative measure of the environmental conditions driving evolution—is very small if we are dealing with a highly optimized genome. The number of mutations that "fix," or are selected to enter the genome, is proportional to the mutation rate times the selection pressure. Thus, the number of "fixed" mutations is very small. A Galapagos finch, a highly evolved creature with a genome optimized for its environment, should not be evolving nearly as rapidly as it does by this formulation.

There is nothing wrong with Darwin's original idea of natural selection. What is wrong is our assumption that the mutation rate is fixed at 10-9 mutations/base-pair generation, and more controversially perhaps that the mutations occur at random on the genome, or that evolution proceeds by the accumulation of single base-pair mutations: Perhaps genomic rearrangements and basepair chemical modifications (a process called "epigenetics") are just as important. Further, we are beginning to understand the role of ecological complexity and the size of the populations. The simple fitness landscape of Figure 23 is a vast and misleading simplification. Even in the 1930s, Seawall Wright realized that the dynamics of evolution had to take into account rough fitness landscapes and multiple populations weakly interbreeding across a rough landscape. Figure 24 dating all the way back to 1932, is a remarkably prescient view of where evolution biological physics is heading in the 21st century.

Sewall Wright sketched the path different populations might take on the fitness landscape.

Figure 24: Sewall Wright sketched the path different populations might take on the fitness landscape.

Source: © Sewall Wright, "The Role of Mutation, Inbreeding, Crossbreeding, and Selection in Evolution," Sixth International Congress of Genetics, Brooklyn, NY: Brooklyn Botanical Garden, 1932. More info

Darwinian evolution in a broader sense is also changing the face of physics as the fundamental concepts flow from biology to physics. Darwinian evolution as modified by recent theories teaches us that it is possible to come to local maxima in fitness in relatively short time frames through the use of deliberate error production and then natural selection amongst the errors (mutants) created. This seems somewhat counterintuitive, but the emergence of complexity from a few simple rules and the deliberate generation of mistakes can be powerfully applied to seemingly intractable problems in computational physics. Applications of Darwinian evolution in computational physics have given rise to the field of evolutionary computing. In evolutionary computing, principles taken from biology are explicitly used. Evolutionary computation uses the same iterative progress that occurs in biology as generations proceed, mutant individuals in the population compete with other members of the population in a guided random search using parallel processing to achieve the increase in net fitness. To be more specific, the steps required for the digital realization of a genetic algorithm are:

  1. A population of digital strings encode candidate solutions (for example, a long, sharp beak) to an optimization problem (needing to adapt to drought conditions).
  2. In each generation, the fitness of every string is evaluated, and multiple strings are selected based on their fitness.
  3. The strings are recombined and possibly randomly mutated to form a new population.
  4. Re-iterate the next generation.

It is possible that by exploring artificial evolution, which came from biology and moved into physics, that we will learn something about the evolutionary algorithms running in biology and the information will flow back to biology.

Evolution and Understanding Disease in the 21st Century

The power influence of evolution is felt in many areas of biology, and we are beginning to understand that the origins of many diseases, most certainly cancer, may lie in evolution and will not be controlled until we understand evolution dynamics and history much better than we do today. For example, shark cartilage is one of the more common "alternative medicines" for cancer. Why? An urban legend suggests that sharks do not get cancers. Even if sharks have lower incidence rates of cancer than Homo sapiens, they possess no magic bullet to prevent the disease. However, sharks possess an important characteristic from an evolution perspective: They represent an evolutionary dead-end. Judging from the fossil record, they have evolved very little in 300 million years, and have not attempted to scale the fitness landscape peaks that the mammals eventually conquered.

Cartilage from the fin of the Mako shark.

Figure 25: Cartilage from the fin of the Mako shark.

Source: © www.OrangeBeach.ws. More info

We can ask two questions based on what we have developed here: Is cancer an inevitable consequence of rapid evolution, and in that sense not a disease at all but a necessary outlier tail of rapid evolution? And is cancer, then, inevitably connected with high evolution rates and high stress conditions and thus impossible to "cure"?

Stress no doubt drives evolution forward, changing the fitness landscapes we have discussed from a basically smooth, flat, and boring plane into a rugged landscape of deep valleys and high peaks. Let us assume that in any local habitat or ecology is a distribution of genomes that includes some high-fitness genomes and some low-fitness genomes. The low-fitness genomes are under stress, but contain the seeds for evolution. We define stress here as something that either directly generates genomic damage, such as ionizing radiation and chemicals that directly attack DNA, viruses, or something that prevents replication of the genome, such as blockage of DNA polymerases or of the topological enzymes required for chromosome replication. Left unchallenged, all these stress inducers will result in the extinction of the quasi-species.

This is the business end of the grand experiment in exploring local fitness peaks and ultimately in generating resistance to stress. The system must evolve in response to the stress, and it must do this by deliberately generating genetic diversity to explore the fitness landscape—or not. Viewed in the perspective of game theory's prisoner's dilemma (see sidebar), the silent option under stress is not to evolve—to go down the senescent pathway and thus not attempt to propagate. Turning on mutational mechanisms, in contrast, is a defection, in the sense that it leads potentially to genomes which can propagate even in what should be lethal conditions and could, in principle, lead to the destruction of the organism: disease followed by death, which would seem to be very counterproductive. But it may well be a risk that the system is willing to make. If ignition of mutator genes and evolution to a new local maximum of fitness increases the average fitness of the group, then the inevitable loss of some individuals whose genome is mutated into a fitness valley is an acceptable cost.