Note: This series of posts is intended as a basic introduction to the science of evolution for non-specialists. You can see the introduction to this series here. In this post we examine some examples of how natural selection has acted on the human genome.
Review: how natural selection works
In the last two posts in this series, we examined how natural and artificial selection shaped the dog genome over time. One example that we discussed was the duplication of the gene for the amylase enzyme in dogs. Recall that amylase is a protein enzyme produced in the pancreas that breaks down starch. This duplication increases the amount of amylase enzyme secreted into the dog’s digestive system, and in turn allows dogs with the duplication to derive more nutrition from the high-starch diet they were scavenging (or receiving) from humans. Since it provided a nutritional benefit to the dogs that carried it, these dogs would reproduce at a slightly higher average rate than dogs without it.
The original duplication event would have occurred in one dog as an error during chromosome replication. Over many generations, the “duplicated amylase gene” variant would become more and more common in the population, since dogs with it would leave more offspring, on average, than dogs without it. Later, additional duplications of the original duplication would arise, giving some dogs an even greater amount of amylase. Eventually, the original non-duplicated variant would disappear from the dog population altogether, though it would persist unchanged in wolves. Now, note well—there is a reasonable probability that a similar duplication has occurred in a wolf at some time—but it was not selected for, since wolves would derive no benefit from an increased ability to break down starch. Such a duplication, if it occurred, would have been lost from the wolf population it arose in.
To summarize, the overall process has a number of steps that can be generalized:
Random mutation: “random” can be a theologically loaded word, but for our purposes, we will use the biological definition of “random”: that the mutation event (the duplication) was “random with respect to fitness.” What this means is that the mutation event was not connected to, nor foreseeing, the benefit that it would provide. It was simply one of many mutations that occurred in ancestral dogs. We know about it because it has been passed down to dogs in the present day (given its selective advantage). Many other mutations that had no effect (or a negative effect) also occurred, but these have not been selected for. I’ve often encountered the misconception among non-biologists that mutations are always harmful, or always remove functions and information. As this example illustrates, however, in many cases mutations can be beneficial, add gene copies, and new functions and information to the organism as well. In a later post in this series, we’ll explore a wide range of different mutations, and examine how they can add or remove functions—but for our present purposes, it’s enough to underscore that not all mutations are harmful, and some are decidedly beneficial.
Natural selection: once the new, duplicated variant arose, it provided a reproductive advantage compared to the non-duplicated version. Any time one variant reproduces at a greater rate than another, natural selection is happening. The duplicated variant became more common in the population (since dogs with it reproduced, on average, more often than dogs without it). Oftentimes natural selection is viewed as a sudden, dramatic slaughter of the “unfit” with only the new, “greatly improved” individuals surviving. This is a popular, but inaccurate understanding—natural selection can be as simple as a slightly increased reproduction rate over many generations. In this case, dogs without the duplicated amylase genes continued to reproduce, but just slightly less frequently than dogs with the duplication.
Change in average characteristics within the population over time: at the start of the process, only one dog had an increased ability to produce amylase. By the end of the process, many, many generations later, all dogs had this ability, because they had all inherited the duplicated version (i.e. it had replaced the non-duplicated variant in the population). Over time, the average ability of the population to digest starch improved. Again, one common misconception of evolution is that it is a dramatic, sudden process, with offspring that differ greatly from their parents. Not so—evolution is a gradual process, with average characteristics shifting slowly over time within a breeding population.
In summary, mutations introduce variation, and not all variants reproduce at the same frequency in a given environment (i.e. the environment acts as a selective filter). Over many generations, these effects can shift the average characteristics of a population.
Has natural selection shaped the human genome?
Sometimes students, having learned about natural selection in other organisms, balk at the notion that this process was involved in our own origins. Despite this hesitation, there is very strong evidence that our own lineage has been subject to natural selection over its long history leading to our species. One example of this evidence comes from the history of our own amylase genes. The story shares similarities to what we have seen for the dog lineage, but also has some interesting differences.
Unlike dogs, humans have two distinct types of amylase genes. Both types have the same enzymatic function (breaking down starch), but they are produced in different locations in the body. One of the types is produced in the pancreas, just like the equivalent enzyme in dogs. Unlike dogs, however, humans have amylase in our saliva as well. This “salivary” amylase works quickly enough that we perceive starchy foods as sweet when we chew them—the amylase enzyme goes to work on the starch, breaking it down into glucose quickly enough for us to taste it. Studies have also shown that salivary amylase continues breaking down starch right through our stomachs and on into our intestines—thereby increasing the amount of glucose we can extract from foods rich in starch.
As you might expect, the human pancreatic and salivary amylase genes sit side-by-side in our genomes, and show the clear signs of being duplicates of each other.* All mammals have pancreatic amylase genes, but only some, such as humans, have salivary amylase genes. This means that the ancestral state was a single pancreatic amylase gene, and the first duplication event produced a second copy, just like what we have seen for dogs. This doubling of pancreatic amylase would likely have been an advantage and come under natural selection in a similar way to what we have seen for dogs. The fact that humans and other great apes share the same duplication event indicates that this event took place in the common ancestor of these species, and thus on the order of 16-20 million years ago.
Once the two pancreatic amylase genes were present in our ancestral lineage, a second event occurred that altered one of the copies: an endogenous retrovirus inserted into the genome next to one of the copies. (Retroviruses are viruses that insert their own genome into the genome of their hosts as part of their infection cycle. Endogenous retroviruses insert into the genome of reproductive cells, such as eggs or sperm—and once inserted, they can persist at a specific spot in a host genome and be passed down from generation to generation.) This retrovirus insertion event altered the DNA sequence that controlled when and where the amylase protein was made—and instead of being made in the pancreas, the altered copy began to be made in salivary glands instead.** Over time, this new combination (one pancreatic copy and one salivary copy) came under natural selection and replaced the previous version that gave rise to it (two pancreatic copies).
So, the story of the human amylase gene cluster thus far shows clear signs of repeated mutations (such as duplications) coupled with natural selection to produce the genes that we see in humans today. Of course, if humans were directly created without common ancestry, there would be no need to create these genes directly and then embed within them the evidence of a convoluted history—yet what we see, time and time again, is the clear evidence of mutation and natural selection. It seems that God was pleased to create this aspect of our biology slowly, through what we perceive as a “natural” process—but of course, what we perceive as “natural” is merely the consistent outworking of God’s ordained and sustained providence that is amenable to scientific investigation. As we became human, and shifted our diet towards agriculture and starchy foods, this God-given mechanism allowed us to take advantage of the shift in our environment.
Tomorrow, we’ll continue this story and use it to discuss the evidence that natural selection is still at work in human populations.
* Space does not permit a detailed discussion of the features of the various amylase gene copies that reveal their duplication and / or mutation history. Readers interested in the details can find them in the following published papers:
Samuelson, L.C. et al., (1996). Amylase gene structures in primates: retroposon insertions and promoter evolution. Molecular Biology and Evolution 13; 767-779. (link)
Meisler, M.H. and Ting, C.N. (1993). The remarkable evolutionary history of the human amylase genes. Crit Rev Oral Biol Med 4; 503-509. (link)
** For readers who follow the Intelligent Design literature closely, the production of the salivary-specific promoter sequence is what ID proponent Michael Behe would describe as a “gain-of-Functional Coded elemenT” (FCT) mutation. The promoter sequence is derived partially from the retrovirus sequence and partially from the DNA sequence next to the insertion site. As such, neither the virus nor the host DNA contain a FCT that can produce expression in the salivary gland. Their combined sequences create the FCT de novo, and this FCT is lost when the virus excises from the one copy, reverting it to expression in the pancreas. Readers may recall that I have critiqued Behe’s arguments based on FCTs in a previous five-part series.