Smallest unit on which selection can act

Question

Traditionally, the individual was considered to be the smallest unit on which Natural Selection (NS) acts. Today, we usually consider the gene as being the unit of NS. Of course, we should also consider all sequences that affect the fitness even though they are not genes (even though the do not code for polypeptide). And theoretically, any sequence of DNA does have an effect on fitness because it influences the time and energy for DNA replication (although it might be negligible). The decision of considering the gene as the smallest unit of NS seems rather arbitrary to me. We might as well consider a group of genes or a given exon of even a smaller sequence.

Here are my questions:

• What factors influence the minimal size of a sequence to be considered as a unit on which NS acts? Mutation rate, generation time, selection differential for this sequence, recombination rate, ...?

• Could we consider a nucleotide as a unit of NS? Why?

• How does the quasispecies model fit into the question of what is the smallest unit of NS? (for those interested, you will also find a very good explanation of this model in Martin Nowak's book called Evolutionnary Dynamic: exploring the equations of life)

• Is it worth talking about that? Is this question biologically relevant? Or is it rather a question based on a choice of definition such as "Is a virus alive?"

As I asked several questions, let me know if I should split my post into several. Otherwise, please do not hesitate to answer only very partially to this post!

UPDATE

Terdon's answer makes sense to me. I should be a bit more accurate in the reas of my question. I read The extended Phenotype from Richard Dawkins quite a long time ago and if I'm not mistaken, Dawkins says the following things

A unit on which selection acts has to be:

• active
• germ-line
• replicator

A replicator has the 3 following properties:

• fecundity
• longevity
• fidelity while being copied

Therefore for fidelity to be respected a unit of selection has to be a sequence which is not too long, so that it is not too often modified by recombination or mutation. He argues in this sense.

He also argues that nucleotides are not possible unit of selection. Indeed, it is hard to imagine a nucleotide being an active replicator. The word active means that it influences its not probability to be replicated. I don't think a nucleotide can do such a thing.

Unfortunately I don't have the book with me right now and I can't check what I have said, give you a citation nor a more accurate a reference. If anyone has some citations from this book, it will be welcome for the discussion!

Thank you!

Answer 1

The smallest unit that can be selected is, of course, the single nucleotide. The most striking examples of this are Single Nucleotide Polymorphisms (SNPs), many of which confer selective (dis)advantages.

To take a simple example, imagine a SNP that introduces a frameshift mutation, rendering a gene incapable of producing its protein. If that protein is something relatively important like p53, the SNP in question will be lethal and will be selected against. So, to take your questions one by one:

1. What factors influence the minimal size of a sequence to be considered as a unit on which NS acts? Mutation rate, generation time, selection differential for this sequence, recombination rate, ...?

   • any sequence unit whose alteration can affect fitness is a candidate for NS.

2. Could we consider a nucleotide as a unit of NS? Why?

   • Yes, if it can affect fitness (which it can, see above) it can be selected for or against.

3. How does the quasispecies model fit into the question of what is the smallest unit of NS?

   • As far as I can tell, it is irrelevant. NS can act on any self-replicating entity as long as the process of replication can produce variation that can affect fitness. If I understand correctly, the quasispecies model simply posits a group of entities with a very very high mutational rate. However, the rate of mutation does not affect the smallest unit that NS can act on, as I said above, anything that can affect fitness will be subject to NS.

4. Is it worth talking about that? Is this question biologically relevant? Or is it rather a question based on a choice of definition such as "Is a virus alive?"

   Yes, it is worth talking about it if you have the misconception that the smallest unit is a gene :). No, but seriously, it is an interesting concept and a very useful tool. The selfish gene hypothesis for example, helped us see the world of evolution in terms of sequences rather than species or individuals. NS did not change, our conception of it did.

These are all useful conceptual tools that allow us as scientists to understand complex questions. However, NS is not a directed or conscious process so it does not care or know what it acts on.

---

PS. Please do not confuse protein-coding sequences with genes. As I said in my comment, there are many genes that do not produce proteins (tRNA genes for example) and that was even before the waters were muddied in the past few years. The current definition of gene is something like:

any nucleotide sequence with a job to do

Read the ENCODE papers for a more technical definition or take this one from wikipedia:

A gene is a molecular unit of heredity of a living organism.

---

UPDATE: First of all, you have to remember that Dawkins wrote The Extended Phenotype more than 30 years ago, in a time when the only complete genomes available were those of a few viruses. The first bacterial genome was sequenced in the mid nineties and the first human genome draft human in 2000. It was a time when selection was seen as acting at the level of the individual, and the importance of genes was underestimated. Dawkins books were instrumental in bringing genes into the spotlight.

However, if he did argue that a nucleotide cannot be selected (I have not read this particular book) he was simply wrong. A nucleotide can indeed affect its chances of being replicated. Think about it, if you have a G at a given position in the genome, and that G is mutated to an A, then the chances of the original G being replicated are obviously very low. Therefore, the nucleotide 'affected' its chances of replication. Alternatively, a mutation of a single nucleotide in the promoter of a gene can also cause the entire gene not to be transcribed and, as I mentioned before, frameshift mutations can wreak havoc as well. All of these are down to single nucleotides.

In any case, our vision of what a gene is has changed enormously since Dawkins wrote his book, we understand the genome much better now, and we have realized that it is far more complex than originally believed. 30-year old literature about genes, even if written by such luminaries as Rihard Dawkins, should be taken with a pinch of salt today.