Evolution is a population-level phenomenon

An argument I have been encountering a lot recently is one that goes something along the lines of:

"Natural Selection cannot be the source of novel adaptations because it only works on what it already present. It does not generate anything and therefore a novel trait cannot be the product of Natural Selection."

These claims are then used to support the primacy of mutation as the driving force behind evolution, often coupled with (unsubstantiated) claims that these mutations are not random. In other words, Natural Selection is a myth and goal-directed mutation is responsible for the evolution of adaptations.

On face value, this can seem like a convincing argument. Natural Selection does only act on existing variation. New variants do have to arise by mutation (for a given definition of mutation), which is independent of selection. However, extrapolating that to mean that adaptive evolution occurs through the source of the new variants, mutation, and not selection is a classic case of confusing individual traits with population/species traits. I suspect that this confusion is the source of misunderstanding for many people who champion directed mutation and denigrate the power or potential of Natural Selection.

We have a tendency to draw phylogenetic trees as single lines for the branches. It is important to remember, however, that these branches - representing the evolution of species of gene sequences - are actually representing whole populations of organisms or molecules.

Evolution is a population-level phenomenon: individuals mutate but they do not evolve. It is true that new variants have to arise in an individual, independent of selection. However, we do not say that a trait has "evolved" until it reaches a high frequency or even reached (effective) fixation in the population.

For example, certain mutations cause polydactyly (extra fingers and toes) in humans but we do not say that humans have "evolved six fingers". For humans, a mutation usually has to reach a frequency of 1% before being considered a polymorphism. This is somewhat arbitrary but it needs to be in at least two generations; otherwise, lethal or sterilising mutations would constitute a polymorphism and this would make the concept pretty useless. Likewise, it would be pretty silly to say that something had "evolved sterility" because a single individual had a sterilising mutation.

Evolution does not need Natural Selection. Random processes are sufficient for a neutral (or nearly neutral) trait to "drift" its way through a population to fixation. However, without invoking an external agent, Natural Selection is the only process that drives a trait through (or from) a population, resulting in adaptive evolution.

Population-level change is still change. A population or species acquiring a new trait is still evolution of a new trait. A novel evolved trait can be the product of Selection.

Is evolution random?

In a recent perusal of The Blogosphere, my eye was caught by a post at A Tippling Philosopher entitled Far from random, evolution follows a predictable genetic pattern, Princeton researchers find. As I suspect most evolutionary biologists would, I got rather annoyed by this title. I should point out, however, that the Tippling Philosopher is not to blame - this is the title of the original Princeton press release.

Unfortunately, my VPN is playing up so I cannot access the original article (Zhen Y, Aardema ML, Medina EM, Schumer M & Andolfatto P (2012). Parallel molecular evolution in an herbivore community. Science 37(6102):1634-7) but I am a bit short of time anyway and don't want to do an in depth study. I suspect the press release actually does a fairly good job of summarising the main points as does the abstract:

Numerous insects have independently evolved the ability to feed on plants that produce toxic secondary compounds called cardenolides and can sequester these compounds for use in their defense. We surveyed the protein target for cardenolides, the alpha subunit of the sodium pump, Na+,K+-ATPase (ATPα), in 14 species that feed on cardenolide-producing plants and 15 outgroups spanning three insect orders. Despite the large number of potential targets for modulating cardenolide sensitivity, amino acid substitutions associated with host-plant specialization are highly clustered, with many parallel substitutions. Additionally, we document four independent duplications of ATPα with convergent tissue-specific expression patterns. We find that unique substitutions are disproportionately associated with recent duplications relative to parallel substitutions. Together, these findings support the hypothesis that adaptation tends to take evolutionary paths that minimize negative pleiotropy.

This is all very interesting and very cool - the power of evolution by Natural Selection demonstrated in replicate. The thing that irked me, though - and the focus of this (probably too hasty) post - is the title and the first paragraph of the Princeton press release:

Far from random, evolution follows a predictable genetic pattern, Princeton researchers find

Evolution, often perceived as a series of random changes, might in fact be driven by a simple and repeated genetic solution to an environmental pressure that a broad range of species happen to share, according to new research.

Evolution is not a series of random changes. At least, adaptive evolution, which is the subject of this paper, is not. Neutral evolution largely is random but that's only one part of evolution as a whole and to imply otherwise is rather misleading. The raw material for evolution is indeed random mutation but this is not the full picture. There was a clever chap who realised that heritable random variation in a trait, if it produced differential survival and/or reproduction, could result in the non-random change of that trait. Good ("fit") traits would increase in frequency and eventually dominate the population, while bad traits would be removed. He realised this over 150 years ago and called it Natural Selection. His name was Charles Darwin and his book is free on Kindle if you want to read it. (Yes, I know, this is a gross simplification of history!)

This is compounded by the title: "Far from being random, evolution follows a predictable genetic pattern, Princeton researchers find". Well, yes, sometimes it does (and in this case) but we've also known that for years. The argument about the predictability of evolution is one that has been going on for a long time. (Read Gould's "A Wonderful Life", for example.) If you were to rewind the clock and let evolution run again, how much would history repeat itself? We know the answer is not "always" and we know the answer is not "never" but we do not know where on the continuum between "always" and "never" reality lies. (Major catastrophic events notwithstanding. These are another role of chance but somewhat different to the one determining evolutionary trajectory.)

In evolution, the opposite of "predictable" is not completely "random". One has to be clear that even if we cannot predict a precise evolutionary trajectory due to the complexity (and, yes, randomness) in the system, trajectories that give rise to exquisite adaptations always have a large non-random component (selection). This may seem like a trivial thing but it's not, for this is the kind of language that feeds the misconceptions spread my advocates of Intelligent Design and other forms of Creationism. (Of course, this study also nicely blows such nonsense out of the water.) If it is "often perceived as a series of random changes", it is only because of misconceptions like this being repeated.

For me, it is the last line of the abstract that is most intriguing and possibly the big discovery:

"Together, these findings support the hypothesis that adaptation tends to take evolutionary paths that minimize negative pleiotropy."

The authors cite "the large number of potential targets for modulating cardenolide sensitivity". It seems that they think that the other possible target genes are more prone to affect other systems as well in a bad way. (I am not sure how they rule out the possibility that the selective advantage of changes in this particular pump are just much, much greater than the other genes and its just driven by the probability and rates of positive selection.) If this turns out to be a widespread phenomenon, it could indeed have implications for the predictability of future adaptation, which could be useful in our changing world!

This is just one example, of course. Another recent paper on parallel and convergent evolution in Proc. R. Soc. B (free this time), "The probability of genetic parallelism and convergence in natural populations" by Gina Conte, Matthew Arnegard, Catherine Peichel and Dolph Schluter, looked at a bunch of studies and concluded that "estimates [of the probability of gene reuse in parallel and convergent phenotypic evolution in nature] using data from published studies. The estimates are surprisingly high, with mean probabilities of 0.32 for genetic mapping studies and 0.55 for candidate gene studies".

I'm sure there's a bit of ascertainment bias towards traits under strong selection (as these are more obvious and thus more studied) but it confirms the Andolfatto study that in the right circumstances convergent evolution can make use of the same gene(s). (They also cite counter-examples, so be quite clear that this is not universal.) It's not a surprise that it happens but given the amount of diversity between genomes - and numbers of genes affecting many traits - the level is possibly surprising. To be honest, I can't decide if I am surprised or not as it is so hard to generate a reasonable a priori expectation.

I still don't think that all this means that evolution in general is predictable (we still need more studies) but it certainly does hammer yet another nail in the coffin of the old Creationist chestnut about evolution being random. Natural Selection is NOT random - that's the whole point!

Footnote: I must concede that in writing this post I realised that explaining the role (and meaning) of "random" in the context of evolution is not quite as simple as I thought. "Random" commonly does mean a lack of predictability but I maintain that it is not helpful to use this language for evolution without some serious explanation of what you mean by random. Random mutation plus Selection means that we are talking about a lack of determinism, not a lack of direction. (This was also supposed to be a quick/short post!)

h/t: A Tippling Philosopher and WEIT

Student-induced hypermutation

Each year around this time I am both amazed and dismayed at the poor numeracy skills of some of our undergraduate students. (Particularly bearing in mind that we don't exactly have soft entry requirements.) To be honest, it's not the bad numeracy alone that upsets me - it's when common sense fails to pick up blatant bad numeracy. In particular, I am talking about spontaneous DNA mutation rates in humans during DNA replication.

The key is the superscript. A little thing but a big effect. The spontaneous error rate during human DNA replication is on the order of (and frequently quoted as) 1 mutation every 10¹⁰ bases. (Bionumbers has the rate at 5 x 10¹¹, which is close enough.) For those not familiar with powers, that's 1 in 10,000,000,000. It is most certainly not 1 in 1010. And yet, every year, a number of students confidently assert that the spontaneous mutation rate is 1 in 1010 bases. To put this in context, we have approx 3 x 10⁹ (3 billion) bases in our genome. This means that each time our genome is replicated, there is (on average) less than one mutation. If, on the other hand, the mutation rate was 1 in 1010, there would be getting on for 3 million mutations per replication. We'd be dead. (Actually, 3 billion bases is our haploid genome size and we're diploids, so we'd actually get twice as many mutations, including about 3000 affecting both copies.)

Another way to put that in perspective. Imagine that DNA was scaled up so that each nucleotide was one mm. On average, human DNA polymerase could copy DNA from the North Pole to the Equator before it made a mistake. This is approx. ten "Megametres", or 10¹⁰mm. (The metre was originally defined to be one ten millionth of this distance. Not really sure why. I guess it seemed like a good idea at the time. Anyway...) If the error rate was 1 in 1010, it probably wouldn't even make it out of the room - just over 1m. Supercripts: they're important!

(Because of the number of germline cell divisions involved in making eggs and sperm, the human per generation mutation rate is about 2.5 x 10⁸ - or a 100ish mutations per generation. Imagine if that was a billion! In real life, the polymerase would make it about 6m versus 0.6µm before making a mistake. That's still pretty impressive when you think how tiny it is! It would take a single polymerase about about 9.6 years to replicate the 6m, though! Just as well we make a whole bunch of them...)

Where do new proteins come from?

As part of my second year lectures on protein evolution, I cover (albeit far too briefly) the topic of where new proteins come from. The most common answer to this question, is "other proteins". Gene duplication is common and happens at a variety of levels, from individual domains (independent folding units of proteins) to full length proteins, multiple proteins (on larger chromosomal stretches) to whole genome duplication, in which every single gene is duplicated.

Duplication is so important because it generates a copy that can remain essentially unchanged, performing the ancestral function, and a copy that is free to accumulate mutations. Often - usually - these mutations will ultimately disable one of the copies, generating what are known as "pseudogenes", the easily identified relics of past mutational dead ends. Occasionally, however, a new function will arise, such as the binding of a new enzyme substrate, the the reception of a different wavelength of light, and both copies will be retained. 

This is fascinating stuff in itself (and I may post more on it later) but it is not what I want to talk about here. Important as they are, duplications rely on something - an existing protein-coding gene - being there to be duplicated. Where do these proteins come from in the first place? Are all modern proteins just duplicated and edited versions of a suite of ancestral proteins? Does this mean that supporters of Intelligent Design community are right and that some kind of intelligent intervention is necessary to increase the repertoire of proteins that are available?

Well, no. Far from being a statistical improbable event unlikely to ever occur, the "de novo" creation of new proteins from non-coding DNA seems to be a reasonably common event. In an article just published in PLoS Genetics, for example, "De Novo Origin of Human Protein-Coding Genes" (PLoS Genet 7(11): e1002379), Wu, Irwin & Zhang describe sixty such proteins that appear to have arisen in the human lineage since we shared a common ancestor with chimps. 

This and other work in the area is nicely summarised in the same issue by Guerzoni & McLysaght in a two page article that is well worth the read, "De Novo Origins of Human Genes" (PLoS Genet 7(11): e1002381). The Wu paper itself builds on earlier work from the McLysaght lab, and they include a nice summary of the type of evidence the work looks for (see the paper for details):



It is important to bear in mind that not all of the sixty novel protein-coding genes might be "true positive" results in the strictest sense. As Guerzoni & McLysaght point out:

"The operational definition of a de novo gene used by Wu et al. means that there may be an ORF (and thus potentially a protein-coding gene) in the chimpanzee genome that is up to 80% of the length of the human gene (for about a third of the genes the chimpanzee ORF is at least 50% of the length of the human gene). This is a more lenient criterion than employed by other studies, and this may partly explain the comparatively high number of de novo genes identified. Some of these cases may be human-specific extensions of pre-existing genes, rather than entirely de novo genes—an interesting, but distinct, phenomenon."

Nevertheless, this is still the creation of de novo protein sequence, just not whole genes. Indeed, if one were to relax the criteria, how much more might this happen within proteins? Domains are frequently linked by disordered regions and also have a tendency to be flanked by intron/exon boundaries, making it quite plausible for extra exons to be picked up along their evolutionary journey. In some ways, therefore, this sixty is likely to be an underestimate. 

There are other issues with identification process itself that make real estimates difficult. For robustness, authors limit themselves to stretches of DNA that are present in both humans and chimps, to rule out the loss of an ancestral protein, rather than the gain of a human-specific one. Insertions and deletions ("indels") are quite common and actually account for most of the difference (in absolute genetic terms) between humans and chimps:

"It is now common knowledge human and chimpanzee DNA differ by only 1% (more accurately, they differ in 1%of alignable regions of genome, with a further 3% divergence due to lineage- specific indels)."

It is quite possible, therefore, that a few extra protein sequences have arisen in non-coding human-lineage DNA that was deleted in the chimp lineage. The other limiting factor - and probably one that has a greater impact - is that of protein annotation. Discovery relies on proteins that are annotated as such in the human genome. The problem here is that such annotations are normally made using conservation of proteins from other organisms - something that is, by definition, lacking for lineage-specific proteins. This issue is not insurmountable and there are sources of evidence that are used but it is still likely to result in underestimation.

An important caveat remains, however. Although we have some examples of de novo proteins that appear to have function, for the majority of them we have no idea what they do, if indeed they do anything at all. A more interesting question opens up of when do you define translated DNA as a protein? Leaky transcription and translation almost certainly happens (although I have no idea how much) and so we might be detecting stuff that is really there but only rarely. Alternatively, it might be consistently transcribed and translated but have no function whatsoever - a neutral mutation that has drifted randomly to high frequency but may have no long-term future.

These questions and issues are unlikely to be resolved soon. For now, we just know that de novo creation of proteins is more common than we used to think and is likely to be a substantial source of raw material on which natural selection (and random drift) can act to evolve new functions. What the ultimate fate of these proteins tends to be, only time will tell.