We continue with the PBS/”It’s ok to be smart” collaboration that’s produced twelve videos on evolution for the holiday season. Here’s #2: “Is evolution random?”
It’s better than yesterday’s video on what evolution is, but of course the petulant PCC(E) has two beefs—though they’re trivial. First, the DNA sequence of an organism isn’t really like a blueprint. For with a blueprint you can take the finished house and reverse-engineer the instructions (the blueprint); but you can’t reverse-engineer (reconstruct) the DNA sequence of an organism from its phenotype. Rather, as Richard Dawkins has pointed out, the DNA is more like a “recipe”: you can’t reconstruct a recipe from tasting or looking at a cake but the recipe does give the instructions for making a cake.
Further, I would have liked a brief definition of “random.” My colleague Paul Sniegowski prefers to use the word “indifferent mutations” rather than “random mutations”, as it’s more precise. That is, mutations occur as if they were indifferent to their effect on the phenotype: they’re not more likely to occur when they’d increase fitness. “Random” is a bit less precise because in some senses mutations aren’t “random”: some genes are more likely to mutate than others, and some sites in the DNA are more likely to change than others.
All told, though, this clip is pretty good.
I have to say, though, that I find it annoying that nearly a quarter of the video is occupied by a blurb for the sponsor, Dropbox.
I agree with you . Too much time devoted to sponsor message
Thanks, PCC(E), for that blueprint/recipe distinction. Very helpful! Now I will watch the video.
Yes, I remember reading Dawkins on this, and my by then vaguely comprehended problem with the ‘blueprint’ metaphor dissolved.
Other Dawkins induced “aha” moments have been the selfish gene (the organism is just the vehicle) and the bayesian learning model of the genome (alleles form a set with a mechanism exactly analogous to some machine learning systems).
Very helpful all of that, it makes me (like some of Jerry’s current analysis) make some head and tails of a complicated process.
Just delimiting this to the germline, are mutations more likely to occur when they’d increase fitness? Or is it that the detrimental mutations lead to spontaneous abortion or fetal demise?
First of all, only mutations in the germline (construing this to include mitochondria) are relevant to evolution, since nothing else is inherited.
Mutations are random with respect to fitness. This means precisely that mutations are NOT more likely to occur because they increase fitness.
It does not mean that all possible mutations are equally likely to occur. To take a trivial example: of the DNA bases, G and A are chemically similar and C and T are similar. Mutations between the similar pairs are known as “transitions”, and they occur more frequently than “transversions”.
And yes, the mechanism of natural selection is subsequent fitness, or “differential mortality” if you like. The effect of a mutation is random, but if it is a mutation that is “bad” for the phenotype of the organism, it is less likely to be passed on to the next generation. Fetal mortality is a large contributor to this, but it the effect may be smaller – the organism may be viable, but may just run 5% more slowly.
@Ralph: I’m in the cancer world where we consider mutations in the germline and mutations in somatic cells. So, I had to turn off the mutations in somatic cells to think through the definitions here. Another topic could be mutations in tumor precursors, clonal expansion, and survival of the clones to tumor hood.
I concur with your thoughts above.
Somatic mutations are also random with respect to cell phenotype.
I think the clearest understanding of cancer comes from an appreciation that multicellular organisms are somewhat odd! For most of our evolutionary history, all life was single-celled. When the cell IS the organism, competition and fitness are straightforward: can a lineage of cells survive and proliferate more rapidly than competing cells?
Multicellular life is odd at first glance, because the vast majority of cells in the organism have no evolutionary descendants. Most cells in the body sacrifice their own future in order to get the DNA of a few egg or sperm cells into the next generation. This works, of course, because all cells in the body share the same DNA. So the somatic cells are still working to get copies of their own DNA into the next generation, even though the particular copy that gets passed on resides in a germline cell.
All of this requires that the DNA itself encodes the sophisticated orchestration of somatic cell types, whereby within the body individual cells are no longer competing to proliferate, but grow in a carefully coordinated and restricted manner.
But within a body, some cells will gradually acquire mutations that disrupt this sophisticated program of subservience to the germline. There are many checks and balances, but if enough mutations occur, cells may revert to the primal nature of single-celled organisms. This is cancer. It can be seen as a natural and ultimately inevitable process of cells reverting to the evolutionary state of competition among single-celled organisms, where individual cellular survival and proliferation are all.
Charleen’s jaw drops in admiration of Ralph’s beautifully penned response.
I’m also detecting a metaphor formulating in me for cooperation as nations, where ISIS, civil wars, and violence are primal, single-cell reversions and rogue tumors.
It makes me wonder. We know there are brain surgeons who deny evolution. Are there any oncologists? It seems to me utterly impossible to understand cancer if you deny evolution. Or maybe that only applies at a research level, I don’t know?
I don’t know oncologists outside the research setting. Before coming to the Fred Hutchinson Cancer Research Center in Seattle for my PhD, I did a fellowship in clinical genetics at the National Cancer Institute and worked with a team of oncologists studying familial cancer predisposition syndromes. It would be unwise for me to reveal their thoughts on Francsis Collins’s religious persona, but I believe I can safely channel the general sentiment that rejecting evolution would be viewed as ridiculous and a waste of time.
I can imagine hospice docs rejecting evolution, but it is, indeed, hard to imagine an oncologist doing so.
There is a good book along these lines by Leo Buss called “The Evolution of Individuality” that maybe you both would find interesting. Its major theme is Ralph’s idea that multicellularity requires the ‘policing’ of access to the germ line, and that what we now call embryonic development is in part a consequence of the evolution of signalling mechanisms that prevent all but one cell lineage in the embryo from forming gametes. All those other cells have to do *something*, so they make a body.
Thanks, I haven’t come across that before.
(I just searched for Leo Buss on Amazon, and was offered a “Lego Bus”, which apparently is a hotter-selling item than his book. Kids these days…)
Complicating this further on multicellular organisms, isn’t the adaptive immune system using an evolutionary process too? My vague memory is that the stem cells of cells that put out successful antibodies are multiplied, the unsuccessful not so much, meaning the system is geared towards fighting the same/similar infections in the organism’s future.
But I can easily have misunderstood the whole process. And since I am just awakened I am too lazy to read up on it right now…
I agree that it’s sensible to clarify of what “random” means. But I don’t like your colleague’s idea of discontinuing use of the word, since it is precisely correct, and it’s an important concept to get right for any student since it applies to all of probability and statistics.
A random variable is associated with a probability distribution that describes the relative probability of the various possible outcomes.
There might be a colloquial sense in which “random” sometimes implies an equal-probability distribution of outcomes. But that’s not what random means in probability and statistics, and the concept is important and easy to explain with (say) the distribution of outcomes from rolling a pair of dice.
How about “probabilistic mutation?” Incorporates both probability (distribution) and element of chance.
But “random” is precisely the correct word that means exactly that.
It’s a concept that has wide application for even the most basic probability and statistics. Surely it’s better to clarify the definition rather than circumlocute?
x is random with respect to y simply does not mean that all possible outcomes of x are equally likely. Any student with that misapprehension should be corrected.
And of course another reason to use “random” is that it is the go-to scare-adjective of cdesign proponentsists. It’s important that the general public learn when and when not it is proper to use it.
For those unfamiliar with the reference:
http://rationalwiki.org/wiki/Cdesign_proponentsists
I LOVE that! Awesome…
Reference to the excessive amount of space taken up by the sponsor’s message reminded me that weit depictions and references to Professor Coyne’s books often overlay offered pictures/video/text in such a way as to obscure them partially. Is this something that would be easy to correct? Nothing major. I just hadn’t noticed it being mentioned previously as a distraction. Maybe it didn’t distract anyone but me.
What I’ve done to view images that are initially partially covered is click on them twice. It requires extra clicking on our part, but I’ve been able to get clear views that way.
WordPress is wonky especially when it comes to formatting. A lot of look and feel issues seem like they’d be easy to fix, but aren’t. Fix something here, something breaks over there. As Charleen said, double clicking the image will reload the image without all the links.
But be sure to pause a bit between clicks, to let the written-over version load first. It’s not your usual double click.
I can see why they used a house blueprint model of evolution, as it does make for an easier metaphor when it comes to understanding the non-random selection of random variation. Though it does seem like they came up with the phrase “…created this beautiful city called life” and worked backwards from there.
Knowing that you can’t reverse-engineer based on phenotype, but you can reverse-engineer a house is an important distinction however.
I don’t think “random” implies equal probabilities for all possible outcomes. For example, when you roll a pair of dice the total will be a random number between two and twelve inclusive, but a seven is six times as likely as a two.
Natural selection is not random but the video asks “Is EVOLUTION random?”
A large part – perhaps even a majority – of evolution involves the fixation and loss of alleles by random genetic drift. This is “random.”
True enough, but virtually of the interesting changes with long-term significance have fitness effects and are influenced by non-random selection. All sophisticated biochemical pathway and organs are built up step-by-step with selection acting on intermediate functionally simpler states along the way. Otherwise you really do have impossibly unlikely scenarios, the “747 from a whirlwind in a junkyard”.
Let me rephrase that slightly:
It’s probably true that there’s a lot of contingency in the path of evolution. Just how much contingency is a matter of debate.
But that doesn’t mean that most features and attributes of the organisms that have survived arose through the accumulation of random mutations without non-random selection acting on intermediate states.
And as Jerry likes to remind us, it is the useful traits that most likely are fixated by selection. (I assume there is a small theoretical chance that genetic drift may fixate a trait that would have been fixated by selection later, or that later becomes useful.)
An easy analogy that comes to mind is electronic signaling (or computer learning, which is closer to the biological function). If you look closely enough the signal is permeated by noise. It is just if you scale back and extract the signal features that you (and the circuit) realize there is a signal, and that the signal-to-noise ratio is low enough for that to happen.
Genetic mutation proposes, natural selection disposes?
Exactly. I loved the line, “Life is the non-random selection of random variation.”
The main thing that is good about this one is that it makes a point to debunk a common creationist trope that evolution builds design by randomness. The key line is at the end: ‘it’s (evolution) the non-random selection of random variation.’
I agree about the various errors, including the view that evolution is entirely by natural selection, but to me they are small compared to the important message it was delivering.
Please correct me if I am wrong, (sayeth the umble skeptic), my understanding is that fundamentally all mutations start out as simple replication errors and are indeed pretty much random. That is, at the level of the rather simple event of the wrong base getting inserted by the polymerase, such an error is equally likely to occur anywhere along a stretch of replicating DNA. The DNA polymerase, with its palm and fingers and thumb, wrapping around the replicating strand like the hand of Joe Pesci is indeed not perfect, and it messes up –randomly. But what happens next starts to become increasingly less random – transitions have less conformational impact and are thus less likely to be corrected than transversions, things like nonsense mediated decay come in toward the last splice site and eliminate some nasty truncation variants, missense changes generally cause less impact and will generally persist more through development than splice site or frameshift-inducing errors in new cells, and on and on… And actual organism-level phenotypic effects usually kick in a bit further downstream… at which point we might view them (the actual DNA sequence variants, that is) as “indifferent”, or not.
Sorry is splitting hairs here…
You need to be more precise about what “random” means here.
(1) Random does not mean or imply equally probable.
(2) Random **with respect to** what?
The underlying principle of evolution is that mutation is random **with respect to** fitness (i.e. phenotype).
In your example: suppose for the sake of argument that, as you describe, during replication all possible base substitution errors are equally probable.
You then consider enzymatic DNA repair processes by saying “what happens next starts to become increasingly less random”. That is wrong. The DNA repair enzymes are more likely to repair transversions, so DNA repair is non-random **with respect to** transition/transversion, thus mutations have a probability distribution in which transitions are more likely. But DNA repair is random **with respect to** phenotype. So the distribution of mutations post-repair is still random with respect to phenotype.
Jerry catches the guy in a massive omission, namely, a definition of random. How about a little respect for the old boy, CD. If you are going to ignore one of the key concepts of the theory, stop riding Darwin’s coattails.
Random variation is so key in Darwin’s theory because in its absence the theory is not scientific. Random variation is admissible as a scientific notion. So if you posit random variation, you will not be dismissed out of hand as being unscientific.
What alternative is there to intrinsic random variation as a cause of variation that is scientifically admissible? If it is not random, it has a definite cause. For ID that cause is agency. For science, it would have to be definite, deterministic natural processes (let’s put probabilistic stuff aside for now).
But, of course,science seems uninterested in anything but random variation.
So, we have to understand what we mean by random variation when we are using Darwinian theory. This is not easy. Many (e.g., Dawkins) as the video just ignore the issue: sorry ma’m, no definition! It is common for the definer to substitute a word for “random.” Gould and Ken Miller like “unpredictable” and Jerry likes “indifferent” (WEIT). That kicks the can down the road besides raising new questions. Kirschner and Gerhart suggest adherents of the Modern Synthesis were partial to the notion that random meant everything is possible. That notion does not jibe well with the mantra “everything from a precursor.” But you have a pretty good idea of what is meant by all things being possible, and it is scientifically admissible.
What strikes me as a serious problem with “any variation being possible” is that it suggests infinite variation. It seems to me that evolution via natural selection might not “work” with infinite variation.
Variation is obviously constrained; it is not random. The question for evolutionary science
is the nature of that constraint. If I read them right, that is the view of Kirscner and Gerhart (Plausibility of Life).
I don’t really understand the underlying point of this word salad, nor can I discern exactly who the apparent disdain is directed toward, or why. Darwin? The video? Gould? Richard Dawkins? Jerry? All of science?
Anyway, here’s a review of the Kirschner/Gerhart book by Brian Charlesworth for those who may not be familiar with it:
http://toriah.org/articles/charlesworth-2005.pdf
Again: “random” does not mean or imply “unconstrained”. A random variable is constrained by its probability distribution.
It is significant that the notion of constraint on variation is not mentioned except in my comment. If random were understood to be consistent with constraint, you might have expected it to appear in all this discussion, not to mention Jerry’s post and the video. If random means anything can happen, that’s one thing. And there is no constraint here. If random refers to the outcome of the flip of a coin, also a random event, that is something different. Obviously the heaviest constraint. We can separate the “random part” from the “constraint part.” Imagine the situation where the organism can produce two forms of a trait (i.e., alleles). Which is the more important, the random or the constraint part?
So recognize constraint and come to understand its nature (I am addressing the science here.)
I’ll admit that “random” is not up front defined as precisely as Ralph does here, but if you dig around the concept is well defined. I could nitpick the usual loose descriptions of “genes” and “nucleotides” as well.
My take is that biology differs from physics (say) in that a lot of knowledge is contextual instead of declared. Maybe that should be changed – despite the complexity of the processes – in order to ease communication, but that is another discussion. (Admittedly, related to the videos discussed here!)
“It seems to me that evolution via natural selection might not “work” with infinite variation.”
I am not sure if this makes any sense. You would need to demonstrate how that claim “works”. =D
Else, bothering about infinity in a finite universe – the observable universe does not admit infinite variation of atom combinations because the latter are finite in numbers* – is an odd corner case. Not only is it not happening, if it was it could make qualitative differences.
Here is a recent example, where you can show that a fundamental characteristic of a quantum system is unsolvable (undecidable) in the Turing (Gödel) sense. E.g. in terms of Jerry’s article we can’t extract the microscopic blueprint from observing the macroscopic phenomena. [ http://www.sciencedaily.com/releases/2015/12/151209142727.htm ]
But that happens in theory only for systems with an infinite number of particles, i.e. non-physical ones. Admittedly I suspect that solution problems would then surface for large systems, so the corner case may be pointing to interesting effects. But problems of principle are thrown out.
* This is, I hope, not word salad. Physicist Max Tegmark builds a constraint on one of his types of multiverses on this observation. E.g. it makes a for repetition, so that we all have distant precise clones if the cosmological universe is large enough.[See Tegmark’s website.]
Or an even shorter analysis:
The genome is finite in size so variation (substitution of bases, chromosome rearrangements, et cetera) is.
The continuity of variation working on an existing system is another sense of “finite”, of finite effect size. You don’t see bacteria magically mutate into tetrapods in one generation…
Well, that sucked.
Nature starts out with finished blueprints of fully-designed houses and just tinkers randomly with the details.
Meh.
Even Erasmus Darwin’s ideas about the “irritations” of “filaments” is more enlightening.
Would it be incorrect to say that, while infinite variations may theoretically be possible, that with respect to any given population (or organism, for that matter) the number of potential variations is perhaps larger than we can count (we have to estimate), and the probability of an error occurring at any locus is equal for all of them, barring a known mechanism for favoring any set thereof?