I thought it was telling that he postulated an original human population of two. Now where could that number have come from?
(Does the maths)
I doubt anyone would agree that 1.00728 is “vastly greater” than 1 in any context.
I can’t recall if this has already been addressed: One assumption you seem to be working under is that there would have to have been more beneficial mutations fixed in the human genome than in the chimp genome in the time since our divergence from our MRCA.
If so, that is another thing you have gotten wrong.
I think maybe Alan Kleinman is mistaking my most recent comments as arguing that neutral evolution accounts for the differences in anatomy etc. between humans and chimps. They were arguing against that and saying that even if most change is neutral, the substantive differences in phenotype are due to natural selection. That could be why Kleinman keeps trying to argue with me that neutral mutation won’t do the job.
Sorry but I don’t accept your speculation. And selection for any mutation in the regulatory portion of the DNA will obey the same DNA evolutionary mathematics that governs mutations in the protein-coding portion of the DNA. If you think otherwise, present your mathematical and empirical evidence.
The flaw in this argument is that you must assume that the genetic differences between humans and chimpanzees can be accounted for by neutral evolution. However, humans and chimpanzees have significantly different reproductive fitnesses as demonstrated by the 20,000 times differences in population size. So if humans and chimpanzees are distantly related, some of those mutations had to be beneficial for humans to account for the differences in reproductive fitness.
Humans have the capability to modify our environments to make them habitable. If we don’t have water, we drill wells or pipe it in. If the environment is too cold or hot, we design and build heating and cooling systems to make the environment habitable. If we need food, we develop agricultural systems that supply that food. And what you call a synergy of small effects has to be a set of adaptive mutations which chimpanzees don’t have for your theory to be correct (that humans and chimpanzees descended from a common ancestor).
Every adaptive evolutionary direction is an evolutionary trajectory and all of these evolutionary trajectories obey the same mathematics.
You are misinterpreting the empirical evidence. Multiple possible evolutionary trajectories do not change the probabilities drastically. To do this probability calculation, you first have to do the math for a single trajectory (which you haven’t done). Then, to the probability calculation for multiple possible evolutionary trajectories, you have to compute the probability of at least one of these trajectories occurring. The Kishony experiment demonstrates this math very clearly:
At the end of the video, they illustrate the phylogenetic evolutionary trajectories for multiple different lineages and there is no reason to believe that each of these lineages is taking identical evolutionary trajectories. These different trajectories all have one thing in common, each node of these phylogenetics represents a colony of bacteria with about a billion members in each colony.
Do you think that extrapolating the mutation rate for a lineage that has a genetic disease is appropriate for a lineage that doesn’t have a genetic disease? Again, I don’t have a problem using the mutation rate you give, that doesn’t solve your mathematical problem.
You won’t get to the correct mathematics using the average values we’ve been using so far. You have to start computing the actual probabilities. And you don’t have any real, measurable, and repeatable experimental evidence to justify your claim.
Does it now? Let’s get back to the simple 2 agriculture mutations thought experiment. You’ve shown that with the mutation rate you use, that 9e7 replications (18e7 genome replications) that you will get every possible base substitution. That means that one of the descendants will one of the agriculture mutations (call it an A mutation) and some other descendent will have the second agriculture mutation (call it a B mutation). The rest of the population will have neither mutation (call those C variants). Now, I want to give you a chance that a recombination event will somehow get these two variants will mate to give a descendent with both the A and B mutations so let’s assume that the mutations don’t have to occur in the same gene otherwise this would require a chimeric recombination for there to be any chance. Now, show us the mathematic of the recombination of event for a single A variant and a single B variant meet and recombine in a lineage of 9e7-2 C variants that have neither mutation.
You’ve done the math for a single mutation. Using the mutation rate you give, 90,000,000 replications give every possible base substitution. If humans and chimpanzees arose from a common ancestor, shouldn’t the chimpanzee get the same mutations humans got?
It shouldn’t be very puzzling for you, after all, don’t people used the Jukes-Cantor, your F81, and a variety of other Markov models to try and make phylogenetic inferences for different species? And Markov chains have an entropy rate.
And why would you think that DNA evolution works differently for humans, chimpanzees, or bacteria? The reason why I push the discussion of evolution towards the Kishony and Lenski experiments (stuff) is that they are real, measurable, and repeatable examples of the evolutionary process. These experiments are demonstrating the laws of thermodynamics if you care to look and try to understand them. These same principles apply to the evolution of humans and chimps.
It makes a huge deal when you are talking about thousands or hundreds of thousands of generations. The two numbers you should consider are:
W=1.0000397289 for chimps gives a population of 300,000 after 300k generations
W=1.0000732564 for humans gives a population of 7 billion after 300k generations
And if W is less than 1, (for example 0.999), the population is going to be declining over generations. There is a context where small differences can make big deals, and this is one of them.
First of all, no one has claimed there were no beneficial mutations. The evidence for the existance beneficial mutations seems pretty clear.
Second, you keep coming back to “… different reproductive fitnesses as demonstrated by the 20,000 times differences in population size.” when the difference in fitness due to intelligence is more quality than quantity. This has been pointed out several times and you haven’t seemed to notice. There is a new thread discussing just this topic that might interest you:
Your most recent argument requires knowledge of just what those beneficial mutations are, and indeed the entire evolutionary sample space since the MRCA to do probability calculations. This is definitely far afield from your original claim.
I didn’t expect you to accept, which is why I presented as speculation. The point is the key ingredients for greater intelligence, tool use, and social behaviors were already in place at the time of the MRCA. Small changes in ability could lead to large changes in behaviors, utilization of tools, and cooperation with others. As I originally mentioned far up the page, these things are fitness multiers. It doesn’t take many multipliers to create large changes in fitness.
Adapting the environment is a game changer, and ought to be a discussion changer. It doesn’t make any sense to compare Chimp and Human fitness across different environments and population sizes. By counter-example: there are a lot more fish than humans (by orders of magnitude). Are fish more fit than humans?
Here you get it. This is what I’m getting at by synergy. We don’t need new mutations for each of those adaptive behaviors, we only need the facility to adapt our behaviors.
Again, I don’t think anyone is saying otherwise. How many potentially adaptive mutations are there? My educated guess is that nobody knows - some assumption would seem to be necessary before doing any probability calculations.
Yes, they do. If half of all mutations result in an increase in fitness that would drastically change the probabilities, wouldn’t it?
That won’t tell you the probability of getting an increase in fitness. To do that, you need to know what all of the trajectories are.
None of those bacteria are reproducing sexually which makes them a poor model for evolution in primates.
I think directly measuring the mutation rate by sequencing is the best method.
Do you have any of these values for the evolution of intelligence in humans? You keep claiming that it’s too improbable, and yet you haven’t produced a single calculation.
Please show that there are only 2 possible mutations that could increase intelligence in humans. Otherwise, your model is useless.
You are forgetting about selection. If both A and B are beneficial then each mutation will quickly increase its frequency making it much more likely that offspring will be born with both alleles.
No. You would get different mutations in each lineage. This is one of the most basic fundamentals of evolution.
Being real, measurable, and repeatable does not automatically make them relevant to every single population.
The fitness difference between humans and chimps is small (approximately three parts in one hundred thousand), but on longer time scales has amounted to a vastly huge greater big deal.
Okay, you can call it a vastly huge greater big deal if it makes you feel good. It’s still about three parts in one hundred thousand. And? Is this supposed to be an impossible fitness differential to evolve?
It occurs to me that given the substantial changes in behavior/adaptability between humans and pan, it’s not clear what constitutes a “neutral” change. Is it neutral to one or both extant species? to the MRCA??
I’m guessing evolutionary biologists must have a way to deal with this.
The E. coli bacteria in your own body vastly outnumber the current humans human population.
Does that mean they have had more beneficial mutations than than the human species has?
I believe I can summarize the argument here:
@swamidass: there are several patterns explained quantitatively if common descent is true and most mutations are largely neutral, and until a better model explains this without common descent, these patterns are very strong evidence for common descent.
@kleinman: But the assumptions are wrong: at least some of those mutations are not neutral. How did they arise?
@swamidass: I didn’t assume anything like that, and in fact the model only requires that most mutations are largely neutral.
Those are the numbers you should consider, particularly with respect to deciding whether 1.0000732564 is “vastly greater” than 1.0000397289.
It is. Since you now say the difference in reproductive fitness is “small”, why you haven’t you retracted your earlier claim that “humans have vastly greater reproductive fitness than chimpanzees”?
Not at all. I completely understood and agree with your post when you said the follow:
But I question a term you used in a response to Faizal_Ali where you wrote:
Just because most of the genome doesn’t code for proteins doesn’t make it junk. Where is the control system that turns on and off these protein-coding genes that cause a stem cell to differentiate into a fully formed properly functioning adult? I think it is primarily the non-coding portion of the genome that determines whether the zygote differentiates into a human or chimpanzee.
So you are claiming that humans and chimps just took different evolutionary trajectories where we got the intelligence and they got the strength? What if humans drive chimpanzees to extinction simply by taking their habitat and using it for farming?
And adapting to an environment is a game-changer, I’d like to see you explain how DNA evolution does that. Why don’t you start by explaining how Kishony’s bacteria adapt to the higher drug-concentration niches in his experiment. And I’ve heard that fish story many times how fish evolve into mammals. Have you ever thought about trying to explain how adaption works in the Lenski experiment? Do you ever think they will ever restart that experiment? I don’t think they will.
And which beneficial mutations confer on us humans the ability to do that adaptive behavior?
And here is where you miss my point. Evolutionary adaptation is mathematically predictable. T_aquaticus has figured out that the first beneficial mutation in an evolutionary trajectory takes 90,000,000 replications. He is balking at doing the math for the second beneficial mutation occurring on a member that has the first. He wants to try and invoke recombination or multiple evolutionary trajectories to address this issue but won’t try and do that math either.
No. Even if half of all mutations result in an increase in fitness, each mutation is part of its own particular evoltionary trajectory and each step on any of these evolutionary trajectories will take about 1/(mutation rate) replications to take the next step on that particular evolutionary trajectory. A good empirical example from the medical field of this is MRSA, drug-resistant staph aureus. There are lots of different variants of MRSA. The reason why this happened is bacteria can attain the population size and recovery rates necessary for multiple different variants and lineages to take their own particular evolutionary trajectory to drug resistance. Humans don’t have anywhere near the population size and recovery rates to take these kinds of evolutionary trajectories.
The environment tells which mutations give improvement in fitness. When sufficient number of replications have occurred so that every possible base substitution has occurred, the environment determines if any of these base substitutions gives improved fitness. Consider the Kishony experiment with a single drug. The colony has grown large enough so that every possible base substitution has occurred. If the drug used in the experiment is trimethoprim, that’s the variant with improved fitness and it can grow in the next higher drug concentration region. The variant with the Ciprofloxacin mutation doesn’t have improved fitness in that environment and its mutation is essentially neutral. However, if the drug used in the experiment is Ciprofloxacin, that variant has the beneficial mutation and the trimethoprim variant has the neutral mutation.
Now, when Kishony tried to run his experiment with both drugs, it didn’t work. The reason is that neither the trimethoprim variant or the Ciprofloxacin variant are beneficial in that environment. A variant with both the Ciprofloxacin mutation and the trimethoprim mutation are needed to give improved fitness. For a reasonable probability of that occurring, you need a colony size of about a trillion. His experiment doesn’t have sufficent carrying capacity to demonstrate that evolutionary process.
Since you can’t or won’t do the mathematics of recombination, perhaps you should think about HIV which does recombination. 3 drug therapy gives durable treatment. Why doesn’t recombination cause treatment failure? Why do combination herbicides suppress the evolution of herbicide resistance despite the fact that plants do sexual reproduction and recombination?
What I have is the real, measurable, and repeatable experimental evidence of how evolutionary processes work and a mathematical explanation of these processes. What this math and empirical evidence shows is that it takes about 1/(mutation rate) replications for each evolutionary step on an evolutionary trajectory. Using your mutation rate, you came up with 90,000,000 replications for the first evolutionary step. You won’t even try to do the math for the second step on an evolutionary trajectory. Instead, you want to say recombination solves this or there are multiple possible evolutionary trajectories without presenting any mathematical or empirical evidence to support your claim. Only under certain circumstances will recombination accelerate DNA evolution but that is only for a single evolutionary step. Once that recombination event occurs, that new variant must sample for new beneficial mutations for the next evolutionary step. If you want to see my calculations, you can find them here:
The basic science and mathematics of random mutation and natural selection
What must selection do to improve the probability of an A+B recombination event occurring? And once that A+B recombination event occurs, what must that new variant do to increase the probability of another beneficial mutation occurring on one of its members?
So quickly you forget, 90,000,000 replications gives every possible base substitution in the genome. Couldn’t the chimpanzee lineage do 90,000,000 replications?
Those experiments are relevant for demonstrating the laws of physics which govern evolutionary processes. Do you think the laws of physics apply differently for humans and chimps than they do for bacteria. If so, such as with recombination, explain the physical difference and show your math. Since you don’t like these experiments, let’s consider an evolutionary experiment that includes the effects of recombination:
The reason why recombination accelerates adaptation in the harsh environment is that the harsh environment is selecting out the less fit variants improving the probability of a recombination event with the more fit variants. So, for the agricultural mutation thought experiment with A and B agricultural more fit variants and C less fit non-agricultural variants, in the more harsh environment, the C variants are selected out increasing the frequency of the A and B variants which in turn improves the probability of an A+B recombination event. In the less harsh environment, the C variants are not selected out and are at much higher frequency than the A and B variants so an A+C or B+C recombination events are much higher probability. When that beneficial recombination event occurs, it only accelerates that one evolutionary step. But the A+B variant population must now do another 90,000,000 replications to get every possible base substitution so that A2 and B2 mutations can occur for the next evolutionary step on the evolutionary trajectory. In other words, recombination only has a chance of operating once the particular alleles are created and that requires DNA evolution by the Markov chain process. By the way, this is the same explanation for Darwin’s finches.
It makes a huge difference if you want to explain the genetic differences between humans and chimpanzees using neutral evolution.
You haven’t been following this discussion very well. I’m challenging the contention that the genetic differences between humans and chimpanzees can’t be calculated using the mathematics of neutral evolution. If one takes the position that neutral mutations do not change the reproductive fitness of that variant with respects to other variants, I asked, why are there 7,000,000,000 humans and only 300,000 chimps if the only genetic differences are neutral. I was asked to compute the reproductive fitness of humans and chimpanzees which are thought to have descended from a common ancestor. I did that calculation using Kimura’s definition but others claim that calculation has take into account the particular environment. So, I said, use the populations of Tanzania and Senegal (75,000,000 humans) where most chimpanzees live. So far, no one else in this discussion has computed their values for the reproductive fitness of humans and chimpanzees, perhaps you want to try? And are you now claiming that E. coli and humans evolved from a common ancestor? I’ve seen some pretty strange phylogenetic trees but that one takes the cake.
I’m not the one who made a video claiming that the genetic differences between humans and chimpanzees is explained by distance = rate x time. As far as I know, nobody knows which genetic difference are beneficial or neutral. And if only a tiny fraction of those mutational differences are beneficial, you don’t have sufficient population size to accomplish that evolutionary process (trajectory). Remember, you said there is 2% genetic difference between the human and chimpanzee genome, that’s 60,000,000 bases. If only 1% of those bases are selective, that’s 600,000 bases. If you have been following my discussion with T_aquaticus, his “back of the envelope” calculation for beneficial mutations requires 90,000,000 replications just for the first beneficial mutation(s) in an evolutionary trajectory.
You are making the false assumption that there is only one mutation at one base that will increase fitness.
This is a bit exaggerated, but let’s say that 50% of offspring in one generation carry a new mutation that increases their fitness. Let’s say the same is true for the next step, that 50% of offspring carrying the first mutation acquire a new mutation that pushes them along that evolutionary trajectory. It doesn’t take long to get those two mutations, now does it?
You haven’t demonstrated that the constrained evolutionary trajectories for drug resistance in bacteria are anything close to the constraints of the evolutionary trajectories that led to human adaptations.
This is a mistake you keep making. You find one type of adaptation in one species that is limited to one or just a few mutations, and then you assume that it is the same for all adaptations in all species. This simply isn’t true.
You don’t have to mutate every single base in order to get beneficial mutations in every single case.
Can you show us a single human adaptation that is analogous to the situation in the Kishony experiment? Can you show us the bases in the human genome?
HIV don’t reproduce sexually.
Do you understand how diploidy works? Do you understand how you carry DNA from both your mother and father?
You have a real, measurable, and repeatable experiment of how antibiotic resistance evolves in asexual organisms. You don’t have an experiment for how intelligence evolves in primates.
Selection would increase the number of people carrying A and B. As those numbers increase it increases the probability of an A mating with a B and having offspring who carry both A and B.
The mutations that happen are not the same as the mutations that get fixed or increase in frequency.
The laws of physics do not demand that there is just one possible mutation that will increase fitness in a primate lineage.
I seriously doubt that Joe Felsenstein, or anyone here, equates “non-coding” with “junk”.
Oh dear. Kleinman reveals, more clearly than ever before, that he doesn’t understand what junk DNA is or how we identify it. Can he be excused because just that misunderstanding is common among molecular biologists? Perhaps. But he still needs to be educated, and that isn’t something he’s willing to do.
Please quote where anyone in this discussion has said that the only genetic differences between us are neutral. I must have missed it.
But to answer your question, perhaps this will illustrate how such a thing could happen:
Imagine a population of, I dunno, let’s say squirrels live on a mountainside. One day there is a landslide, and the population is split into two. Over time, the genomes of the two populations diverge such that they are readily identifiable as two subspecies. However, there is no difference between the two in terms of how many adaptive mutations were fixed.
Then, one very sad day, another landslide occurs and almost completely wipes out one of the subgroups, leaving the other unaffected.
The population of one group is now much greater than that of the other. But there is no differences in terms of the number of beneficial mutations either has accrued.
Not really that complicated at all, is it?
Joe never claimed that. “Non-coding” DOES NOT EQUAL “junk.”
It’s in the non-coding, non-junk category. It’s still a tiny proportion of the genome.
“Non-coding” DOES NOT EQUAL “junk.” You are employing the straw-man fallacy again.