The Failures of Mathematical Anti-Evolutionism
Download 0.99 Mb. Pdf ko'rish
|
The Failures of Mathematical Anti-Evolutionism (Jason Rosenhouse) (z-lib.org)
AA
AA AA AB BB, then 7 of the 10 genes are As and 3 of the 10 are Bs. That makes 70% A s and 30% Bs. It is customary to refer to these percentages as “frequencies” and to represent them as decimals. In my example, the frequency of A is 0.7 and the frequency of B is 0.3. By writing things this way, we have the convenient fact that the frequencies add up to 1. We can also think of frequencies as probabilities, in the sense that if we reach into our bag of As and Bs and pull one out at random, the probability of pulling an A is 0.7, and the probability of pulling a B is 0.3. Return now to the general set up. We have the three types, AA, AB , and BB, and we will assume that A and B appear with frequencies p and q, respectively. (Of course, we know that q = 1 − p, but it will be simpler just to use the single letter q rather than the more complex 1 − p.) If we now let the animals do what they do, how will the three types be represented in the next generation? Our answer to this question will depend on the answers to several other questions: • Do the animals choose their mates at random with respect to A and B, or do they prefer to mate only with others of the same type? • Do all three types have the same chance of finding a mate, or does one of the types have an advantage over the others? • Do we want to allow for the possibility that this gene will mutate in the next generation? Are we allowing for possible migration into and out of the population? These factors would change the gene frequencies in the next generation. • Can we assume the population is very big, so that there is a large number of couplings in each generation? If we do not make this assumption, then there is a danger that the gene frequencies will change in the next generation just by chance. If that last point is unclear, think about flipping a coin. It would not be terribly surprising if we flip the coin three or four times and get all heads or all tails. In short runs, the frequencies of heads and tails might differ substantially from fifty-fifty. But in a long run these 5.3 the hardy–weinberg law 119 frequencies will eventually settle down to their expected values. In the biological context, in a small population, gene frequencies might change just by luck in the same way that we might toss three heads in a row. But in a large population this is much less of a factor. At this point we should recall what we said about mathematical modeling: we throw out most of the messy reality and then hope the bit that remains includes the really important stuff. So, let us assume that the animals do, indeed, choose their mates randomly, that they all have an equal chance of finding a mate, that there is no mutation or migration, and that the population is very large. What are the consequences of these assumptions? Each of the offspring in the next generation will inherit one copy of the gene from its mother and one from its father. Since each gene individually has a probability p of being A, and since the gene inherited from the mother is independent of the gene inherited from the father, we can multiply these together to get a probability of p 2 that the offspring is of type AA . By similar reasoning, the probability is q 2 that the offspring is of type BB. Working out the probability of getting type AB is a little trickier because there are two ways for this to happen. It might be that the offspring inherits A from the mother and B from the father. This will happen with probability pq. But it might happen instead that the off- spring gets B from the mother and A from the father, and this also happens with probability pq. Adding these together gives us 2pq, and that is the probability that the offspring is of type AB. Our model predicts that a randomly chosen offspring in the next generation will be type AA with probability p 2 , type BB with probability q 2 , and type AB with probability 2pq. But notice that this conclusion depended only on the values of p and q and did not depend on the proportions of the three types in the initial population. And since the values of p and q have not changed from the parent’s to the child’s generation, because no As or Bs were gained or lost in the transition, the representation of the three types will remain constant for all subsequent generations. 120 5 probability theory (For the sake of accuracy, let me mention that I have omitted a few technical details that very slightly affect this conclusion, but including these details would be more trouble than it is worth relative to my purposes in this section. I say a little more about this in Section 5.11.) This finding – that the three types will be represented with probabilities p 2 , q 2 , and 2pq in the next generation – is known as the Hardy–Weinberg law, after the mathematician (Hardy) and geneticist (Weinberg) who first published it. The law’s conclusion was based on such a simplistic model that you might think it could not possibly be useful for anything. But you would be wrong! There have been many studies of gene frequencies in natural populations, and with surprising regularity the findings are in accord with what the Hardy– Weinberg law tells us to expect. When this happens, we conclude that the assumptions underlying the model hold true in the population with regard to that gene. When the data is not in accord with the law, we conclude that one of the assumptions does not hold. Either way, we have learned something about the animals we are studying. We could make our model more realistic by introducing some further variables. In real populations there is immigration and emi- gration, and this affects the frequencies of A and B. These frequencies could also change as a result of mutation. Most importantly, it might be that there is a selective advantage to possessing either A or B, and when this is the case the Hardy–Weinberg frequencies certainly will not hold in the next generation. We could construct more elaborate models that take these variables into account, and population geneti- cists have been very assiduous at doing precisely that. Alas, since each of these variables is inevitably represented by a different letter in the ensuing equations, textbooks on population genetics quickly come to look like alphabet soup. The mathematical formalism is formidable, but for our purposes it will not be necessary to consider the details. Models of this sort are useful to biologists engaged in field work because they provide a starting point for research into the 5.4 the art of counting 121 genetic makeup and behavior of actual populations. They also have theoretical value, in that they can help reveal what is possible under different assumptions. For example, in the early twentieth century, many scientists questioned whether natural selection was a powerful enough force to account for large-scale evolutionary change. If a genetic variation gave one animal a tiny reproductive advantage over another, would selection really be able to cause its spread through the population? Mathematical models were able to resolve this question by showing that under a variety of plausible assumptions, even minuscule selective advantages would be sufficient to drive the gene to fixation (meaning, essentially, that every animal in the population comes to possess that gene). We should also be mindful of the domain of application for these models. Over a small number of generations, the variables represented in the model are likely to remain constant, and it is for this reason that population genetics is sometimes said to model short-term gene flow. However, as time passes environments and populations change, to the point where the models are no longer useful. That is why these models are not used for predicting the long-term course of evolution. Developing and testing these models is tedious, painstaking work, but it is illustrative of the care that goes into serious scientific research. Most of it is not glamorous, and most scientists will go their whole careers without revolutionizing their disciplines or addressing the public. These models also serve as a helpful counterpoint to the modeling-on-the-cheap approach taken by the anti-evolutionists. But we must attend to one more bit of table setting before coming to that. 5.4 the art of counting Counting is simple when we have a small number of objects laid out sequentially before us. We just point to the objects in turn and say, “One, two, three, …” We teach small children how to do that. But counting can become surprisingly difficult in more abstract settings. 122 5 probability theory The branch of mathematics devoted to counting is called “com- binatorics,” and we saw in Section 5.2 that it is closely related to probability. Going forward, we will need a few basic combinatorial principles. We have already helped ourselves to one such principle: If we have n ways to do one thing, and k ways to do another, then there are nk ways of doing both together. For example, since each die has 6 possible outcomes by itself, the total number of outcomes for two dice is 6 × 6 = 36. Had we rolled a six-sided die and a twenty-sided die at the same time, there would have been 6 × 20 = 120 possible outcomes. This formula has an interesting consequence. Suppose we ask for the number of ways of arranging the six letters d, a, r, w, i, n. One such arrangement is “darwin,” but there are many others: niwrad, dirwan, and awinrd, for example. If we listed all such arrangements, how long a list would we have? We can use our basic principle like this: Arranging the letters is a six-step process. There are six ways of carrying out the first step, since any of the six letters could appear first. Once we have chosen the first letter, any of the remaining five can be the second letter. So there are five ways of carrying out the second step. Continuing in this manner, the total number of ways of ordering the letters is: 6 × 5 × 4 × 3 × 2 × 1 = 720. There was nothing special about using six letters. If instead we had n distinct letters we would have multiplied all the numbers from n down to 1. This multiplication is written as n! and is called “n factorial.” The exclamation point indicates our surprise at how fast these numbers grow. For example, 10! is already more than three million. Let us try something more complex. We have seen that the number of ways of ordering n distinct letters is n!. What happens if the letters are not distinct? For example, how many ways can we order 5.4 the art of counting 123 the letters e, v, o, l, u, t, i, o, n? Obviously, evolution is one possible ordering, but, again, there are many others. To answer our question, suppose that we could distinguish the two occurrences of the letter o. We could imagine using a capital O in one place and a lowercase o in the other. Then we would have 9 different letters and the answer would just be 9!. Let us call this the “modified problem,” since we have modified the original problem to obtain this answer. We now ask the following question: How should we adjust the answer to the modified problem to account for the fact that in our original problem the two occurrences of o cannot be distinguished? The difference between the two problems is this: In the modi- fied problem, the strings evOlution and evolutiOn are counted as different, but in the original problem they are the same. That means every string in the original problem gets counted twice when we shift to the modified problem. Since the answer to the modified problem was 9!, the answer to the original problem must have been 9! /2. Now we move on to the final exam. How many ways can we arrange the letters in the word Mississippi? This is trickier than our other examples, but the same logic applies. If we modify the problem so that all 11 letters are dis- tinguishable, then the answer is 11!. Alas, the modified problem counts MississiPpi as different from MississipPi, whereas the original problem counts them as the same. Thus, just as in the previous example, we need to divide by 2. The modified problem also treats the four occurrences of s as different from one another. There are 4! = 24 ways of arranging these four symbols. That means that every sequence in the original problem gets counted 24 times in the modified problem, implying that after we divide by 2, we still have to divide by 24. We have to do this again to account for the 24 ways of ordering the four occurrences of i. Bringing everything together, and keeping in mind that 2! = 2 and 4! = 24, the answer to the original problem is |
Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©fayllar.org 2024
ma'muriyatiga murojaat qiling
ma'muriyatiga murojaat qiling