Natural selection is probably the most well known of the evolutionary mechanisms, and it is the one that most people think of when someone says, “evolution.” It is, however, often misunderstood, and people frequently fail to appreciate its complexity. Therefore, I am going to provide a brief introduction and overview of this fascinating mechanism as well as debunking several common misconceptions about it (note: sexual selection is best understood as a type of natural selection, but it is important and interesting enough that I will deal with it later in a separate post that is devoted to it).
How it works
The basic concept of evolution by natural selection is really quite simple. In most populations, some individuals will be able to survive better and produce more offspring than others. As a result, those individuals will pass on more genetic material to the next generation than other individuals do. Their offspring will, of course, posses the same alleles which allowed them to do well, so their offspring will also produce lots of offspring. Thus, with each generation, the alleles that allow individuals to produce lots of offspring become more abundant in the population (remember, evolution is just a change in allele frequencies).
To describe this in a more technical way, natural selection requires three conditions in order to work:
- There is variation for a trait
- That trait is heritable
- There is a selection differential for that trait
The first condition simply means that within a population, different individuals have different values for a give trait. For example, if the trait is height, then within a population, not all individuals will be the same height. Similarly, if the trait is color, then the population must contain alleles for at least two different colors. If this condition is not met, then natural selection simply cannot happen. Darwin correctly noted, however, that nearly all real populations have tremendous variation for most traits.
The second condition means that the trait can be passed from parents to offspring. Darwin did not understand how this happened, but he knew from captive breeding experiments that it did. You see, Darwin spent an incredible amount of time experimenting with artificial selection (especially with pigeons) and he noticed that this condition generally held true. For example, in a group of pigeons, he might find a few that had little tufts of feathers around the head, and if he bred them, their offspring would also have tufts. Thus, the trait got passed from the parents to the offspring. This is, of course, the same way that we have produced our modern dog breeds, crops, etc.
The final condition simply means that some individuals have a higher fitness than others. Importantly, “fitness” in this context does not mean physical prowess. Rather, in evolutionary terms, fitness refers to the number of genes that you are able to get into the next generation (this is typically thought of as the number of offspring who survive to a reproductive age, but the reality is more complicated because of the genes of your siblings, cousins, etc.). Therefore, if a trait does not affect your ability to produce offspring, it cannot be selected for.
All of that can be summarized by saying that natural selection occurs anytime that there is natural variation for a heritable trait, and that trait affects the number of genes that you can pass on to the next generation. Importantly, natural selection is a mathematical certainty. Anytime that all three of those conditions are met, natural selection will occur (though it can sometimes be trumped by other evolutionary mechanisms like genetic drift and gene flow). It’s also worth noting that even young earth creationists accept that natural selection occurs, the just place arbitrary and logically invalid limits on it (details here).
Simulating selection
To illustrate how this works, I wrote a simulator that I will use to model selection. For the sake of simplicity, the simulator models a trait that is controlled by two alleles and the trait is inherited via complete dominance. For the first example, let’s see what happens to populations of 100 individuals with starting allele frequencies of 1:1 (i.e., in each population there are just as many recessive alleles as there are dominant alleles). However, individuals with a dominant phenotype (i.e., individuals who have two dominant alleles or one dominant allele and one recessive allele) have a 100% chance of surviving to a reproductive age, whereas individuals who have two recessive alleles only have a 25% chance of surviving to a reproductive age. Now, let’s tell the simulator to make ten populations like that, and run the simulation for 100 generations. What you can see, is that the populations evolve extremely rapidly and by the 38th generation, the recessive allele is completely gone from all of the populations (Figure 1).
Figure 1: Results of simulations of natural selection. Each line represents the average of 10 simulations, and each simulation began with a population of 100 dominant alleles and 100 recessive alleles. Individuals with a dominant phenotype always had a 100% chance of surviving to a reproductive age, but the probability of individuals with a recessive phenotype surviving varied between sets of simulations.
We can also use the simulator to demonstrate the intuitively obvious fact that the strength of natural selection will depend on how much a trait affects fitness. For example, I ran the same simulations several more times, but I changed the chance of survival for the recessive individuals (50%, 75%, 90%, 95%). As you can see, the weaker the selection pressure, the longer natural selection takes. This should make good sense. When the selection pressure is very strong (i.e., the trait has a large effect on fitness), then selection can act very quickly, but when the selection pressure is weak (i.e., the trait has a small effect on fitness) then selection acts slowly because most of the individuals with the disadvantageous alleles are still able to survive and reproduce.
You should also note that selection can be a very powerful force. Even when most recessive individuals survived, selection was ultimately able to remove the disadvantageous alleles. In some cases, however, weak selection pressure can be trumped by genetic drift (more on that in a future post).
Figure 2: Mean results of 10 simulations in which dominant phenotypes had an 80% chance of surviving and recessive individuals had a 70% chance of survival.
To really drive home what is happening here, I also ran the simulator with a 100% survival probability for recessives. You’ll notice that the line for that simulation is the only one in which recessives are not removed from the population. This is because selection cannot act when traits don’t affect fitness. So, instead of selection, the changes in allele frequencies are from genetic drift, which is a random process (again, more on that in a future post).
Figure 3: Mean results of 10 simulations with a starting population of 150 dominant alleles and 50 recessive alleles. Dominant phenotypes had a 50% chance of survival, and recessives had a 100% chance.
Now, just in case someone takes issue with me setting the survival probabilities to 100% for the dominant phenotype, it is worth noting that selection will happen anytime that a trait affects fitness. For example, Figure 2 shows the mean results from a simulation in which dominant individuals had an 80% chance of survival and recessive individuals only had a 70% chance of survival.
Also, there is no reason why the dominant trait should be the beneficial one, and selection can act even when the beneficial alleles are rare in the initial population. For example, Figure 3 shows the mean results from a simulation in which 75% of the alleles in the initial population were dominant, but recessive individuals had a 100% survival probability, whereas dominant individuals only had a 50% chance of survival. Once again, evolution by natural selection occurred.
Natural selection causes populations to adapt
It’s also important to realize that natural selection always adapts populations. In other words, it makes them more well suited to their current environment/way of life. As evidence of that, in Figure 3, I included a line showing the percent of individuals that survived to a reproductive age in each generation, and you will notice that it increases as the percent of harmful alleles decreases. Thus, the populations are adapting. This contrasts with all of the other evolutionary mechanisms which can be either harmful or beneficial.
Natural selection removes variation
To me, this is one of the most fascinating aspects of selection: it reduces the genetic variation in a population. Look at Figure 1 again. Each population started with 50% dominant alleles and 50% recessive alleles, but by the end, each of the populations that were under selection had completely lost the recessive alleles. If you remember back to the start, however, selection requires variation in order to operate. Thus, if left to itself, selection will ultimately remove all of the variation from a population, at which point it will grind to a halt. It is, therefore, entirely reliant on mutations to provide it with new variation. Mutations are vitally important because they actually create new genetic material which selection can act on. Without them, selection would quickly run out of variation and cease to function (yes, beneficial mutations do exist). Gene flow can also play an important role in providing variation, but it can only move alleles from one population to another, rather than actually making novel alleles. Thus, mutations are still ultimately necessary to fuel selection.
Types of natural selection
There are three basic types of natural selection with regards to their outcome: directional, disruptive, and stabilizing. Before I explain these, however, it’s important to remember that most traits are polygenic, meaning that they are controlled by multiple genes, and most of those genes have multiple alleles. This results in a wide range of variety, and if you graph a quantitative trait for all of the individuals in a population, you will generally get a bell curve (see Figure 4). For example, if you graphed the heights of a human population, you would generally find that there are a few very tall individuals, a few very short individuals, and most people were in the middle. With that in mind, let’s talk about the three types of selection. To illustrate them, I am going to use the lengths of lizards in a fictional population (Figure 4).
Figure 4: The three types of natural selection. Arrows indicate the direction of selection.
Directional selection is exactly what it sounds like: selection moves the trait in a single direction by selecting for one of the extreme phenotypes. For example, let’s say that in generation 1 for the population in Figure 4 (top), small individuals tend to get eaten by predators, but large individuals can escape. This will result in large individuals producing a disproportionate number of offspring because they live longer. As a result, nature will select for large lizards and the average size of the population will increase over time.
Disruptive selection is very similar to directional selection, but instead of one extreme phenotype being selected, both extremes are selected. For example, let’s say that for generation 1 of the population in Figure 4 (middle), large individuals are once again able to outrun predators, but very small individuals are able to escape by hiding in small holes. Thus, it is the intermediate sized lizards which get eaten because they can neither fit down the holes nor outrun the predators. In that situation, selection would act on both the large lizards and the small lizards, resulting in the population evolving in two separate directions. This type of selection is very important because it often results in speciation (i.e., the formation of new species) especially when it occurs during sexual selection.
The final type of selection (stabilizing selection) is basically the opposite of disruptive selection. In this type of selection, it is the intermediate phenotype that is selected. For example, let’s imagine a situation for the first generation of the population in Figure 4 (bottom) in which there are intermediate sized holes to hide in, and very large lizards are too big to fit inside the holes and can’t run for long enough to escape predators. Also, very small lizards are not fast enough to get into the holes before being captured. Thus, large lizards get eaten because they have nowhere to hide, and small lizards get eaten because they are too slow, but intermediate lizards are fast enough to get to the holes and small enough to fit inside. Thus, selection will act against the two extremes. Importantly, for both of the other types, the mean value of the trait actually changes (in disruptive you essentially get two means); however, in stabilizing selection, the mean value stays the same, but variation is lost.
Misconceptions about natural selection
For the remainder of this post, I am going to talk about several common misconceptions about natural selection.
Misconception 1: Natural selection is random
Natural selection is not in any way a random process. In other words, which individuals survive and reproduce and which individuals die is not random. Rather, it is determined by their alleles. In other words, the individuals with the best alleles for the current environment survive and reproduce more readily than the individuals without those alleles. Thus, individuals are selected rather than being determined at random. To be clear, both mutations and genetic drift are random, but natural selection is not.
Misconception 2: Survival of the fittest
Describing selection as, “survival of the fittest” is really terrible for two reasons. First, “fitness” in evolution refers to the number of genes that you pass on to the next generation, but “survival of the fittest” is nearly always used to mean that the most physically fit individuals survive.
Second (and related to the first), selection is about reproduction not survival. Survival is only important in that it gives you more time to reproduce. If, for example, a group of individuals had alleles that made them immortal, but they never reproduced, selection could not act because those alleles would not get passed onto the next generation. Further, traits that have nothing to do with survival can still get selected. For example, a mutation that resulted in a bird laying 3 eggs instead of 2 could be selected because the individuals with that mutation would produce more offspring that their neighbors. Further, there are many species that produce thousands of offspring, but only live for a very short period of time. So yes, selection will increase the frequency of traits that help individuals to survive, but it only does that because surviving longer allows you to produce more offspring. If a trait allows individuals to survive longer, but they don’t use that time to produce offspring, then selection cannot act (note: for the sake of this post, I am essentially ignoring kin selection. Yes, selection could still act if the non-reproductive individuals spent the time helping their siblings, but that is a complexity that is far beyond the scope of this post).
Misconception 3: Something/someone is doing the selection
For reasons that I truly don’t understand, some people get confused when we talk about “nature selecting a trait.” They seem to think that this implies that there must be some entity or force driving the selection (i.e., God). This is a complete misunderstanding of how selection works. It’s a simple numbers game. Those who produce the most offspring get the most genes into the next generation and, therefore, are “selected.” There is no entity doing the selecting, it’s just probabilities and gene frequencies.
Misconception 4: Selection gives organisms what they need
People often seem to be under the impression that selection provides organisms with the traits that they need, but in reality, there is no relationship between what an organism needs and what selection gives it. Remember, selection is constrained by the genetic variation that is available to it, and that variation is produced by random mutations. Thus, although an individual may need a given trait, natural selection cannot do anything about it if the alleles for that trait aren’t available. Therefore, although selection does adapt populations to their environment, it does not give them what they need (more details here).
Misconception 5: Selection has a goal or direction
This is closely related to #4, and it basically proposes that selection is working towards some ultimate endpoint or goal (this is the misconception on which irreducible complexity is based). In reality, selection is blind. In other words, it simply adapts populations for their current environment, and it has no way of telling what will be beneficial in the future. Thus, if the environment changes, a trait which has been beneficial may suddenly become very harmful and selection will quickly reverse its direction (more details here).
Conclusion
Natural selection is simply the mechanism by which the individuals with the best alleles produce the most offspring and, therefore, pass on the most genetic material to the next generation. As a result, the alleles that allowed those individuals to do so well gradually increase in frequency. Selection is, however, constrained by the genetic information that is available to it, and it relies on mutations to provide new genetic material. Finally, it is not a random process, but it is also not a process that is being guided by an entity, nor does it move towards a particular endpoint or goal. Rather, it simply adapts populations to their current environments, and it is incapable of predicting future environments or giving organisms the traits that they need.
Other posts on evolutionary mechanisms:
The Logic of Science 