I mentioned “directional selection” in an earlier post and someone asked about it, so I decided to take the opportunity to explain the process of natural selection in a bit more detail. I think there’s a tendency for people to have a very straightforward, simplified view of natural selection when they think about evolution. As with everything in biology, the real story is more nuanced and therefore more interesting. I’ll develop these ideas over a series of posts, starting with the different modes of selection in this post and moving on to different mechanisms in a future post. Understanding how the different aspects of natural selection interact is an important part of appreciating the complex and dynamic process of evolution, which is at the very core of biology and has generated all of the beautifully fascinating life we see around us.
Evolutionary biologists usually think in terms of a population, which is more or less a group of individuals that interbreed. Within a population, there can be a range of different values for any trait (like size or eye colour); some values are more or less frequent in a particular population. For example, dark hair and eyes are much more common in humans from the Mediterranean area than in those from northern Europe. A graph of how common a given trait is in a population is called the “distribution” of that trait; the famous “bell curve” is a distribution where fewer individuals have extremely high or low values and more individuals have intermediate values. Of course, the bell curve is just one possible example; in practice, traits may have a distribution with a different shape. One way to think about evolution is to look at how these distributions change over time.
There are basically three different “modes” of selection, which are descriptions of the different effects selection can have on the distribution of a trait. The modes describe the outcome of selection, while the mechanisms (which I’ll discuss in a later post) describe the process. We can often directly measure the change in distribution of a trait, making the mode of selection something we can observe and even quantify; on the other hand, we usually have to try to infer the mechanism (or combination of mechanisms). There isn’t necessarily a link between a mechanism and a mode; in principle, any mechanism of selection could result in any mode being observed. In other words, we can forget about mechanisms for the moment and just discuss the outcome, since the different outcomes can be the result of any mechanism.
Directional selection is the kind most people probably think of when they’re talking about evolution. This is when one particular value for a trait is more successful than others, so over time the average for the entire population shifts in that direction. For example, it might be that in a particular population faster individuals are more successful than slower ones (because they are better able to outrun predators or catch prey); over the course of generations, the average speed of the population would increase, since the faster individuals would have more offspring than their slower counterparts.
Stabilizing selection is selection which maintains the status quo because individuals with extreme values (high or low) for a trait do less well. For example, many animals face a trade-off between quantity and quality of offspring in a clutch — they can either have many poor quality offspring or a few high quality offspring. Because these two factors play off against each other, there is some optimum intermediate value. Stabilizing selection maintains clutch size at this intermediate value; individuals that have clutches which are too large or too small do less well, so the clutch size doesn’t change over generations. It’s important to realize that even though the situation appears static (i.e., we don’t see clutch size change over time), this is actually the output of a continuous, dynamic process of selection maintaining the trait at a particular value.
Disruptive selection is the opposite of of stabilizing selection. Instead of the individuals with the current average value for a trait doing better, individuals with extreme values do better which results in selection away from the status quo. The difference from directional selection is that the change isn’t in a particular direction, but is simply away from the current average value. Whereas directional selection will cause the distribution for a trait to shift either higher or lower, disruptive selection results in a distribution with more individuals at both higher and lower values, since they do better than individuals with intermediate values. The most famous example is probably Darwin’s finches, which have evolved different sizes and shapes of beaks adapted to different kinds of food; for example, birds with big or small beaks can eat big or small seeds, respectively, but intermediate sized beaks would be too big from small seeds and too weak for big seeds.
It’s important to realize that selection is likely to be acting on all kinds of different traits in different ways at the same time. Some of these traits might be somehow correlated with each other, leading to more complicated outcomes. For example, there might be directional selection to increase the biomass of seed produced by a plant and stabilizing selection to maintain the overall biomass. Since the two traits are correlated (seed biomass is a percentage of overall biomass), the outcome would depend on the relative strength of the two selective pressures and how flexible the development of the plant is (i.e., how easy it is to change what percentage of overall biomass goes towards seed production).
Another wrinkle is that strength and type of selection may not be static from one generation to the next. Of course, selection can change due to a variety of factors, but the one of the more interesting cases is frequency dependent selection, where the fitness of a particular value for a trait depends on how common it is in the population. If a bird uses to recognize its prey by visual cues, then prey individuals with an uncommon appearance will do better. For example, if most of the prey is red, then blue individuals will do better; however, when the blue individuals become more common, the birds will learn to look for them and the red (or yellow or green…) ones will do better. In this example, individuals with uncommon values do better, but it’s also possible for the reverse to happen — the fitness of a trait might increase as it becomes more common. A classic example of the trait’s fitness increasing with its frequency is warning colouration (like brightly coloured, poisonous frogs) — it only works if enough individuals are brightly coloured for the predator to learn the association. Once warning colouration has evolved, other species may evolve to mimic the warning colours without actually producing the toxin, but this only works as long as the mimics are relatively infrequent — another example of frequency dependent selection favouring a trait when it is uncommon.
I hope this post has managed to clarify some of the details about the process of natural selection. As I tried to point out towards the end, there are lots of interesting possibilities for interaction between the different modes of selection and the process is quite dynamic — selection is never static, but is shaped by the outcome from previous generations. These sorts of dynamic, layered interactions are what produce the richly disposed world in which we find ourselves, a world teeming with creatures that have evolved to make their living in just about every imaginable place and manner, shaping their world and themselves as they do so.