Natural selection and adaptation

Natural selection is the only known process that produces adaptations: the eye, the wing, the camouflage of a moth on a tree trunk, the antifreeze proteins of Arctic fish. It is a simple idea with far-reaching consequences, and it is also one of the most misunderstood ideas in science. This lesson sets out the logic of natural selection, how biologists measure fitness and selection, the different forms selection takes, and examples that have been watched happening in real populations.

The logic of natural selection

Darwin's argument can be stated as a short chain of observations and inferences. Each observation is easy to check; the conclusion follows almost automatically.

Two features of this logic are worth stressing. First, selection acts on individuals, but the evolutionary change appears in the population: no finch grew a deeper beak, but the population's average beak depth changed because deep-beaked birds left more offspring. Second, selection does not create the variation it acts on. Variation comes from mutation and the shuffling of genes in sexual reproduction (lesson 3), and arises without regard to what would be useful. Selection then sorts the variation that exists.

Fitness

In evolutionary biology, fitness has a precise meaning: an individual's contribution of offspring to the next generation, relative to others. It is not strength, speed or health in themselves. A strong animal that leaves no offspring has zero fitness; a modest one that raises many young has high fitness. Biologists usually express fitness as relative fitness, symbolized \(w\), with the most successful type set to 1.

How fast does selection work?

A simple model shows how an advantageous allele spreads. Consider a haploid organism such as a bacterium, or think of alleles directly. If an allele A has frequency \(p\) and gives a fitness advantage \(s\) over the alternative, each generation its frequency changes according to:

Even a small advantage spreads remarkably quickly by evolutionary standards. The code below follows an allele that starts rare, at 1 in 1,000, with advantages of 1, 5 and 10 per cent.

# Spread of an advantageous allele: p' = p(1+s)/(1+ps)
def generations_to(p0, s, target):
    p, g = p0, 0
    while p < target:
        p = p * (1 + s) / (1 + p * s)
        g += 1
    return g

for s in (0.01, 0.05, 0.10):
    g50 = generations_to(0.001, s, 0.5)
    g99 = generations_to(0.001, s, 0.99)
    print(f"s = {s:.2f}: from 0.1% to 50% in {g50:4d} generations, to 99% in {g99:4d}")

s = 0.01: from 0.1% to 50% in  695 generations, to 99% in 1156
s = 0.05: from 0.1% to 50% in  142 generations, to 99% in  236
s = 0.10: from 0.1% to 50% in   73 generations, to 99% in  121

With a 10 per cent advantage the allele goes from rare to nearly universal in about a hundred and twenty generations, a blink of an eye in geological time. Even a 1 per cent advantage completes the job in about 1,150 generations. This is why antibiotic resistance can appear within a few years in bacteria that divide many times a day, and why insect and weed populations become resistant to pesticides within a decade or two.

Forms of selection

For traits that vary continuously, such as body size or beak depth, selection can change a population in three main ways depending on which individuals do best.

Stabilizing selection is probably the most common form in a stable environment, which is why many species look much the same for long periods. Directional selection dominates when the environment changes. Disruptive selection is rarer, but it can be a first step toward the origin of new species (lesson 4).

Real examples of natural selection

Selection has been measured directly in wild and laboratory populations. Three cases are especially well documented.

Peppered moths in industrial England. Before the industrial revolution, most peppered moths in England were pale and speckled, camouflaged against lichen-covered bark. As soot from coal burning darkened trees in the 1800s, a dark (melanic) form became common near industrial cities, reaching over 90 per cent around Manchester. When clean-air laws reduced pollution after the 1950s, the pale form returned. Field experiments showed that birds ate more of the moths that stood out against the bark, and in 2016 researchers identified the mutation responsible: a transposable element inserted into a gene called cortex.

Galápagos finches. Since 1973, Peter and Rosemary Grant, evolutionary biologists at Princeton University, measured every medium ground finch on the small island of Daphne Major. In the 1977 drought, small soft seeds ran out and birds with deeper beaks survived better; the next generation's average beak depth was about 4 per cent larger. In later wet years, the advantage reversed. Selection can change direction as conditions change.

Antibiotic resistance. When a population of bacteria is exposed to an antibiotic, any cells carrying resistance genes survive and multiply while the rest die. Resistance genes may arise by mutation or be picked up from other bacteria by horizontal gene transfer (lesson 7). Overuse and misuse of antibiotics speed the process, which is why public health agencies promote antibiotic stewardship: using antibiotics only when needed and completing prescribed courses as directed.

In Canada, the Public Health Agency of Canada tracks antimicrobial resistance through the Canadian Antimicrobial Resistance Surveillance System, and the 2023 Pan-Canadian Action Plan on Antimicrobial Resistance sets national goals for stewardship in human and animal health.

import numpy as np
# Simulated antibiotic treatment: resistant cells survive, sensitive cells mostly die
rng = np.random.default_rng(42)
pop = 1_000_000
resistant_freq = 1e-5                      # 10 resistant cells per million to start
survive_sensitive, survive_resistant = 0.001, 0.9

for day in range(1, 6):
    resistant = rng.binomial(pop, resistant_freq)
    sensitive = pop - resistant
    r_after = rng.binomial(resistant, survive_resistant)
    s_after = rng.binomial(sensitive, survive_sensitive)
    resistant_freq = r_after / (r_after + s_after)
    print(f"day {day}: resistant fraction after treatment = {resistant_freq:.3f}")
    # survivors regrow to a million cells before the next dose

day 1: resistant fraction after treatment = 0.011
day 2: resistant fraction after treatment = 0.908
day 3: resistant fraction after treatment = 1.000
day 4: resistant fraction after treatment = 1.000
day 5: resistant fraction after treatment = 1.000

In this illustrative run, resistant cells go from 10 per million to about 1 per cent of the survivors after the first dose, and to essentially all of the population by the third. Real infections are more complicated, because the immune system also clears bacteria and doses vary, but the direction is the same: every exposure enriches the resistant fraction.

Artificial and sexual selection

Darwin opened the Origin with artificial selection: breeders choosing which animals and plants reproduce. Every breed of dog descends from wolves, and broccoli, cabbage, kale, cauliflower, kohlrabi and Brussels sprouts were all bred from one wild mustard, Brassica oleracea. Artificial selection shows how much heritable variation populations hold and how quickly selection can reshape them, though here humans rather than the environment decide who reproduces.

Sexual selection is selection for traits that improve mating success. It takes two forms: competition within one sex, usually males, such as antlers used in fights between male elk or moose; and choice by the other sex, usually females, such as the peacock's tail or the elaborate songs of many birds. Sexual selection can favour traits that are costly for survival, because the reproductive benefit outweighs the cost. It explains many differences between males and females, called sexual dimorphism.

Adaptation and its limits

An adaptation is a heritable trait that increases fitness in a particular environment and that evolved by natural selection for its current function. The streamlined body of a dolphin, the thick fur and small ears of an Arctic fox and the water-storing stems of cacti are adaptations. But not every trait is an adaptation, and selection cannot produce perfect organisms. Several constraints apply:

It is also important to avoid saying that organisms evolve a trait because they need it. Selection has no foresight. Bacteria do not become resistant because they want to survive; resistant variants already present simply survive and leave more descendants. Saying a trait evolved in order to do something is convenient shorthand, but the real process is the sorting of random variation by its consequences.