Chapter 4 Drift at its strongest
Founder and bottleneck effects
When a population is founded by a few individuals, or crashes to a few and recovers, drift does in a generation what normally takes thousands.
The founder effect is the change in gene frequencies that happens when a new population is started by a small number of individuals: the founders carry only a sample of the source population’s variation, and a sample of a few is often unrepresentative. The bottleneck effect is the same thing happening to an existing population when its numbers crash. In both, rare variants are usually lost, a few become far more common than they were, and genetic diversity falls. Both are genetic drift, the subject of the previous chapter, compressed into a short window when the population is tiny. A one-page comparison of the two is here.
A boatload is a small sample
Suppose n people leave a large population in which a recessive disease allele has frequency p. Between them they carry 2n copies of the gene, and each is the disease allele with probability p, so the number they carry is Binomial(2n, p), the same distribution as one generation of drift in equation (3.1). The founders’ allele frequency has the source frequency as its average, but the spread is large when n is small.
The figure shows the two faces of the founder effect. Most small founding groups lose a rare allele altogether, which is why isolated populations often lack variants common on the mainland. A minority start with it at many times the source frequency, and if the new population then grows while staying isolated, that accident becomes permanent. Across many founding events, the average frequency is unchanged, as drift requires; what changes is that the populations become very different from the source and from each other. And the founder event is not the end: a population of a few dozen continues to drift strongly for generations afterwards, which can push the lucky allele further up or remove it.
The Old Order Amish
Ellis–van Creveld syndrome is a recessive form of short-limbed dwarfism, often with extra fingers and heart defects. In most of the world it occurs in 1 in 60,000 to 200,000 newborns; among the Old Order Amish of Lancaster County, Pennsylvania, it is much more common[1]. Victor McKusick and colleagues described it in the Amish in 1964, in the first of a series of papers on dwarfism in the Amish[2]. The Amish are a classic founder population: descended from eighteenth-century immigrants and marrying largely within their own communities, or demes, ever since. documented
The detail that makes the case is that Ellis–van Creveld is common in only one Amish community. Ridanpää, McKusick and colleagues contrast it with cartilage–hair hypoplasia, another recessive condition first described among the Amish, which is frequent in at least three separate Amish communities, indicating several carriers among the founders; Ellis–van Creveld, by contrast, is limited in its high frequency to a single deme[3]. That is the pattern the founder model predicts: one carrier who happened to be in one founding group. documented
Pingelap
Pingelap is a small atoll in Micronesia whose people have a high incidence of complete achromatopsia, a recessive condition in which the cone cells of the retina do not work: people see no colour at all and are dazzled by daylight. Elsewhere it is rare. In 2000 Olof Sundin and colleagues found the Pingelapese mutation, a single amino-acid change in the gene CNGB3, which codes for part of an ion channel the cones need to respond to light[4]. The population history assembled by Newton Morton and colleagues in the 1970s attributes the high frequency to a bottleneck: a typhoon in the late eighteenth century left only a small number of survivors on the atoll, among them at least one carrier of the allele[5][6]. reconstructed Whether the frequency also rose through drift in the generations after the typhoon, which is likely in a population of a few hundred, cannot be separated from the founding event itself.
Bottlenecks: diversity lost is not regained
A bottleneck is a founder event that happens in place. Its genetic cost can be calculated with the heterozygosity rule from the previous chapter, applied one generation at a time. If the population has Nt individuals in generation t, then
The product is dominated by the smallest factors. A population of 1,000 loses 0.05 % of its heterozygosity per generation; at 10 individuals it loses 5 % per generation, a hundred times faster. Five generations at size 10 cost more than two hundred generations at 1,000. That is why the effective size of a fluctuating population is the harmonic mean of its sizes, which is dragged down toward the smallest[7].
The flat line after the crash is the important part. Recovery in numbers slows the further loss of variation, but it does nothing to restore what was lost. New variation can come back only by mutation, at rates around 10−8 per base per generation, or by migration from somewhere else, and a population that has just recovered from near-extinction usually has nowhere else.
Northern elephant seals
Northern elephant seals were hunted for their oil through the nineteenth century, and taken by collectors, until by 1892 only about 20 animals may have survived. Protected since, they have recovered to more than 220,000[8]. In 2024 Rus Hoelzel and colleagues sequenced the genomes of 260 modern seals and of 8 museum specimens collected before the bottleneck. Average heterozygosity per genome fell from 0.00142 in the historical seals (five genomes with enough data) to 0.000176 in 180 modern adults[8]: the species kept about one eighth of its genetic variation, despite a thousandfold recovery in numbers. established
Can equation (4.1) produce an eightfold loss? Only with a severe bottleneck. A drop from H to 0.124 H needs Σ 1/2Nt ≈ 2.1, for example about 40 generations at an effective size of 10. Elephant seals live long and breed slowly, so 40 generations is far too many for a bottleneck that lasted decades. Two things narrow the gap. The seals were already declining for decades before 1892, so the bottleneck was longer than its lowest point. And elephant seals are extremely polygynous: a few dominant males father most pups, so by equation (3.4) the effective size was much smaller than the head count[8]. The historical estimate also rests on only five genomes with a wide spread (standard deviation 0.00092), so the true pre-bottleneck value is uncertain. A simple model gets the direction and the order of magnitude right, and the mismatch in detail points to exactly the factors, prolonged decline and skewed breeding, that the researchers emphasise. active
Hoelzel and colleagues also asked whether the bottleneck has a cost today. They report that inbreeding, a rise in the frequency of alleles with lost function, and distorted allele frequencies reduced the breeding success of modern males and females and the performance of adult females on their foraging migrations[8]. Some harmful alleles rose by drift while the population was too small for selection to hold them down. That interaction, drift raising alleles whose effects are too weak for selection to see, is the theme of Chapter 6. active
Founders and bottlenecks show drift making populations different from where they started. The next force does the opposite: migrants carry genes between populations and pull them back together.