
calculate allele and genotype frequencies with the Hardy-Weinberg equations, test whether a population is in equilibrium, and model how genetic drift and bottlenecks change allele frequencies in small populations.
Instruction
Evolution as a change in allele frequencies. In a population, each gene has alleles at certain frequencies. If a gene has two alleles, A and a, their frequencies are p and q, with p + q = 1. Population geneticists define evolution as a change in allele frequencies from one generation to the next.
The Hardy-Weinberg equilibrium. If mating is random and nothing changes the frequencies, the genotype frequencies in the next generation are p2 (AA), 2pq (Aa) and q2 (aa), and p2 + 2pq + q2 = 1. The frequencies then stay the same generation after generation. This equilibrium holds only when five conditions are met: no mutation, no migration (gene flow), a very large population (no drift), random mating, and no natural selection. Real populations rarely meet all five, and that is what makes the model useful: when a population departs from equilibrium, one of the five is at work. It is a null model, like the 50 : 50 expectation in the pill bug lab.
Worked example 1: from a recessive phenotype. In a population, 4% of individuals show a recessive trait. So q2 = 0.04 and q = 0.2; then p = 0.8. Carriers (Aa) are 2pq = 2 × 0.8 × 0.2 = 0.32, or 32%. Notice that most copies of the recessive allele are hidden in carriers, not in individuals who show the trait.
Worked example 2: testing equilibrium. A sample of 1,000 individuals has 360 AA, 480 Aa and 160 aa. Count alleles: p = (2 × 360 + 480) ÷ 2,000 = 1,200 ÷ 2,000 = 0.6, so q = 0.4. Expected genotypes: 0.36 × 1,000 = 360, 0.48 × 1,000 = 480 and 0.16 × 1,000 = 160. They match exactly, so the sample is consistent with equilibrium. Now a sample of 100 with 50 AA, 20 Aa and 30 aa: p = (100 + 20) ÷ 200 = 0.6 again, so expected counts are 36, 48 and 16. χ2 = 196/36 + 784/48 + 196/16 ≈ 5.44 + 16.33 + 12.25 = 34.0. With 1 degree of freedom (three genotypes, minus one, minus one for estimating p), the critical value is 3.84, so this population is far from equilibrium. There are too few heterozygotes, which could mean non-random mating (for example, individuals mating with similar individuals) or a mixed sample from two populations.
Genetic drift and bottlenecks. In a small population, chance alone changes allele frequencies, a process called genetic drift. Flip a coin 10 times and 7 heads is common; flip it 1,000 times and 700 heads essentially never happens. In the same way, a small population’s allele frequencies wander, and alleles can be lost entirely. A standard result from population genetics is that each generation a population of N breeding individuals keeps, on average, a fraction 1 − 1/(2N) of its heterozygosity. With N = 20 that is 0.975 per generation, and after 10 generations 0.97510 ≈ 0.78: about a fifth of the variation lost to chance alone. With N = 100, five generations keep about 0.975 of it. A bottleneck, like the one Missouri’s and Illinois’s prairie-chickens passed through, cuts variation sharply, and it does not come back quickly even when the population grows again, because new variation arises only slowly by mutation.
Selection versus drift. 9-12.LS4.B.2 asks you to show statistically that advantageous traits increase in proportion. Drift makes that harder to see in small populations: chance can outweigh a small advantage. Large samples and repeated populations, as in your bead model, separate the steady push of selection from the random wander of drift.
📖 Read for this lesson
The reading opens over the lesson; close it to come back to exactly where you were. The checks below draw on it.
Lab: Bead model of genetic drift and a bottleneck
Question. How does population size affect how much allele frequencies change by chance, and what does a bottleneck do to genetic variation?
Materials
- Two opaque bags and at least 200 beads in each of two colors (representing alleles A and a); beans or buttons work too
- Cups or bowls to hold the new generation, and a data sheet or free spreadsheet
- A small cup (the “bottleneck” scoop) that holds about 10 beads
- Keep beads off the floor to prevent slips, and keep small beads away from young children.
- Do not put beads or beans in your mouth, and keep beans used for the model separate from food afterward.
Procedure
- Large population: put 100 beads of each color in bag 1 (p = 0.5). Without looking, draw 100 beads to form the next generation’s gene pool, count each color, and calculate p.
- Refill bag 1 with 200 beads in the new proportions (double each count) and repeat for five generations.
- Small population: put 10 beads of each color in bag 2. Draw 10 beads, count, calculate p, refill with 20 beads in the new proportions, and repeat for five generations.
- Repeat the small-population run twice more. Record whether either allele is ever lost (p = 0 or p = 1).
- Bottleneck: from a large, well-mixed bag with four colors of beads (four alleles), scoop about 10 beads and record how many different colors survive. Repeat five times.
- Graph p against generation for every run on the same axes.
Data
| Generation | Large pop. p | Small pop. run 1 p | Small pop. run 2 p | Small pop. run 3 p |
|---|---|---|---|---|
| 0 | 0.50 | 0.50 | 0.50 | 0.50 |
| 1 | ||||
| … | ||||
| 5 |
Bottleneck: number of allele colors surviving in each of five scoops ___.
Illustrative drift data (Option B). Constructed teaching numbers, not real draws. The large population drew 100 beads each generation; each small population drew 10.
| Generation | Large pop. p | Small pop. run 1 p | Small pop. run 2 p | Small pop. run 3 p |
|---|---|---|---|---|
| 0 | 0.50 | 0.50 | 0.50 | 0.50 |
| 1 | 0.52 | 0.60 | 0.40 | 0.50 |
| 2 | 0.49 | 0.50 | 0.30 | 0.60 |
| 3 | 0.51 | 0.70 | 0.20 | 0.40 |
| 4 | 0.53 | 0.80 | 0.10 | 0.50 |
| 5 | 0.50 | 0.90 | 0.00 | 0.30 |
Bottleneck, five scoops of about 10 beads from a well-mixed bag holding four allele colors: 4, 3, 4, 3 and 2 colors survived.
Analysis
- Compare how much p changed in the large and small populations. Use your graph and the range of p values as evidence.
- Did any small population lose an allele? Explain why this is more likely in small populations and why the allele cannot come back without mutation or migration.
- Use your bottleneck results to explain why the Illinois prairie-chickens lost alleles when their population crashed, and why adding birds from Kansas, Nebraska and other states could bring some of those alleles back.
- Identify one Hardy-Weinberg condition that your model breaks on purpose and one it keeps.
If you cannot do this lab in person: Choose one. All three earn full credit and end in the same graph of p against generation, the same comparison of large and small populations and the same bottleneck conclusion. Option A — the bead model. The procedure above. Any two colors of dried beans, pasta or buttons work, and two paper bags are enough; use four kinds for the bottleneck. Option B — the illustrative drift data in the data section above. No materials and no internet. Graph all four runs on one set of axes, give the range of p for each, and answer every analysis question, including which run lost an allele and why it cannot come back. Option C — coins or a spreadsheet. Generate your own random draws. With coins: for the small population flip 10 coins per generation and count heads as A; for the large population flip 100, or flip 10 coins ten times and pool them. With a free spreadsheet: fill 10 cells (and 100 cells) with =RAND(), count how many are below the current p, and use that count to set p for the next generation. Run five generations for one large and three small populations. For the bottleneck, number four alleles 1-4, draw 10 random numbers from 1-4 with a die you reroll on 5 and 6 (or with =RANDBETWEEN(1,4)), and count how many of the four survive; repeat five times.
Formative check
Work through these before moving on. They are not graded — they tell you whether the standard below has landed yet.
If 9% of a population shows a recessive trait, q = , p = and carriers are % of the population.
A sample has 245 AA, 210 Aa and 45 aa. What is p, the frequency of A?
Genetic drift changes allele frequencies more in small populations than in large ones.
Which condition is NOT required for Hardy-Weinberg equilibrium?
How confident are you that you can use Hardy-Weinberg equations and a drift model to decide whether and why a population is evolving?
Practice
Work these on paper or in your notebook, then open Check your answer. Aim for all of Fluency and Application; try at least one Challenge.
Printable version: this unit’s practice workbook (PDF)
Fluency
Build speed and accuracy with the core skill.
- In a population in Hardy-Weinberg equilibrium, 12.25% show a recessive phenotype. Find the allele frequencies and the proportion of carriers (heterozygotes).
Check your answer
Answer: q = 0.350, p = 0.650; carriers 2pq = 0.4550 (45.50%)q2 is the recessive phenotype frequency; p + q = 1; heterozygotes are 2pq. - In a population in Hardy-Weinberg equilibrium, 9.00% show a recessive phenotype. Find the allele frequencies and the proportion of carriers (heterozygotes).
Check your answer
Answer: q = 0.300, p = 0.700; carriers 2pq = 0.4200 (42.00%)q2 is the recessive phenotype frequency; p + q = 1; heterozygotes are 2pq. - In a population in Hardy-Weinberg equilibrium, 1.44% show a recessive phenotype. Find the allele frequencies and the proportion of carriers (heterozygotes).
Check your answer
Answer: q = 0.120, p = 0.880; carriers 2pq = 0.2112 (21.12%)q2 is the recessive phenotype frequency; p + q = 1; heterozygotes are 2pq. - In a population in Hardy-Weinberg equilibrium, 1.00% show a recessive phenotype. Find the allele frequencies and the proportion of carriers (heterozygotes).
Check your answer
Answer: q = 0.100, p = 0.900; carriers 2pq = 0.1800 (18.00%)q2 is the recessive phenotype frequency; p + q = 1; heterozygotes are 2pq.
Application
Use the skill in context. Show your reasoning.
- A population has 490 AA, 420 Aa and 90 aa individuals. Find p and q, and decide whether the genotypes fit Hardy-Weinberg proportions.
Check your answer
Answer: p = 0.7, q = 0.3; expected 490, 420, 90, an exact fitp = (980 + 420) ÷ 2,000 = 0.7. Expected: 0.49, 0.42, 0.09 of 1,000. - A sample has 30 AA, 10 Aa and 10 aa. Test whether it is in Hardy-Weinberg equilibrium (1 degree of freedom, critical value 3.84).
Check your answer
Answer: p = 0.7; expected 24.5, 21, 4.5; χ2 ≈ 13.7, so reject equilibrium. There are too few heterozygotes.(5.52/24.5) + (112/21) + (5.52/4.5) = 1.23 + 5.76 + 6.72 = 13.7. - A breeding population has 10 individuals. On average, what fraction of its heterozygosity is kept after one generation and after five generations of drift?
Check your answer
Answer: 0.95 after one; about 0.77 after five1 − 1/(2 × 10) = 0.95; 0.955 = 0.774.
Challenge
Stretch problems. Expect to think before you write.
- Compare the heterozygosity kept after 10 generations by a population of 50 breeders with one of 20 breeders. Use the result to explain why conservation plans for rare species try to keep breeding populations large.
Check your answer
Answer: About 0.90 for 50 breeders versus about 0.78 for 20. Smaller populations lose variation much faster by chance, reducing their ability to adapt to disease or environmental change, so managers aim to keep effective population sizes high and connect populations.(1 − 1/100)10 = 0.904; (1 − 1/40)10 = 0.776.
Review
Keep earlier skills sharp.
- (Biology A) An ear of 200 kernels from Pp × Pp has 140 purple and 60 yellow. Test the 3 : 1 model with chi-square.
Check your answer
Answer: Expected 150 and 50; χ2 ≈ 2.67, below 3.84, so consistent with 3 : 1100/150 + 100/50 = 0.67 + 2.00 = 2.67.
Is This Population in Equilibrium? Real Genotype Counts from the 1000 Genomes Project
Ensembl REST API, variation rs4988235, population genotype counts, 1000 Genomes Project phase 3 populations; retrieved Sept. 23, 2026 · SLS original, with genotype counts from the 1000 Genomes Project (phase 3, openly released data) as served by Ensembl
Why this reading. The lesson’s two worked samples were invented so the arithmetic would come out clean. The objective, though, says “test whether real genotype counts fit equilibrium,” and that needs real counts. Here they are: the genotypes of one well-studied DNA variant in several human populations sampled by the 1000 Genomes Project, retrieved from the Ensembl database with the query given above so you can repeat it. Two of the tests below come out the way the lesson’s second example did, and for the same reason.
The variant
rs4988235 is a single-base change near the gene for lactase, the enzyme that digests milk sugar; in European-ancestry populations its A allele is the one associated with keeping lactase activity into adulthood. The two alleles are G and A, so the three genotypes are GG, GA and AA. The 1000 Genomes Project sequenced volunteers from populations around the world and released the data openly; the counts below are people, not alleles.
Genotype counts in eight sampled populations
| Population (1000 Genomes code) | GG | GA | AA | n |
|---|---|---|---|---|
| Utah residents of northern and western European ancestry (CEU) | 7 | 38 | 54 | 99 |
| British in England and Scotland (GBR) | 9 | 33 | 49 | 91 |
| Finnish in Finland (FIN) | 14 | 53 | 32 | 99 |
| Iberian in Spain (IBS) | 34 | 48 | 25 | 107 |
| Toscani in Italy (TSI) | 90 | 15 | 2 | 107 |
| Mexican ancestry in Los Angeles (MXL) | 37 | 23 | 4 | 64 |
| Peruvian in Lima (PEL) | 69 | 14 | 2 | 85 |
| Han Chinese in Beijing (CHB) | 103 | 0 | 0 | 103 |
| All five European samples pooled (EUR) | 154 | 187 | 162 | 503 |
Worked test 1: the Utah sample (CEU)
- Allele frequency. Count A alleles: 2 × 54 + 38 = 146 of 2 × 99 = 198 alleles. p(A) = 146 ÷ 198 = 0.737; q(G) = 0.263.
- Expected genotypes. AA: 0.7372 × 99 = 53.8. GA: 2 × 0.737 × 0.263 × 99 = 38.3. GG: 0.2632 × 99 = 6.8.
- Chi-square. (54 − 53.8)2 ÷ 53.8 + (38 − 38.3)2 ÷ 38.3 + (7 − 6.8)2 ÷ 6.8 = 0.01. One degree of freedom (three genotypes, minus one, minus one for estimating p); critical value 3.84. The sample is consistent with Hardy-Weinberg equilibrium.
Worked test 2: the five European samples pooled (EUR)
- p(A) = (2 × 162 + 187) ÷ 1,006 = 511 ÷ 1,006 = 0.508; q = 0.492.
- Expected: AA 0.5082 × 503 = 129.8; GA 2 × 0.508 × 0.492 × 503 = 251.4; GG 0.4922 × 503 = 121.8.
- χ2 = (162 − 129.8)2 ÷ 129.8 + (187 − 251.4)2 ÷ 251.4 + (154 − 121.8)2 ÷ 121.8 = 8.0 + 16.5 + 8.5 = 33.0. Far above 3.84.
Look at what happened. Every one of the five European samples passes on its own (CEU 0.01, GBR 0.93, FIN 1.14, IBS 0.99, TSI 1.91; compute them). Pooled, they fail badly, with too few heterozygotes and too many of both homozygotes, exactly the pattern of the lesson’s second worked example. The lesson offered two explanations for that pattern: non-random mating, or “a mixed sample from two populations.” Here we know which it is. The A allele is at 0.74 in Utah and 0.09 in Tuscany; mixing a population that is mostly AA with one that is mostly GG produces a sample short of GA, because the two groups were not mating with each other. A Hardy-Weinberg test that fails is telling you one of the five conditions is broken; it does not tell you which, and the same signature can come from a sampling decision rather than from biology.
Two more to try
- Beijing (CHB): 103 GG, no other genotype. p(A) = 0. Expected GG = 103. χ2 = 0. Equilibrium is trivially satisfied when one allele is absent; there is nothing for drift or selection to act on at this position in this sample.
- Los Angeles (MXL): p(A) = (8 + 23) ÷ 128 = 0.242; expected 3.8, 23.5, 36.8; χ2 = 0.03. A population with recent ancestry from two continents can still sit in equilibrium if mating within it has been random for a few generations.
Worked checks from the lesson
- 9 percent show a recessive trait: q2 = 0.09, q = 0.3, p = 0.7, carriers 2 × 0.7 × 0.3 = 42 percent.
- 245 AA, 210 Aa, 45 aa: p = (490 + 210) ÷ 1,000 = 0.70.
- The lesson’s second sample (50, 20, 30): expected 36, 48, 16; χ2 = 5.44 + 16.33 + 12.25 = 34.0.
- Drift: N = 20 breeders keep 1 − 1 ÷ 40 = 0.975 of heterozygosity per generation; 0.97510 = 0.78.
Your turn
- Run the test for GBR, FIN, IBS and TSI and confirm the four chi-square values above.
- Pool CEU and TSI only (97 GG, 53 GA, 56 AA). Compute p, the expected counts and χ2. Explain the result using the two separate allele frequencies.
- State which Hardy-Weinberg condition the pooled test violates, and why the same table could not tell you whether selection was also acting.
Sources: Ensembl REST API, variation rs4988235 with population genotypes (1000 Genomes Project phase 3) (rest.ensembl.org); Ensembl, rs4988235 population genetics page (ensembl.org); International Genome Sample Resource, 1000 Genomes Project data (data use statement) (internationalgenome.org)