Heterozygosity
More actions
Heterozygosity is the condition of carrying two different alleles at a locus, one received from each parent. A plant with two copies of the same allele at a locus is homozygous there. For a population, heterozygosity is the proportion of plants that are heterozygous at a locus or that proportion averaged over many loci.[1] The observed heterozygosity of a sample is counted from its genotypes. The expected heterozygosity, also called gene diversity, is computed from its allele frequencies on the assumption that the population is in Hardy–Weinberg proportions.[2]
Cannabis is pollinated by the wind and is in most forms dioecious. Its plants are highly heterozygous. Many drug-type strains are propagated as clones to keep their genetic identity.[3] Across the whole genome, the share of heterozygous bases was 0.18–0.26 per cent in four cultivars sequenced in 2011[4] and at 1–2.5 per cent in the genomes of a pangenome published in 2025.[5] Most marker studies have found hemp more heterozygous than drug-type cannabis.[6] Clones of hybrid drug-type breeds carry more heterozygotes than random mating would produce. Inbred cultivars keep only a little residual heterozygosity.[7][8]
Heterozygotes at a single locus
At a locus with two alleles, A and a, a diploid plant has one of three genotypes: AA, Aa or aa. The heterozygote Aa passes A to half of its gametes and a to the other half.[1]
At the B locus, which carries CBDA synthase, each of the three genotypes gives a different chemotype in the model of inheritance that Aina and colleagues used. Plants homozygous for the functional allele B1 produce mostly cannabidiol (CBD), chemotype III. Plants homozygous for the non-functional allele B2 produce mostly tetrahydrocannabinol (THC), chemotype I. Heterozygotes produce both in similar amounts, chemotype II. Among about 1,400 feral plants from 90 populations in 12 states of the United States, 1,120 were B1B1, 91 were B2B2 and 208 were heterozygous; the two alleles had frequencies of 0.86 and 0.14.[9] Of four cultivars whose genomes were assembled in 2021, the one that produced both THC and CBD, FL48, was heterozygous at the THCA synthase locus and expressed both synthases. The three high-CBD cultivars expressed only CBDA synthase.[8]
A rare allele is found mostly in heterozygotes. Under random mating the fraction of the copies of an allele that sit in homozygotes equals the frequency of the allele, so an allele at a frequency of 0.0003 has 99.97 per cent of its copies in heterozygotes. This bears on how natural selection acts on rare recessive alleles.[10]
Expected heterozygosity
Under random mating, a locus whose alleles have frequencies p and 1 − p carries the genotypes AA, Aa and aa in the proportions p2, 2p(1 − p) and (1 − p)2. These are the Hardy–Weinberg proportions, found independently in 1908 by the mathematician G. H. Hardy and the obstetrician Wilhelm Weinberg.[1] The expected heterozygosity at such a locus is 2p(1 − p). It reaches its highest value, one half, when the two alleles are equally common and falls towards zero as either allele becomes rare. With more than two alleles, each homozygote AiAi has the expected frequency pi2 and each heterozygote AiAj the frequency 2pipj.[10] The expected heterozygosity is then one minus the sum of the squared allele frequencies. It can approach one at a locus with many alleles of similar frequency. High expected heterozygosity indicates many alleles at the locus. Low values can follow inbreeding or a population bottleneck.[2]
In a large population mating at random, Mendelian inheritance has no tendency to lose variation. Genotype frequencies disturbed without a change in allele frequency return to Hardy–Weinberg proportions in the next generation.[1] A population of finite size loses heterozygosity by genetic drift: in an idealised population of N plants, a fraction 1/(2N) of the heterozygosity is lost each generation on average.[11] New mutations add it back. In the infinite-alleles model, where every mutation creates a new allele, drift and mutation balance where the probability that two gene copies are identical by descent is about 1/(1 + 4Nu), u being the mutation rate. The equilibrium therefore depends on population size and mutation rate only through their product.[12] In a real population the census size N is replaced by the effective population size, Ne.[13]
Fewer or more heterozygotes than expected
The gap between observed and expected heterozygosity is measured by Wright's FIS, calculated as (HE − HO)/HE, where HO is the observed and HE the expected heterozygosity.[5] It is zero in Hardy–Weinberg proportions, positive where heterozygotes are fewer than expected and negative where they are more.
Inbreeding produces a deficit. The inbreeding coefficient f of a plant is the probability that its two gene copies at a locus are identical by descent, descended from the same gene copy in an ancestor. A plant with inbreeding coefficient f is expected to be heterozygous with probability 2p(1 − p)(1 − f).[14] Each generation of self-fertilisation halves the heterozygosity that a line holds.[15] Crossing inbred lines restores much of the heterozygosity of the population they came from.[16] The loss of fitness that can come with inbreeding, from the expression of harmful recessive alleles, is inbreeding depression.[17]
A sample that mixes plants from several populations, each of them in Hardy–Weinberg proportions, also shows a deficit. Its homozygotes exceed their expected frequency by the variance of the allele frequency among the populations, the Wahlund effect. If the populations have a mean allele frequency of 0.3 with a standard deviation of 0.1, homozygotes for that allele make up 0.10 of the pooled sample against an expected 0.09. Felsenstein counts this among the most important causes of departure from Hardy–Weinberg proportions in samples from nature.[18] Where plants are grouped into subpopulations, the departure is divided by level. FIS compares individuals with their own subpopulation, FST compares subpopulations with the whole and FIT compares individuals with the whole. The three are related by (1 − FIT) = (1 − FIS)(1 − FST). FST measures how far the subpopulations have diverged.[19] Its use in cannabis is described under population structure and genetic diversity.
An excess of heterozygotes gives a negative FIS. Several drug-type breeds of indica or mixed origin consisted of identical genotypes, clones, with an allelic richness of 2, an observed heterozygosity of 0.5 and FIS values reaching −1. Dufresnes and colleagues attributed this to the clonal propagation of hybrids between two parental strains.[7] At a locus where the heterozygote is fitter than either homozygote, a condition called overdominance, selection keeps both alleles in the population.[20]
Measuring heterozygosity
Estimates differ with the markers used and with what is counted. Hemp varieties typed at 13 microsatellite loci averaged 4.0 alleles per locus (scaled to eight plants) and had a mean observed heterozygosity of 0.59.[7] Feral plants typed at about 21,000 single-nucleotide polymorphisms (SNPs) had one of 0.20.[9] In studies that use genotyping by sequencing, the heterozygosity of a single plant is its number of heterozygous sites divided by its number of sites with a genotype call.[3][6] The count runs over the SNPs genotyped in the study, not over every base of the genome.
Sawler and colleagues expected some heterozygous genotypes in their data to be called homozygous, a systematic error that follows from the minimum sequencing depth of ten reads they required.[3] Filters also remove sites with too many heterozygotes. Sawler and colleagues dropped SNPs whose excess of heterozygotes departed from Hardy–Weinberg proportions. Grassa and colleagues kept only sites with an observed heterozygosity below 0.60.[3][8]
Whole-genome sequencing gives the share of heterozygous bases in one plant. Reads from the four genomes sequenced in 2011 were aligned to the Purple Kush draft. They gave 0.20 per cent for Purple Kush, 0.26 per cent for Chemdawg, 0.25 per cent for the hemp cultivar Finola and 0.18 per cent for the monoecious hemp USO-31.[4] Counts of k-mers gave 0.38 to 0.58 per cent for four cultivars assembled in 2021.[8] In the pangenome of 2025, built from 193 genomes of 144 plants, SNP heterozygosity ranged from 1 to 2.5 per cent. Counting structural variants and regions that could not be aligned, 20.6 per cent of the length of a genome was heterozygous on average.[5]
Heterozygosity complicates genome assembly. In the 2011 draft, scaffolds that were nearly identical over most of their length were taken to be the two copies of one heterozygous region, assembled separately, and only the larger was kept.[4] In 2021 the synthase genes of FL48 would not assemble into one contig, which Grassa and colleagues put down to heterozygosity in that region. The distribution of k-mer counts had one peak, with a slight shoulder, in the three highly inbred cultivars, and two peaks in the more heterozygous FL48.[8] The 2025 pangenome included 78 haplotype-resolved assemblies at chromosome scale, which give the two copies of each chromosome separately.[5]
Heterozygosity in cannabis
| Study | Plants | Markers | Measure | Value |
|---|---|---|---|---|
| van Bakel et al. 2011 | Purple Kush, Chemdawg, Finola, USO-31 (one plant each) | whole-genome short reads | heterozygous bases | 0.18–0.26% |
| Sawler et al. 2015 | 43 hemp, 81 drug-type | 14,031 SNPs | heterozygosity of each plant | higher in hemp |
| Dufresnes et al. 2017 | 24 hemp and 15 drug-type varieties, 1,324 plants | 13 microsatellites | observed, per variety | hemp 0.59; drug type 0.41 |
| Soorni et al. 2017 | 68 plants from Iran and Afghanistan | SNPs from genotyping by sequencing | heterozygosity of each plant | 0.15 (mean) |
| Grassa et al. 2021 | CBDRx, FL18, FL48, FL49 | k-mer counts | heterozygous bases | 0.38–0.58% |
| Aina et al. 2025 | 346 feral plants, United States | 21,037 SNPs | observed; expected | 0.20; 0.20 |
| Lynch et al. 2025 | 193 genomes of 144 plants | genome assemblies | SNP heterozygosity | 1–2.5% |
Mating system
The 2011 genomes had similar heterozygosity except USO-31, a monoecious hemp, whose lower value the authors presumed to come from self-pollination.[4] Because the plant is dioecious and pollinated by the wind, outcrossing is its only natural means of reproduction. Natural populations, and breeds made by mass selection, are usually made up of highly heterozygous genotypes that share a common gene pool.[21] The breeding system and its exceptions are described under outcrossing and sex determination in cannabis.
Hemp and drug-type cannabis
Sawler and colleagues found heterozygosity significantly higher in 43 hemp samples than in 81 drug-type samples. They offered two explanations, which may act together: a broader genetic base for hemp cultivars and more breeding among close relatives in drug-type cannabis.[3] In a survey of 1,324 plants at 13 microsatellite loci, hemp varieties had a mean observed heterozygosity of 0.59 and drug-type varieties one of 0.41. Every hemp variety had a positive FIS, 0.19 on average, which Dufresnes and colleagues linked to bottlenecks in current breeding. They connected the lower diversity of drug-type cannabis with clonal propagation and with the discarding of male plants.[7] Earlier estimates from SNPs, microsatellites and AFLPs had mostly put hemp above drug-type cannabis. One study of 195 samples and 2,894 SNPs found the reverse, which has been put down to the small number of hemp samples in it.[6]
The 68 plants from Iran and Afghanistan that Soorni and colleagues genotyped had a mean heterozygosity of 0.15, similar to the value Sawler and colleagues found for drug-type samples.[6] Feral plants of the United States had observed and expected heterozygosities of 0.20, which Aina and colleagues placed nearer the published range for hemp, 0.16 to 0.27, than that for drug-type cannabis, 0.12 to 0.14.[9]
Feral populations
Aina and colleagues sorted 346 feral plants from the United States into five genetic clusters.[9] Over all of them, FIS was −0.03. Three clusters, from along the Mississippi River and from the West North Central states, had observed heterozygosities of 0.24 and FIS values from −0.05 to −0.08. The authors read these negative values as a sign of random mating. The New York cluster of 14 plants had an observed heterozygosity of 0.09 and an FIS of 0.06; the Indiana cluster had 0.19 and 0.00.[9] The low value in New York is discussed under genetic drift.
Inbred lines and hybrids
To breed F1 hybrid varieties, inbred lines made by repeated selfing or mating of full siblings are crossed to give highly heterozygous hybrids, whose vigour is heterosis. Crossing two divergent inbred female lines, one of them masculinised to serve as the pollen parent, gives seed that is all female and of one highly heterozygous genotype.[21] For the mapping population used to assemble the CBDRx genome, the drug cultivar Skunk No. 1 and the fibre hemp Carmen were each bred by crossing siblings for five generations, to raise their homozygosity. A fifth-generation Skunk No. 1 female was crossed with a fifth-generation Carmen male. One F1 female was propagated as clones. The clones were induced to bear male flowers with silver nitrate and self-pollinated to give the F2.[8] The inbred cultivars CBDRx, FL18 and FL49 kept a residual heterozygosity of 0.38 to 0.54 per cent of their genomes.[8] The 2025 pangenome study described in detail EH23, an F1 hybrid of two parents that differ in phenotype and genotype.[5]
The pangenome also found a chromosomal inversion of 19.5 megabases on chromosome 1 that was mostly carried in heterozygous form. Of its 17 copies among the plants sampled, 15 were in heterozygotes and two in one homozygous plant. The region holds about 1,203 genes, among them PSEUDO RESPONSE REGULATOR 3, which has been linked to autoflowering. Lynch and colleagues suggested that the inversion may act as a supergene kept as a balanced polymorphism by associative overdominance.[5]
See also
References
- ↑ 1.0 1.1 1.2 1.3 Felsenstein, J. (December 2019). Theoretical evolutionary genetics. Seattle: Department of Genome Sciences, University of Washington. pp. 4–8.
- ↑ 2.0 2.1 Davinack, D. (2024). Molecular Ecology & Evolution: An Introduction. Norton, Massachusetts: Wheaton College. pp. 110–112.
- ↑ 3.0 3.1 3.2 3.3 3.4 Sawler, J.; Stout, J.M.; Gardner, K.M.; Hudson, D.; Vidmar, J.; Butler, L.; Page, J.E.; Myles, S. (2015). "The genetic structure of marijuana and hemp". PLoS ONE. 10 (8) e0133292. doi:10.1371/journal.pone.0133292.
- ↑ 4.0 4.1 4.2 4.3 van Bakel, H.; Stout, J.M.; Cote, A.G.; Tallon, C.M.; Sharpe, A.G.; Hughes, T.R.; Page, J.E. (2011). "The draft genome and transcriptome of Cannabis sativa". Genome Biology. 12 (10) R102. doi:10.1186/gb-2011-12-10-r102.
- ↑ 5.0 5.1 5.2 5.3 5.4 5.5 Lynch, R.C.; Padgitt-Cobb, L.K.; Garfinkel, A.R.; Knaus, B.J.; Hartwick, N.T.; et al. (2025). "Domesticated cannabinoid synthases amid a wild mosaic cannabis pangenome". Nature. 643 (8073): 1001–1010. doi:10.1038/s41586-025-09065-0. PMC 12286863.
- ↑ 6.0 6.1 6.2 6.3 Soorni, A.; Fatahi, R.; Haak, D.C.; Salami, S.A.; Bombarely, A. (2017). "Assessment of genetic diversity and population structure in Iranian Cannabis germplasm". Scientific Reports. 7 (1) 15668. doi:10.1038/s41598-017-15816-5.
- ↑ 7.0 7.1 7.2 7.3 Dufresnes, C.; Jan, C.; Bienert, F.; Goudet, J.; Fumagalli, L. (2017). "Broad-scale genetic diversity of Cannabis for forensic applications". PLoS ONE. 12 (1) e0170522. doi:10.1371/journal.pone.0170522.
- ↑ 8.0 8.1 8.2 8.3 8.4 8.5 8.6 Grassa, C.J.; Weiblen, G.D.; Wenger, J.P.; Dabney, C.; et al. (2021). "A new Cannabis genome assembly associates elevated cannabidiol (CBD) with hemp introgressed into marijuana". New Phytologist. 230 (4): 1665–1679. doi:10.1111/nph.17243. PMC 8248131.
- ↑ 9.0 9.1 9.2 9.3 9.4 Aina, Ademola; Wenger, Jonathan P.; Stanton, Eliot; Majumdar, Chandrani Gon; ElSohly, Mahmoud; Weiblen, George D.; Ellison, Shelby (1 July 2025). "Genetic diversity, population structure, and cannabinoid variation in feral Cannabis sativa germplasm from the United States". Scientific Reports. 15 (1) 20423. doi:10.1038/s41598-025-07912-8. PMC 12215616.
- ↑ 10.0 10.1 Felsenstein, J. (December 2019). Theoretical evolutionary genetics. Seattle: Department of Genome Sciences, University of Washington. pp. 10–11.
- ↑ Felsenstein, J. (December 2019). Theoretical evolutionary genetics. Seattle: Department of Genome Sciences, University of Washington. pp. 254–256.
- ↑ Felsenstein, J. (December 2019). Theoretical evolutionary genetics. Seattle: Department of Genome Sciences, University of Washington. pp. 284–285.
- ↑ Felsenstein, J. (December 2019). Theoretical evolutionary genetics. Seattle: Department of Genome Sciences, University of Washington. pp. 289–290.
- ↑ Felsenstein, J. (December 2019). Theoretical evolutionary genetics. Seattle: Department of Genome Sciences, University of Washington. p. 219.
- ↑ Felsenstein, J. (December 2019). Theoretical evolutionary genetics. Seattle: Department of Genome Sciences, University of Washington. pp. 231–233.
- ↑ Felsenstein, J. (December 2019). Theoretical evolutionary genetics. Seattle: Department of Genome Sciences, University of Washington. pp. 242–244.
- ↑ Davinack, D. (2024). Molecular Ecology & Evolution: An Introduction. Norton, Massachusetts: Wheaton College. pp. 172–174.
- ↑ Felsenstein, J. (December 2019). Theoretical evolutionary genetics. Seattle: Department of Genome Sciences, University of Washington. pp. 175–176.
- ↑ Felsenstein, J. (December 2019). Theoretical evolutionary genetics. Seattle: Department of Genome Sciences, University of Washington. pp. 275–277.
- ↑ Felsenstein, J. (December 2019). Theoretical evolutionary genetics. Seattle: Department of Genome Sciences, University of Washington. pp. 88–91.
- ↑ 21.0 21.1 Barcaccia, G.; Palumbo, F.; Scariolo, F.; Vannozzi, A.; Borin, M.; Bona, S. (2020). "Potentials and challenges of genomics for breeding cannabis cultivars". Frontiers in Plant Science. 11 573299. doi:10.3389/fpls.2020.573299.