Quantitative trait locus
More actions
A quantitative trait locus (QTL) is a stretch of chromosome whose inherited variants are statistically associated with differences in a quantitative trait, a character that varies by degree among plants. It is inferred from genetic markers. Plants are genotyped and measured, and a region is called a QTL where marker genotype predicts the measurement beyond a set threshold.[1] A QTL report gives a position, an interval around it and an estimated effect in one population. The gene responsible cannot be settled from the association alone.[2]
In Cannabis sativa QTLs have been mapped in crosses between hemp and drug-type plants, in crosses within hemp and in panels of accessions, for chemotype, cannabinoid and terpene content, flowering time, sex expression, plant size, seed size, fibre quality and disease resistance. One QTL at the synthase genes accounted for 92 per cent of the variance in the ratio of tetrahydrocannabinol (THC) to cannabidiol (CBD) in the F2 of one such cross.[3] Landrace material has been mapped directly in a panel of Iranian populations and in a cross with a Syrian accession as one parent.[4][5]
A locus known by its marker
Quantitative traits generally rest on many genes, each of small effect. The environment and genotype-by-environment interaction act on them too, so plants do not fall into classes that reveal their genotype.[1] Mapping reaches the genes through linkage. Alleles close together on a chromosome are usually inherited together, so a marker near a causal locus carries information about it. Karl Sax recognised the value of such markers in the 1920s, for selecting traits that are hard to assay. Testing the idea across a whole genome had to wait about 60 years, until molecular markers arrived in the 1980s.[6][1]
Heritability estimates the genetic share of a trait's variance without placing any of it on a chromosome.[7] QTL mapping tries to place part of that share. It can only find what segregates in the population at hand. A cross between two inbred lines samples two alleles at each locus, so its QTLs are differences between those two parents.[1]
Populations for mapping
Each mapping population trades speed against the number of recombination events it holds. Recombination sets how finely a position can be resolved.[1]
| Population | How it is made | Gain | Cost |
|---|---|---|---|
| F2 | F1 selfed | Fast; three genotypes at a locus, so dominance can be estimated | One generation of effective recombination, so low resolution unless very large |
| Backcross | F1 crossed to one parent | Fast | Low resolution; two genotypes at a locus, so additive and dominance effects cannot both be analysed |
| Recombinant inbred lines | F1 selfed for generations until homozygous | More recombination; the same lines can be grown in repeated trials | Slow |
| Advanced intercross lines | F2 randomly intermated for further generations | Confidence intervals about one fifth the size of an F2's by the F10 | Many generations of mating |
| Doubled haploids | Chromosomes of a haploid doubled | Fully homozygous in one step; reproducible | Difficult and expensive; one round of recombination |
| Nested association mapping | Diverse lines crossed to one common parent, many lines per family | Many alleles and fine resolution; combines linkage and association | Costly in time, crosses and genotyping |
Dioecious plants cannot simply be selfed and hemp is highly heterozygous and outcrossing, so cannabis mappers have adapted these designs.[8] For the cross between the drug cultivar Skunk No. 1 and the hemp cultivar Carmen, each parent line went through five generations of full-sib mating. One F1 female was then propagated as clones. Silver nitrate brought fertile male flowers on some of them to pollinate the others.[9] A cross between a Syrian landrace accession and the Chinese hemp SI-1 used silver thiosulfate in the same way, which gave an all-female F2.[5] Carmagnola × USO31 used a monoecious F1 plant, and its F2 plants were cloned to replicate each genotype in the field.[10] Other studies map an F1 directly. Purple Kush × Finola was mapped in 99 F1 plants by tracing all four parental haplotypes, which gives a separate map for each parent.[11] An F1 saves time, though each genotype can be measured only once.[8]
Crossing over in cannabis is concentrated near the chromosome ends. In Purple Kush × Finola large blocks across the middle of the chromosomes typically lacked recombination events.[11] In two later families most chromosomes held regions of 40 megabases or more with under 0.5 centimorgans per megabase.[12]
Testing for an effect
Eric Lander and David Botstein introduced interval mapping in 1989, adapting the LOD score (logarithm of the odds) of human genetics to estimate the position and effect of a QTL from a linkage map.[13] Later methods handle several QTLs at once.
| Method | What it does | What limits it |
|---|---|---|
| Single-marker analysis | Compares trait means among the genotype classes of each marker | Effect size and distance from the marker are confounded |
| Simple interval mapping | Tests positions between markers by maximum likelihood or regression; estimates position and effect | Closely linked QTLs are hard to separate, which can produce false QTLs or hide real ones |
| Composite interval mapping | Interval mapping with markers outside the interval as cofactors, to control for QTLs elsewhere[14] | Linked QTLs with opposite effects can cancel out |
| Multiple interval mapping | Fits several intervals at the same time and can test epistasis | Heavy computation; choosing the model is difficult |
| Bulk segregant analysis | Compares allele frequencies between pooled plants from the two extremes of a trait, so only the pools are sequenced[15] | No individual genotypes; suited to loci of large effect[16] |
A QTL is declared where the LOD score passes a set threshold, commonly 2 to 3.[1] Several cannabis mapping papers set the value by permuting the data 1,000 times.[9][10][5] The scope of the test matters. In the first Skunk No. 1 × Carmen map a putative QTL for total cannabinoid content had a LOD of 3.30, above the threshold of 2.18 for its own linkage group and below the threshold of 3.42 for the whole map.[9]
Position is reported as an interval. The Carmagnola × USO31 cross gave intervals of 1.5 LOD. Among its 34 agronomic QTLs only one, for stem biomass, had an interval as narrow as about 50,000 base pairs.[10] In the Autoflower1 and Early1 scans peaks also appeared on chromosomes 8 and X, which their authors took to be artefacts of mapping or of the reference genome.[16] Realigning the Early1 data to newer assemblies narrowed a scatter of peaks spread over 40 megabases to about 2 megabases.[12]
Association mapping
A genome-wide association study looks for QTLs in a panel of existing lines without making a cross. Linkage mapping depends on the linkage disequilibrium that a recent cross creates, while association mapping uses what remains of older disequilibrium after many generations of recombination. It samples many alleles at a locus and resolves positions more finely, but it needs far more markers. Population structure can produce false associations. Testing thousands of markers produces others by chance, so analyses add covariates for structure and correct the significance threshold.[1]
Disequilibrium decays at different rates in different cannabis panels. In the Iranian panel the squared correlation between markers fell to half its maximum within 300 base pairs. In commercial Canadian drug-type cultivars the decay distance was 22.6 to 89 kilobases.[4] Petit and colleagues counted as one QTL a cluster of collinear significant markers, represented by the one explaining the most variance.[17]
What the figures measure
The share of variance a QTL explains is an estimate from one sample and tends to run high. Only effects that pass the threshold are reported, so the reported effects come from a truncated distribution and are biased upward. In a simulation by William Beavis, the variance attributed to correctly identified QTLs was greatly overestimated with 100 progeny, slightly with 500 and fairly close to the true value with 1,000, a result since called the Beavis effect.[18] Small effects go the other way and are missed. In Carmagnola × USO31 survival of the clones in the field was low and most agronomic models summed to about 50 per cent of the variance, which the authors took to suggest further small QTLs beyond the experiment's power.[10]
In Skunk No. 1 × Carmen 62 plants were enough for the ratio of THC to CBD, a result the authors attributed to the extreme difference between the parents, while the same plants could not settle total cannabinoid content.[9]
An effect can also change with the environment. QTL-by-environment models test for this directly.[19] The 123-accession hemp panel was grown at three sites in Europe: of 16 fibre-quality QTLs shared across sites, eight appeared at all three and eight at two.[20] Estimates of QTL effects for complex traits are often inconsistent. Far fewer mapped QTLs have been exploited in breeding than have been reported.[21]
The chromosome holding the synthase genes is 6 in the Purple Kush × Finola numbering and 7 in the CBDRx assembly.[2] The BKR region is chromosome 9 on the Purple Kush assembly and chromosome 4 on CBDRx.[4]
Measured in cannabis
| Population | Type | Size | Markers | Measured |
|---|---|---|---|---|
| Skunk No. 1 × Carmen | F2 of inbred parents | 62, then 96[3] | 121, then 1,235 | Cannabinoids[9] |
| Purple Kush × Finola | F1 | 99 | Haplotype blocks from resequencing | Genetic map[11] |
| Carmagnola × USO31 | F2, cloned | 372 lines | 1,817 SNPs | Eight agronomic traits, 17 cannabinoids and terpenes[10] |
| IPK_CAN_57 × SI-1 | F2 | 222 | 455 SNPs | Seed and plant traits[5] |
| Carmagnola × Carmagnola (two), Uso 31 × Fedora 17 | F1 | AFLPs | Sex expression[22] | |
| FL 58 × TJ's CBD | F2 | 706 | SNP array | Powdery mildew[23] |
| Bubble Kush × Aquawoman | F2 | 300 | Bulks of 50, then 15 KASP markers | Flowering time[24] |
| Felina 32 × FINOLA | F2 and F3 | Bulks | Flowering time[25] | |
| Jindao-15 × Fire No. 1 | F1 | 305 monoecious plants | Bulks of 30 | Fibre content[8] |
| W03 × AC, N88 × AC | F1 | Bulks of 25 | Powdery mildew[15] | |
| European hemp panel | Panel | 123 accessions | More than 600,000 SNPs | Flowering, sex, fibre quality[17] |
| Canadian drug-type panel | Panel | 174 and 176 accessions | About 282,000 SNPs | Cannabinoids; agronomic traits[2][26] |
| Ecofibre collection | Pools of extremes | 711 plants, 72 accessions | Resequencing | Cannabinoid side chain[27] |
| Iranian landrace panel | Panel of pools | 145 pools, 25 populations | 233,624 variants | 42 traits[4] |
Skunk No. 1 × Carmen
The F2 of this cross segregated 127 drug-type, 264 intermediate and 105 hemp-type plants. On a map of 62 plants the QTL for the THC to CBD ratio reached a LOD of 40.44 on one linkage group, the group to which both synthase genes mapped. The authors read it as separate linked loci, not alleles of one gene.[9] The inheritance behind it is the subject of Genetic traits. A later map from 96 F2 females, with marker bins from whole-genome sequencing, placed the QTL on the synthase arrays of chromosome 7. Six QTLs on five chromosomes together explained 53 per cent of the variance in total cannabinoid content. The arrays themselves did not reach significance for it (LOD 3.02 against a threshold of 4).[3]
Carmagnola × USO31
This cross joined an Italian dioecious fibre cultivar with a Ukrainian monoecious dual-purpose one. It was grown in one field at Fort Collins, Colorado, in 2017 and gave 34 agronomic and 35 biochemical QTLs. The largest single effects were 34 per cent of the variance in days to maturity, 31 per cent in stem biomass and 45 per cent in cannabidiolic acid content. Agronomic and biochemical QTLs did not share locations. The traits of the two classes were barely correlated. In 2021 no other whole-genome mapping population existed to show whether that separation holds outside the cross.[10]
IPK_CAN_57 × SI-1
The parents were chosen from 84 accessions for their contrast: one female plant of IPK_CAN_57, a small Syrian landrace accession with very small seed, and one male plant of SI-1, a tall Chinese hemp with large seed. The F2 was grown under controlled conditions. Among 53 QTLs, 25 explained more than 10 per cent of their trait's variance, among them 47 per cent for plant height in a QTL spanning 1 centimorgan and 49 per cent for seed density. Seed traits were not correlated with plant traits and their main QTLs lay on other chromosomes.[5] In Carmagnola × USO31 seed mass was correlated with plant size and shared its QTL clusters. Manansala-Siazon and colleagues read the two results as showing that the control of seed traits depends on genetic background or growing conditions.[5][10]
Loci of large effect from pools
Pooled scans placed Autoflower1 on chromosome 1 in the F2 family GVA-H-20-1080 and Early1 on the same chromosome in the cultivar Umpqua.[16] Autoflower2, on chromosome 8, was the only QTL found in both the F2 and the F3 of Felina 32 × FINOLA.[25] In Bubble Kush × Aquawoman a QTL on the X chromosome was narrowed from 2.2 megabases to 637 kilobases and explained 28.5 per cent of the variation in flowering time.[24] Pools of extreme plants, drawn from 711 of the Ecofibre collection, recovered the known CBDA synthase region as a control and pointed to a candidate gene, BKR, for the length of the cannabinoid side chain.[27] PM2, a dominant locus for resistance to powdery mildew, mapped to chromosome 9 in two F1 families.[15]
Canadian drug-type panel
In 174 accessions from the legal Canadian market, 33 markers were associated with 11 cannabinoid traits, 19 of them above 10 per cent of the variance and one as high as 96 per cent, for cannabigerolic acid. A haplotype of about 60 megabases on chromosome 7 was carried by most THC-dominant accessions. The panel held 152 THC-dominant, 17 intermediate and five CBD-dominant accessions. The authors warned that this imbalance may have amplified the haplotype's association and reduced their power at the synthase genes, which gave no clear signal.[2]
Iranian landrace panel
Seed from 25 native populations, bought in local markets or from farmers, was grown in one greenhouse complex at Mashhad. Plants of one sex in each block were pooled for genotyping, which gave 145 samples. Association mapping found 91 loci for 40 traits, 15 of them associated with several traits. A marker on chromosome 5, with a minor allele at 5 per cent, explained 94 per cent of the variance in total THC potential. Chromosome 7 gave only minor signals. Babaei and Torkamaneh took the main loci on chromosomes 5, 3 and 1 to be fixed in modern cultivars and polymorphic in landraces.[4] These results describe the Iranian populations in that greenhouse.
From a locus to a gene
A QTL interval usually holds many genes. The step to one of them takes separate evidence. In Carmagnola × USO31 the alleles of olivetol synthase differed by nine amino acids. Expressed in yeast, the Carmagnola allele made more olivetol, which fits the effect of the QTL. A second candidate in the same cross, TINY, remained a hypothesis.[10] In Bubble Kush × Aquawoman the 637-kilobase interval held 14 genes, one with a frameshift in the day-neutral parent. Overexpressed in rice, it delayed flowering from about 100 days to between 115 and 120.[24] In FL 58 × TJ's CBD the marker most associated with resistance lay near CsMLO1, which carries a 6.8-kilobase insertion in the resistant parent.[23]
Markers can be used in breeding before any gene is known, through marker-assisted selection.[1] Assays for Early1 matched phenotype in all 175 Umpqua plants tested, while markers for Autoflower1 did not predict day-length insensitivity across all the germplasm screened.[16]
See also
References
- ↑ 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 Lübberstedt, T.; Beavis, W.; Suza, W. (2023). "Cluster Analysis, Association, and QTL Mapping". In Suza, W.; Lamkey, K. (eds.). Molecular Plant Breeding. Iowa State University Digital Press. doi:10.31274/isudp.2023.133.
- ↑ 2.0 2.1 2.2 2.3 de Ronne, M.; Torkamaneh, D. (2025). "Discovery of major QTL and a massive haplotype associated with cannabinoid biosynthesis in drug-type Cannabis". The Plant Genome. 18 (2) e70031. doi:10.1002/tpg2.70031. PMC 12104491.
- ↑ 3.0 3.1 3.2 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.
- ↑ 4.0 4.1 4.2 4.3 4.4 Babaei, M.; Torkamaneh, D. (2026). "Genetic architecture of phenological, morphological, and phytochemical traits in Cannabis landraces". The Plant Genome. 19 (2) e70243. doi:10.1002/tpg2.70243.
- ↑ 5.0 5.1 5.2 5.3 5.4 5.5 Manansala-Siazon, S.E.; Siazon, P.M.; Tandayu, E.; Garcia-de Heer, L.; Burn, A.; et al. (2025). "Seed the difference: QTL mapping reveals several major loci for seed size in Cannabis sativa L." Plants. 14 (24) 3853. doi:10.3390/plants14243853. PMC 12737016.
- ↑ Beavis, W.; Mahama, A.A.; Suza, W. (2023). "Linkage". In Suza, W.; Lamkey, K. (eds.). Quantitative Genetics for Plant Breeding. Iowa State University Digital Press. doi:10.31274/isdup.2023.146.
- ↑ Beavis, W.; Lamkey, K.; Espinosa, K.; Mahama, A.A.; Suza, W. (2023). "Components of Variance". In Suza, W.; Lamkey, K. (eds.). Quantitative Genetics for Plant Breeding. Iowa State University Digital Press. doi:10.31274/isdup.2023.146.
- ↑ 8.0 8.1 8.2 Zhao, Y.; Sun, Y.; Cao, K.; Zhang, X.; Bian, J.; et al. (2022). "Combined use of specific length amplified fragment sequencing (SLAF-seq) and bulked segregant analysis (BSA) for rapid identification of genes influencing fiber content of hemp (Cannabis sativa L.)". BMC Plant Biology. 22 (1) 250. doi:10.1186/s12870-022-03594-w. PMC 9123736.
- ↑ 9.0 9.1 9.2 9.3 9.4 9.5 Weiblen, G.D.; Wenger, J.P.; Craft, K.J.; ElSohly, M.A.; Mehmedic, Z.; et al. (2015). "Gene duplication and divergence affecting drug content in Cannabis sativa". New Phytologist. 208 (4): 1241–1250. doi:10.1111/nph.13562.
- ↑ 10.0 10.1 10.2 10.3 10.4 10.5 10.6 10.7 Woods, P.; Campbell, B.J.; Nicodemus, T.J.; Cahoon, E.B.; Mullen, J.L.; McKay, J.K. (2021). "Quantitative trait loci controlling agronomic and biochemical traits in Cannabis sativa". Genetics. 219 (2) iyab099. doi:10.1093/genetics/iyab099.
- ↑ 11.0 11.1 11.2 Laverty, K.U.; Stout, J.M.; Sullivan, M.J.; Shah, H.; Gill, N.; Holbrook, L.; Deikus, G.; Sebra, R.; Hughes, T.R.; Page, J.E.; van Bakel, H. (2019). "A physical and genetic map of Cannabis sativa identifies extensive rearrangements at the THC/CBD acid synthase loci". Genome Research. 29 (1): 146–156. doi:10.1101/gr.242594.118. PMC 6314170.
- ↑ 12.0 12.1 Stack, G.M.; Quade, M.A.; Wilkerson, D.G.; Monserrate, L.A.; Bentz, P.C.; et al. (2025). "Comparison of recombination rate, reference bias, and unique pangenomic haplotypes in Cannabis sativa using seven de novo genome assemblies". International Journal of Molecular Sciences. 26 (3) 1165. doi:10.3390/ijms26031165. PMC 11818205.
- ↑ Lander, E.S.; Botstein, D. (1989). "Mapping mendelian factors underlying quantitative traits using RFLP linkage maps". Genetics. 121 (1): 185–199. doi:10.1093/genetics/121.1.185. PMC 1203601.
- ↑ Zeng, Z.B. (1994). "Precision mapping of quantitative trait loci". Genetics. 136 (4): 1457–1468. doi:10.1093/genetics/136.4.1457. PMC 1205924.
- ↑ 15.0 15.1 15.2 Seifi, S.; Leckie, K.M.; Giles, I.; O'Brien, T.; MacKenzie, J.O.; et al. (2025). "Mapping and characterization of a novel powdery mildew resistance locus (PM2) in Cannabis sativa L." Frontiers in Plant Science. 16 1543229. doi:10.3389/fpls.2025.1543229. PMC 11966446.
- ↑ 16.0 16.1 16.2 16.3 Toth, J.A.; Stack, G.M.; Carlson, C.H.; Smart, L.B. (2022). "Identification and mapping of major-effect flowering time loci Autoflower1 and Early1 in Cannabis sativa L." Frontiers in Plant Science. 13 991680. doi:10.3389/fpls.2022.991680.
- ↑ 17.0 17.1 Petit, J.; Salentijn, E.M.J.; Paulo, M.-J.; Denneboom, C.; Trindade, L.M. (2020). "Genetic architecture of flowering time and sex determination in hemp (Cannabis sativa L.): a genome-wide association study". Frontiers in Plant Science. 11 569958. doi:10.3389/fpls.2020.569958.
- ↑ Xu, S. (2003). "Theoretical basis of the Beavis effect". Genetics. 165 (4): 2259–2268. doi:10.1093/genetics/165.4.2259. PMC 1462909.
- ↑ Malosetti, M.; Ribaut, J.-M.; van Eeuwijk, F.A. (2013). "The statistical analysis of multi-environment data: modeling genotype-by-environment interaction and its genetic basis". Frontiers in Physiology. 4 44. doi:10.3389/fphys.2013.00044. PMC 3594989.
- ↑ Petit, J.; Salentijn, E.M.J.; Paulo, M.-J.; Denneboom, C.; van Loo, E.N.; et al. (2020). "Elucidating the genetic architecture of fiber quality in hemp (Cannabis sativa L.) using a genome-wide association study". Frontiers in Genetics. 11 566314. doi:10.3389/fgene.2020.566314. PMC 7527631.
- ↑ Bernardo, R. (2008). "Molecular markers and selection for complex traits in plants: learning from the last 20 years". Crop Science. 48 (5): 1649–1664. doi:10.2135/cropsci2008.03.0131.
- ↑ Faux, A.-M.; Draye, X.; Flamand, M.-C.; Occre, A.; Bertin, P. (2016). "Identification of QTLs for sex expression in dioecious and monoecious hemp (Cannabis sativa L.)". Euphytica. 209 (2): 357–376. doi:10.1007/s10681-016-1641-2.
- ↑ 23.0 23.1 Stack, G.M.; Cala, A.R.; Quade, M.A.; Toth, J.A.; Monserrate, L.A.; et al. (2024). "Genetic mapping, identification, and characterization of a candidate susceptibility gene for powdery mildew in Cannabis sativa". Molecular Plant-Microbe Interactions. 37 (1): 51–61. doi:10.1094/mpmi-04-23-0043-r.
- ↑ 24.0 24.1 24.2 Tang, L.; Fan, C.; Yang, L.; Yuan, H.; Cheng, L.; et al. (2026). "Identification of a major QTL and candidate gene for flowering time in industrial hemp via BSA-seq and fine mapping". Scientific Reports. 16 (1) 32880. doi:10.1038/s41598-026-62855-y.
- ↑ 25.0 25.1 Dowling, C.A.; Shi, J.; Toth, J.A.; Quade, M.A.; Smart, L.B.; et al. (2024). "A FLOWERING LOCUS T ortholog is associated with photoperiod-insensitive flowering in hemp (Cannabis sativa L.)". The Plant Journal. 119 (1): 383–403. doi:10.1111/tpj.16769.
- ↑ de Ronne, M.; Lapierre, É.; Torkamaneh, D. (2024). "Genetic insights into agronomic and morphological traits of drug-type cannabis revealed by genome-wide association studies". Scientific Reports. 14 (1) 9162. doi:10.1038/s41598-024-58931-w. PMC 11033274.
- ↑ 27.0 27.1 Welling, M.T.; Liu, L.; Kretzschmar, T.; Mauleon, R.; Ansari, O.; et al. (2020). "An extreme-phenotype genome-wide association study identifies candidate cannabinoid pathway genes in Cannabis". Scientific Reports. 10 (1) 18643. doi:10.1038/s41598-020-75271-7. PMC 7596533.