CoffeariumBeta
  1. Plant
  2. Farm
  3. Process
  4. Trade
  5. Roast
  6. Brew
  7. Cup
Botany · Genetics

The arabica bottleneck

Cultivated arabica is one of the least genetically diverse crops. How a single hybrid origin, self-pollination and a few founding plants narrowed it, and what that means for disease and breeding.

Across the chain: History · Science

Source-checkedby Coffearium verification pass· Last source-checked 22 Oct 2026· Prepared with AI assistance· 11 sources (10 tier A)· 2 contested· Updated 22 Oct 2026
Reading depth
Contents
  • A 2020 genome study found in arabica the lowest genetic diversity reported so far in any crop; its two parent species are about 5–11 times more diverse (our calculation).1
  • Arabica probably arose from a single hybridisation and mainly fertilises itself, which leads to high levels of inbreeding.1,2
  • Most cultivars worldwide descend from the Typica and Bourbon bases, which spread from Yemen from the early 18th century and carry very little variation.4,5
  • The narrow base leaves cultivated arabica susceptible to pests and diseases such as coffee leaf rust, and limits what breeders can work with.2,10
  • Ethiopian accessions are more diverse than commercial varieties and carry alleles found in no other group.6

What a genetic bottleneck is#

A genetic bottleneck happens when a population shrinks to a few individuals. The descendants carry only the variation those few plants had. Domestication usually involves such bottlenecks, which leave cultivated plants less diverse than their wild relatives.2

Arabica went through several of them. A 2020 genome study found in Coffea arabica the lowest genetic diversity reported so far in any crop species.1 Its two parent species, C. canephora and C. eugenioides, were about 5–11 times more diverse in the same study, by our calculation from its figures.1 The same holds for wild arabica: the narrow genetic base affects both cultivated and wild plants.2

Why arabica started narrow#

Arabica is a hybrid of C. canephora, the species grown as robusta, and C. eugenioides.8,2 The cross most likely happened only once.1 A 2024 genome study dates this founding event to 350,000–610,000 years ago.2 A species that begins from one hybrid begins with very little variation. How its two genomes sit side by side is covered in Polyploidy and the arabica genome.

The species then stayed small for a long time. The 2024 study found a bottleneck in the wild that began about 350,000 years ago and ended about 15,000 years ago.2

Arabica also mainly fertilises itself. Insects may occasionally carry pollen between plants.1 According to a 2025 review of studies under Colombian conditions, self-pollination can account for 56.4–95.3% of fruit set.9 Self-fertilisation leads to high levels of inbreeding, so each plant holds little variation of its own.1 Flowering and pollination has the details of the flower.

From Ethiopia to Yemen and the world#

The ancestors of today’s cultivars split from the wild population about 30,500 years ago.2 People later took a few plants from Ethiopia to Yemen. A 2022 genetic study calls this a first severe bottleneck, in the 15th century.5 Isolation of some populations within Yemen then caused a second one, before coffee was smuggled out in the 18th century.5

From Yemen, two genetic bases spread from the early 18th century: Typica and Bourbon. They gave rise to most arabica cultivars grown worldwide.4 The Bourbon group descends from a single plant that survived on Réunion by 1720.2 The full journey is told in Arabica varieties and their lines.

Each base carries very little variation, as historical records of a few founding plants would predict.4 Breeding then worked from the same few lines. In Brazil, seven ancestors make up 97.55% of the genetic base of the country’s arabica cultivars.7

12310456789
  1. Parent species. C. canephora and C. eugenioides.8,2
  2. One hybridisation. The founding event, 350,000–610,000 years ago.1,2
  3. Long wild bottleneck. From about 350,000 to about 15,000 years ago.2
  4. Cultivar founders split off. About 30,500 years ago.2
  5. Into Yemen. A first severe bottleneck, 15th century.5
  6. Typica base. Spread from Yemen from the early 18th century.4
  7. Bourbon base. From a single plant that survived by 1720.2
  8. Cultivars worldwide. Mostly from the Typica and Bourbon bases.4,5
  9. Timor Hybrid. Robusta genes raise diversity in some cultivated lines.3
  10. Wild Ethiopian arabica. More diverse than cultivars and a possible source of new alleles for breeding (dashed arrow); narrowing again since about 5,000 years ago.6,2
How arabica’s gene pool narrowed. Band height stands for genetic diversity, qualitative only; studies disagree on how strong the Yemen bottleneck was (see Contested). Schematic, not to scale.

Why it matters#

Low diversity leaves a crop exposed. The 2024 genome study links arabica’s low diversity to its susceptibility to many pests and diseases, such as coffee leaf rust.2 As a result, the classic Bourbon and Typica lines grow successfully in only a few regions.2

Breeders feel it too. A 2015 review names limited access to further arabica genetic resources and the breakdown of rust resistance among the newer challenges.10 According to World Coffee Research, most introgressed varieties owe their rust resistance to one shared parent, the Timor Hybrid.11 Robusta genes from that hybrid have widened diversity in some cultivated lines.3 The hybrid is covered in The Timor Hybrid.

Wild Ethiopia as a reservoir#

Ethiopia holds more of the species’ variation. In a 2019 study, Ethiopian accessions were more diverse than commercial varieties.6 Accessions from west of the Great Rift Valley carried 34 alleles, gene variants, found in no other group.6 The study names such accessions as a source of new alleles for breeding.6 How they are grown and collected is covered in Ethiopian landraces.

The reservoir has limits. The 2024 genome study calls the narrow base of both cultivated and wild arabica an obstacle to breeding with wild diversity.2

Measured diversity#

Studies measure diversity in different ways. Marker studies count variants at a few loci: amplified fragment length polymorphisms (AFLP) or microsatellites (simple-sequence repeats, SSR). Genome studies estimate nucleotide diversity (π), the average share of DNA positions that differ between two copies of the genome.1,3

Marker studies. A 2002 study compared 11 accessions derived from the Typica and Bourbon bases, four cultivars from Yemen and 11 subspontaneous accessions from the species’ primary centre of diversity, using 107 AFLP markers and six microsatellites.4 Except for one AFLP marker, every marker in the cultivated material also occurred in the subspontaneous accessions. Polymorphism was much higher among the subspontaneous accessions and very low within the genetic bases.4 The Yemen cultivars grouped with the Typica-derived accessions.4

A 2019 study used 30 microsatellite markers on 33 arabica genotypes, 25 of them from the Ethiopian collection.6 Unbiased expected heterozygosity (uHe) was 0.183 in the western Ethiopian group, 0.170 in the eastern group and 0.106 in cultivars and inbred lines.6

A 2022 study genotyped 555 samples, mostly cultivated trees from Ethiopia and Yemen, with 10 microsatellite markers, and found six genetic clusters.5 Two “Core Ethiopia” clusters held the most alleles, 65 and 45, of 77 in total.5 The Ethiopian Legacy cluster held 39; the Typica/Bourbon, New-Yemen (109 samples) and Harrar (30 samples) clusters held 29–30 each.5 Cultivated varieties worldwide come from the “Ethiopian Legacy” and Typica/Bourbon clusters, none from New-Yemen or Harrar.5

01020304050607080allelesCore Ethiopia 1655 Core Ethiopia 2455 Ethiopian Legacy395 Typica/Bourbon305 New-Yemen (109 samples)305 Harrar (30 samples)295
Data table
allelesSource
Core Ethiopia 1655
Core Ethiopia 2455
Ethiopian Legacy395
Typica/Bourbon305
New-Yemen (109 samples)305
Harrar (30 samples)295
Alleles found in each of six genetic clusters of arabica from Ethiopia and Yemen, at 10 microsatellite markers. Clusters differ in size, and the study does not state that counts were standardised for it.5

Genome studies. A 2020 study estimated diversity in 736 arabica accessions by genotyping by sequencing.1 It found π = 2.3 × 10⁻⁴, from 652 SNPs across 193,873 informative nucleotides. The parent species gave π = 2.6 × 10⁻³ (C. canephora) and 1.1 × 10⁻³ (C. eugenioides), one order of magnitude higher.1 The authors found such low levels in crops only in bread wheat, another recent allopolyploid.1

A 2024 whole-genome study of 95 arabica accessions detected 1,877,440 SNPs and a median π of 3.12 × 10⁻⁴; the parent species were again one order of magnitude higher.3 It found that cultivars, landraces and spontaneous germplasm all have a limited genetic base.3

051015202530π × 10⁻⁴C. arabica (2020 study)2.31 C. arabica(2024 study, median)3.123 C. eugenioides (2020)111 C. canephora (2020)261
Data table
π × 10⁻⁴Source
C. arabica (2020 study)2.31
C. arabica (2024 study, median)3.123
C. eugenioides (2020)111
C. canephora (2020)261
Nucleotide diversity (π, × 10⁻⁴) of arabica in two genome studies and of its two parent species. The 2020 study used genotyping by sequencing of 736 accessions; the 2024 study whole genomes of 95.1,3

Timing of the bottlenecks#

A second 2024 genome study sequenced 41 arabica accessions, among them 17 wild and three wild or cultivated accessions from both sides of the Great Rift Valley.2 Modelling population history, it found the oldest bottleneck beginning abruptly about 350,000 years ago and ending about 15,000 years ago. A more recent bottleneck began gradually about 5,000 years ago and lasts to this day.2 Cultivated accessions showed the oldest bottleneck but not the recent one.2

The same study puts the split between the wild population and the cultivar founding population about 1,450 generations ago, about 30,000 years.2 In cultivars, nucleotide diversity was only slightly lower than in wild populations, which the authors read as only minor bottlenecks during domestication.2 Bourbon was less diverse than Typica, probably because of its known single-plant bottleneck.2 This reading differs from the marker studies; both are listed under Contested.

New variation in a narrow species#

Diversity is not fixed. The 2024 whole-genome study found that diversity rises in some cultivated genotypes to levels seen in one of the parent species, presumably through introgression from the Timor Hybrid.3 It also found cryptic Timor Hybrid introgression in one accession of the Bourbon/Typica group.3

The same study found recent chromosomal changes still varying within the germplasm: aneuploidies, deletions, duplications and exchanges between the two subgenomes. Four accessions showed whole-chromosome aneuploidies.3 The authors propose these changes as a fundamental source of variation in this low-diversity species.3

Consequences for breeding#

Breeding has drawn on few parents. A 2013 pedigree study traced the 121 arabica cultivars released in Brazil between 1939 and 2009 to 13 ancestors. Seven of them make up 97.55% of the genetic base, and Bourbon Vermelho alone 52.76%.7

A 2015 review lists limited access to additional arabica genetic resources, breakdown of host resistance to leaf rust, worsening insect pests and climate change as challenges for breeders.10 According to World Coffee Research, cultivars from Typica and Bourbon still make up much of Latin American production, contributing to the bottleneck.11 Breeders’ answers, introgression and F1 hybrids with wild Ethiopian parents, are covered in The Timor Hybrid and F1 hybrids.

Reference values#

Parameter Value Source
Nucleotide diversity, C. arabica, 736 accessions (genotyping by sequencing) π = 2.3 × 10⁻⁴ Scalabrin et al. 20201
Nucleotide diversity, C. canephora π = 2.6 × 10⁻³ Scalabrin et al. 20201
Nucleotide diversity, C. eugenioides π = 1.1 × 10⁻³ Scalabrin et al. 20201
Nucleotide diversity, C. arabica, 95 accessions (whole genomes) median π = 3.12 × 10⁻⁴ Scalabrin et al. 20243
SNPs in 95 arabica accessions 1,877,440 Scalabrin et al. 20243
Founding hybridisation 350,000–610,000 years ago Salojärvi et al. 20242
Oldest wild bottleneck ≈350,000 to ≈15,000 years ago Salojärvi et al. 20242
Recent wild bottleneck from ≈5,000 years ago Salojärvi et al. 20242
Wild and cultivar founders split ≈30,500 years ago (≈1,450 generations) Salojärvi et al. 20242
Self-pollination share of fruit set (Colombia) 56.4–95.3% Unigarro et al. 20259
Insect cross-pollination share of fruit set (Colombia) 9.4–26.7% Unigarro et al. 20259
uHe, western Ethiopian / eastern Ethiopian / cultivars 0.183 / 0.170 / 0.106 da Silva et al. 20196
Private alleles, west of the Rift Valley 34 da Silva et al. 20196
Alleles, Core Ethiopia 1 / Typica/Bourbon (10 SSR) 65 / 30 of 77 Montagnon et al. 20225
Yemen samples by cluster: New-Yemen / Typica/Bourbon / Ethiopian Legacy 57% / 27% / 16% Montagnon et al. 20225
Brazil’s cultivar genetic base from seven ancestors 97.55% (121 cultivars, 1939–2009) Setotaw et al. 20137

Values from different methods are not directly comparable: π from genotyping by sequencing and from whole genomes, uHe and allele counts each measure diversity differently.1,3,5,6

Open questions#

  • Strength of the domestication bottleneck. Marker studies describe strong losses on the way out of Ethiopia; a 2024 genome study finds cultivars only slightly less diverse than wild plants.2,4,5
  • Chromosomal changes as a source of variation. How much aneuploidies and exchanges between subgenomes add to usable diversity is proposed, not yet measured in breeding.3
  • Breeding from wild diversity. Ethiopian accessions carry private alleles, but the narrow base of wild arabica itself limits what they can add.2,6

Key studies#

Anthony et al., 2002 – Where cultivated arabica came from. Compared accessions from the Typica and Bourbon bases, Yemeni cultivars and subspontaneous coffee from the centre of diversity with AFLP and microsatellite markers. Almost every marker in the cultivated material also occurred in the subspontaneous accessions, and variation within each base was very low.

Theoretical and Applied Genetics 104, 894–900 · doi:10.1007/s00122-001-0798-8

Scalabrin et al., 2020 – One polyploidisation, little variation. Estimated diversity in 736 arabica accessions and in both parent species. Arabica showed the lowest diversity reported in any crop, one order of magnitude below its parents, which the authors trace to a single polyploidisation event and self-fertilisation.

Scientific Reports 10, 4642 · doi:10.1038/s41598-020-61216-7

Scalabrin et al., 2024 – Chromosome changes as a source of diversity. Built a chromosome-scale arabica genome and resequenced a large panel. Found low diversity even in the centre of origin, higher diversity in some Timor Hybrid-derived lines, and many recent chromosomal changes that may generate new variation.

Nature Communications 15, 463 · doi:10.1038/s41467-023-44449-8

Salojärvi et al., 2024 – The arabica genome. Assembled genomes of arabica and its parents and modelled population history from wild and cultivated accessions. Dated the founding hybridisation and two wild bottlenecks, and found cultivars only slightly less diverse than wild plants.

Nature Genetics 56(4), 721–731 · doi:10.1038/s41588-024-01695-w

Montagnon et al., 2022 – Ethiopia and Yemen genotyped. Genotyped 555 samples, mostly cultivated coffee from Ethiopia and Yemen. Found six clusters, with the most alleles in core Ethiopian groups, and placed a first severe bottleneck at the introduction into Yemen.

Agronomy 12, 3203 · doi:10.3390/agronomy12123203

Contested
  • How much domestication narrowed arabica: marker studies describe successive losses of diversity as coffee spread from Ethiopia, with a severe bottleneck on its introduction into Yemen4,5; a 2024 genome study found nucleotide diversity in cultivars only slightly lower than in wild populations and suggests that only minor bottlenecks occurred during domestication.2
  • Diversity in the centre of origin: microsatellite studies find Ethiopian accessions clearly more diverse than cultivars5,6; a 2024 whole-genome study reports low diversity in the centre of origin itself, with Ethiopian accessions and the Bourbon/Typica varieties together encompassing very low diversity.3
Share or cite this entry

Sources

  1. AScalabrin, S.; Toniutti, L.; Di Gaspero, G.; et al.; Bertrand, B. (2020). A single polyploidization event at the origin of the tetraploid genome of Coffea arabica is responsible for the extremely low genetic variation in wild and cultivated germplasm. Scientific Reports 10, 4642. doi:10.1038/s41598-020-61216-7 · other copies
  2. ASalojärvi, J.; Rambani, A.; Yu, Z.; Guyot, R.; et al. (2024). The genome and population genomics of allopolyploid Coffea arabica reveal the diversification history of modern coffee cultivars. Nature Genetics 56(4), 721–731. doi:10.1038/s41588-024-01695-w · other copies
  3. AScalabrin, S.; Magris, G.; Liva, M.; Vitulo, N.; Vidotto, M.; Scaglione, D.; et al.; Di Gaspero, G.; Morgante, M. (2024). A chromosome-scale assembly reveals chromosomal aberrations and exchanges generating genetic diversity in Coffea arabica germplasm. Nature Communications 15, 463. doi:10.1038/s41467-023-44449-8 · other copies
  4. AAnthony, F.; Combes, M. C.; Astorga, C.; Bertrand, B.; Graziosi, G.; Lashermes, P. (2002). The origin of cultivated Coffea arabica L. varieties revealed by AFLP and SSR markers. Theoretical and Applied Genetics 104, 894–900. doi:10.1007/s00122-001-0798-8 · other copies
  5. AMontagnon, C.; Sheibani, F.; Benti, T.; Daniel, D.; Bote, A. D. (2022). Deciphering early movements and domestication of Coffea arabica through a comprehensive genetic diversity study covering Ethiopia and Yemen. Agronomy 12, 3203. doi:10.3390/agronomy12123203 · other copies
  6. Ada Silva, B. S. R.; Sant’Ana, G. C.; Chaves, C. L.; Androcioli, L. G.; Ferreira, R. V.; Sera, G. H.; Charmetant, P.; Leroy, T.; Pot, D.; Domingues, D. S.; Pereira, L. F. P. (2019). Population structure and genetic relationships between Ethiopian and Brazilian Coffea arabica genotypes revealed by SSR markers. Genetica 147(2), 205–216. doi:10.1007/s10709-019-00064-4 · other copies
  7. ASetotaw, T. A.; Caixeta, E. T.; Pereira, A. A.; Oliveira, A. C. B. de; Cruz, C. D.; Zambolim, E. M.; Zambolim, L.; Sakiyama, N. S. (2013). Coefficient of parentage in Coffea arabica L. cultivars grown in Brazil. Crop Science 53, 1237–1247. doi:10.2135/cropsci2012.09.0541 · other copies
  8. ALashermes, P.; Combes, M. C.; Robert, J.; Trouslot, P.; D’Hont, A.; Anthony, F.; Charrier, A. (1999). Molecular characterisation and origin of the Coffea arabica L. genome. Molecular and General Genetics 261(2), 259–266.
  9. AUnigarro, C. A.; Cayón Salinas, D. G.; León-Burgos, A. F.; Flórez-Ramos, C. P. (2025). Flowering and Fruiting of Coffea arabica L.: A Comprehensive Perspective from Phenology. Plants 14(21), 3396. doi:10.3390/plants14213396 · other copies
  10. Avan der Vossen, H.; Bertrand, B.; Charrier, A. (2015). Next generation variety development for sustainable production of arabica coffee (Coffea arabica L.): a review. Euphytica 204(2), 243–256.
  11. BWorld Coffee Research History of Arabica. World Coffee Research, Arabica Coffee Varieties catalogue.

Tier A · authoritative · B · trade and expert · C · general. Supporting quotes for every source are in the data version of this entry. If a publisher’s site refuses access, try the doi link or “other copies”.

Suggest a change

On The arabica bottleneck. Tell us what you would change and why. A source that supports it helps most.