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

How coffee is bred

How breeders create new coffee varieties – the traits they select for, where new genes come from, and how self-pollinating arabica and outcrossing robusta are taken from a first cross to a released variety.

Across the chain: Science

Source-checkedby Coffearium verification pass· Last source-checked 17 Nov 2026· Prepared with AI assistance· 20 sources (16 tier A)· 1 contested· Updated 17 Nov 2026· Literature last searched 17 Nov 2026

In brief

  • Arabica mainly self-fertilises, so breeders fix crosses into uniform lines by selfing and selection over generations; Castillo’s parent progenies were evaluated in the F5 and F6.2,3
  • Robusta mainly outcrosses; programmes release clones as well as seed varieties, and Côte d’Ivoire has used reciprocal recurrent selection since 1984.15,14
  • From start to release takes decades: Catuaí was crossed in 1949 and its release article appeared in 1972; Cenicafé’s programme began in 1968 and released Castillo in 2005.4,3
  • Genomic selection is new in coffee: a 2019 study found no earlier work in arabica, and Brazilian robusta programmes have begun to use it.13,15
  • Breeding is mostly done by public institutes in producing countries: the Instituto Agronômico de Campinas released 90 of Brazil’s 121 arabica cultivars of 1939–2009.1,9
Go to the deep dive ↓
Contents

What breeders select for#

Coffee breeding creates new varieties by crossing plants and choosing the best offspring. In arabica, breeding for disease resistance combined with vigour, productivity and quality began in India in the early 1920s.1 In the second half of the 20th century, comprehensive programmes followed in several other producing countries.1

The main goals have stayed much the same. They are yield, cup quality and resistance to diseases, above all coffee leaf rust and coffee berry disease.1,18 Plant size matters too: Catuaí was bred from the short Caturra Amarelo and Mundo Novo, and gave compact, productive lines.4 Newer trials also test how varieties respond to different climates.17 Breeding for disease resistance in detail is a subject of its own, planned as a separate entry.

Where new variation comes from#

A breeder can only select among differences that exist. Arabica has very little genetic variation, explained in The arabica bottleneck.2 In Brazil, the cultivars trace back to seven ancestors.9 Breeders therefore bring in material from four kinds of source.

  • Landraces and wild arabica. Traditional varieties and wild trees from the species’ home region can serve as parents. In Central America, F1 hybrids were bred from wild Ethiopian accessions.6
  • The Timor Hybrid. A natural cross of arabica and robusta, it brought robusta genes for rust resistance into arabica.9 It is covered in The Timor Hybrid.
  • Other crosses with robusta. In Brazil, Icatu came from a cross of tetraploid C. canephora with Bourbon Vermelho.9
  • Wild relatives. Resistance to the coffee leaf miner came from a natural cross of arabica with C. racemosa, made in the 1950s.11 A rust resistance gene, SH3, came from C. liberica.12

Breeding arabica: fixing a line#

C. arabica mainly fertilises itself.2 After a cross, breeders let the offspring self-pollinate for several generations. In each generation they keep the best single plants.3,4 The result is a true-breeding line: its seed gives plants like the parent.1

This is called pedigree selection; World Coffee Research uses the term for Marsellesa, selected in Nicaragua.19 Colombia’s Castillo was bred in a similar way. Caturra was crossed with the Timor Hybrid CIFC 1343, and the offspring were selfed through several generations.3 In the fifth and sixth generations (F5 and F6), breeders evaluated the progenies of 39 selected parent plants.3 Generations are counted from the cross: the F1 plants, selfed, give the F2 and then the F3.3

Breeders also backcross: they cross a selected plant again with a cultivated parent, then select again.5 Genes from other species reach arabica this way.11,12 A finished line is released as seed. Some varieties, such as Castillo, are released as a mixture of related lines.3

123456
  1. Cross. Two arabica parents with different traits, such as plant size.3,4
  2. F1 plant. The first generation after the cross.3
  3. Selfing and selection. Each generation self-pollinates; the best single plants are kept, and rows grow more uniform.2,3,4
  4. Backcross. A selected plant is crossed again with a cultivated parent.5,11
  5. Trials at several sites. The same lines are compared in different places.3,5,16
  6. Release as seed. A true-breeding line, or a mixture of lines.1,3
Breeding an arabica line: a cross, generations of self-pollination and selection, trials at several sites and release as seed. Schematic, no time scale.2,3,4

Breeding robusta: choosing and cloning trees#

C. canephora (robusta) works the other way. Each plant is genetically mixed, and the species mainly cross-pollinates.15 Such a tree does not pass its traits unchanged to its seed (our note). Programmes have released clonal varieties, multiplied vegetatively, as well as seed-propagated synthetic varieties.15

To improve whole populations, some programmes cross two genetic groups. Côte d’Ivoire has done so since 1984 by reciprocal recurrent selection: parents are chosen by how well their crosses perform.14 The named clones and how they are planted are covered in Robusta varieties and clones.

123
  1. Tree selection. The best trees in a varied field are chosen.15
  2. Cloning and clone trials. Chosen trees are multiplied vegetatively and tested as clones.15
  3. Reciprocal recurrent selection. Two robusta populations are crossed; parents are chosen by how their crosses perform, in repeated cycles.14
Breeding robusta: choosing and cloning single trees, and crossing two populations in repeated cycles. Schematic, no time scale.14,15

F1 hybrids#

A third route skips the long fixing of a line. Breeders cross two parents and plant the first generation itself, an F1 hybrid. In Central America, American cultivars were crossed with wild Ethiopian accessions.6 Its offspring segregate, so F1 plants are cloned or their seed is produced anew from the parents.6 Kenya’s Ruiru 11 is made by hand pollination; Starmaya uses a mother tree that makes no pollen.7,6 The details are in F1 hybrids.

12345
  1. Cross. A cultivated arabica and a wild parent, such as a wild Ethiopian accession.6
  2. F1 hybrid. The first generation itself is planted.6
  3. Cloning. F1 plants can be multiplied as clones.6
  4. Seed made anew. Or the seed is produced from the parents each time: by hand pollination for Ruiru 11, with a mother tree that makes no pollen for Starmaya.7,6
  5. Own seed not replanted. The hybrid’s offspring segregate.6
The F1 hybrid route: the hybrid is cloned or its seed is made anew from the parents. Schematic.6,7

From cross to release#

Breeding a coffee variety takes decades. In Brazil, the cross that gave Catuaí was made in 1949; the article that marked its release appeared in 1972.4 In Colombia, Cenicafé’s programme for rust-resistant varieties began in 1968, and Castillo was released in 2005.3 In Kenya, breeding at the Ruiru station began in the 1970s; Ruiru 11 was released in 1985 (1986 in one source).7,20,6 Batian, which also combines resistance to coffee berry disease and leaf rust, followed in 2010.20

Robusta is not faster. According to a 2026 review of robusta programmes, developing a variety takes more than 20 years.15

194519551965197519851995200520152025yearCatuaí, Brazil(IAC): cross to release article1949–19724 Castillo, Colombia(Cenicafé): rust-resistanceprogramme to release1968–20053 Starmaya, Nicaragua(CIRAD–ECOM): male-sterile treefound to first commercial seed2001–20196
Data table
yearSource
Catuaí, Brazil (IAC): cross to release article1949–19724
Castillo, Colombia (Cenicafé): rust-resistance programme to release1968–20053
Starmaya, Nicaragua (CIRAD–ECOM): male-sterile tree found to first commercial seed2001–20196
Years from the start of breeding work to release for three arabica varieties. The events differ: a cross (Catuaí), the start of a programme (Castillo) or the discovery of a pollen-free tree (Starmaya), to a release article (Catuaí), a release (Castillo) or the first commercial seed, expected in 2019 (Starmaya).4,3,6

Who breeds coffee#

Most coffee varieties in the sources consulted come from public research institutes.9,10,15 A 2015 review expects breeding of locally adapted varieties to stay national or regional.1 In Brazil, the Instituto Agronômico de Campinas (IAC) released 90 of the 121 arabica cultivars of 1939–2009.9 India’s Central Coffee Research Institute (CCRI) has developed 13 arabica and three robusta selections.10

Other players work with or beside them. Starmaya was bred by the French research centre CIRAD with the coffee company ECOM, which supplied plant material and field support in Nicaragua.6 World Coffee Research (WCR), a research organisation, leads Innovea, a global arabica breeding network.15,18 According to WCR, Innovea is funded by the coffee companies that form its membership.18 Seed supply after release is covered in Seed systems.

Deep dive

Selection over generations in arabica#

The pedigree method rests on the mating system. Arabica is primarily self-fertilising, with occasional outcrossing by pollinators.2 Its cultivars are released as true-breeding lines or as F1 hybrids.1 The release documents show how many generations of selfing and selection a line took.

Castillo (Cenicafé, Colombia). Caturra (female parent) was crossed with the Timor Hybrid CIFC 1343. F1 plants were selfed to give the F2 and F3.3 The F4 and F5 were handled as a seed mixture of the plants selected before.3 In the F5 and F6, the progenies of 39 pre-selected parents were evaluated beside three commercial check varieties.3 The evaluation ran at five sites with contrasting climate and soils: Naranjal (Caldas), La Trinidad (Tolima), Paraguaicito (Quindío), El Rosario (Antioquia) and Santa Bárbara (Cundinamarca).3 According to the bulletin, the selection drew on what was learnt about rust in the producing regions over more than 22 years of cultivation in the presence of the disease.3 Castillo is a composite cultivar, uniform in its agronomic characteristics. Regional composites were drawn from it for each selection area.3 It was released by resolution of the Federación Nacional de Cafeteros on 13 May 2005.3

Catuaí (IAC, Brazil). Caturra Amarelo, short with yellow fruit, was crossed with Mundo Novo in 1949.4 A 2018 reanalysis of the IAC records found that the choice of what to recommend was made in the F1:2 generation, on two harvests.4 From 1,371 F2:3 plants, 101 progenies were derived into the F3:4 generation, from 3,986 plants in total.4 According to the authors, the selection begun in 1949 allowed compact, productive lines to be distributed 15 years later. The release article followed in 1972; the authors describe this as 24 years after the cross.4 The dates as given are 23 years apart (our note).

Further examples. Marsellesa, a Sarchimor from Timor Hybrid 832/2 × Villa Sarchi CIFC 971/10, was selected by CIRAD and ECOM to the F8.19,6 In 2017 the same group decided to produce F1 hybrid seed from parent cultivars in the F10.6 MG Travessia, from Minas Gerais, was tested as progenies in the fourth generation of selfing after a second backcross.5

Backcrosses and genes from other species#

Backcrossing returns a hybrid to a cultivated parent over several generations. It keeps a wanted gene while recovering the cultivated type.

  • Within arabica. MG Travessia came from IAC crosses of Catuaí Amarelo IAC H 2077-2-12-70 with Mundo Novo IAC 515-20, followed by backcrosses and selfing.5 Its progenies were assessed over eight harvests at two EPAMIG farms and six harvests at two private farms in Minas Gerais. It was registered as MGS Travessia.5
  • From C. racemosa. Plants from a natural cross of arabica with C. racemosa, made in the 1950s, are the main source of resistance to the leaf miner. Plants of the second generation of natural backcrosses to arabica were crossed with commercial cultivars to breed resistant cultivars.11 The resulting cultivars are named in Coffee leaf miner.
  • From robusta and other species. Brazil introduced the Icatu cross and the Timor Hybrid, both interspecific, for rust resistance.9 In India, spontaneous tetraploid hybrids between arabica and diploid species have also allowed introgressive breeding.10
  • From C. liberica. A 2017 Brazilian study worked with progeny of Indian selections crossed with arabica cultivars. It tracked SH3, a rust resistance gene from C. liberica.12

Material from wild relatives is described in Stenophylla and wild relatives.

Robusta schemes and cycle length#

Robusta breeding follows the biology of an outcrossing, heterozygous species.15 In Côte d’Ivoire, reciprocal recurrent selection between the Guinean and Congolese populations has been used since 1984.14 In the first cycle, intergroup progenies in three trials planted in 1985 and 1986 were recorded tree by tree for six years.14 Six promising crosses and 65 single trees within 19 crosses were selected; the expected yield gains are given in Robusta varieties and clones.14

The 2026 review describes the length of each step. The seed-to-seed generation, including evaluation and selection, typically spans almost 20 years.15 Brazil’s state institute Incaper documents 10 steps over 12 years to develop clonal cultivars of conilon.15 In single-tree trials summarised in the review, evaluations lasted 2–4 years in Côte d’Ivoire, 5–7 years in Brazil and 14 years in Cameroon.15 The authors propose cutting field evaluation from six to four years, based on trait correlations and predictions.15

Molecular markers and genomic selection#

Two tools use DNA to choose plants earlier.

Marker-assisted selection (MAS) uses a DNA marker linked to one known gene. In a 2017 Brazilian study, validated markers identified 11 resistant plants homozygous for the rust gene SH3 in a breeding progeny.12 Seven of them also carried a resistance gene from the Timor Hybrid. Two plants carrying Ck-1, a gene for resistance to coffee berry disease, were also found.12 The authors name such plants as sources for gene pyramiding, the stacking of several resistance genes in one variety.12 In India, CCRI reports that DNA marker tools are being integrated into breeding.10

Genomic selection (GS) predicts a plant’s value from thousands of markers across the genome. A 2019 Brazilian study states that no earlier work had addressed arabica.13 It genotyped 195 plants of 13 families with 21,211 SNP markers.13 For its comparison, it assumed 24 years for the release of an arabica cultivar: four selection cycles of six years.13 The authors conclude that genomic selection is promising for shortening the selection cycle.13 For robusta, the 2026 review reports that Brazilian programmes developed and implemented genomic selection. It recommends that other programmes integrate it, as Brazil is exploring.15

The 2015 arabica review and the Innovea network both name genomics-assisted or genomic selection as a tool to adopt.1,18 No source consulted for this entry reports a released coffee variety chosen by genomic selection.

Multi-location and international trials#

A variety that does well at one station may fail at another. Breeders therefore test lines in several environments before release.3,5 Castillo was tested at five sites in Colombia, MG Travessia at four sites in Minas Gerais.3,5 Catuaí lines went on being compared after release: the 2018 reanalysis catalogued 116 trial summaries from 1973 to 2017, from 22 municipalities in nine Brazilian states.4 Starmaya was compared in seven environments in Nicaragua over five years, 2013–2018.6

The International Multilocation Variety Trial (IMLVT), run by World Coffee Research, tests varieties across borders. In 2015, WCR gathered 31 arabica varieties from 11 breeding programmes.17,16 They were planted at 29 sites in 18 coffee-growing countries.16,17 In 2025 the programme was extended with a smaller group of countries.17 A 2025 analysis of leaf rust scores found highly significant effects of both the variety and the variety-by-site interaction.16 EC16 and Ruiru 11 were the most resistant across all sites.16 The authors conclude that no single variety is optimal everywhere and that selection should be done locally.16

Funding and institutions#

Most programmes in the sources consulted are run by public research institutes.1,9,10,15 The figure shows the Brazilian case.

020406080100cultivarsIAC(Instituto Agronômico de Campinas)909 EPAMIG/UFV129 IAPAR119 Fundação Procafé89
Data table
cultivarsSource
IAC (Instituto Agronômico de Campinas)909
EPAMIG/UFV129
IAPAR119
Fundação Procafé89
Arabica cultivars released in Brazil from 1939 to 2009, by the institution that released them (121 in total).9

In robusta, the 2026 review describes Uganda’s National Coffee Research Institute as a public research institute and Embrapa as a federal research institution.15 In Colombia, Castillo was released by the Federación Nacional de Cafeteros.3 In Kenya, Ruiru 11 was bred by the Coffee Research Foundation, now the Kenya Agricultural and Livestock Research Organization (KALRO).7

Private money enters in two ways in the sources consulted. Companies co-breed, as ECOM did with CIRAD for Starmaya and Marsellesa.6,19 Companies also fund shared networks. According to WCR, Innovea is funded by the coffee companies that form its membership.18 A 2026 review describes it as pre-competitive: it provides improved breeding material to participating countries, which develop it into finished varieties.15 WCR’s announcement sets out a schedule of further crosses and possible first releases; these are plans.18 No source consulted for this entry gives total public or private spending on coffee breeding.

Reference values#

Parameter Value reported Source
Mating system, C. arabica Primarily self-fertilising Scalabrin et al. 20202
Mating system, C. canephora Heterozygous, predominantly outcrossing Kawuki et al. 202615
Castillo: generations F1 selfed to F2–F3; F4–F5 as seed mixture; progenies of 39 selected parents evaluated in F5–F6 Alvarado-Alvarado et al. 20053
Castillo: programme start and release 1968; 13 May 2005 Alvarado-Alvarado et al. 20053
Catuaí: cross and release article 1949; 1972 Guerreiro Filho et al. 20184
Ruiru 11: programme start and release 1970s; 1985 (1986 in one source) WCR7; Gichuru et al. 201220; Georget et al. 20196
Batian: release 2010 Gichuru et al. 201220
Starmaya: male-sterile tree found; first commercial seed 2001; 2019 (expected) Georget et al. 20196
Marsellesa: generation at selection F8 Georget et al. 20196
MG Travessia: generation when tested Fourth selfed generation after a second backcross Carvalho et al. 20175
Arabica release time assumed in a GS study 24 years (4 cycles × 6 years) Sousa et al. 201913
Robusta variety development More than 20 years; seed-to-seed almost 20 years Kawuki et al. 202615
Conilon clonal cultivars, Incaper 10 steps, 12 years Kawuki et al. 202615
Reciprocal recurrent selection, Côte d’Ivoire Since 1984 Leroy et al. 199714
Arabica cultivars released in Brazil, 1939–2009 121: IAC 90, EPAMIG/UFV 12, IAPAR 11, Fundação Procafé 8 Setotaw et al. 20139
Selections released by CCRI, India 13 arabica, 3 robusta Prakash et al. 201610
IMLVT 31 varieties, 11 programmes, 29 sites, 18 countries; from 2015 Berny Mier y Teran et al. 202516; WCR17

Open questions#

  • Shorter cycles. Whether genomic selection and shorter field evaluations can cut the decades a variety takes, without losing accuracy for yield over many harvests.13,15
  • Local or global varieties. The international trial found strong interaction between variety and site for rust resistance; how far varieties bred in one country can serve others is not settled.16,1
  • Durability of resistance. Resistance to coffee berry disease in Ruiru 11 and similar cultivars had no confirmed field breakdown by 2009.8 Host resistance to leaf rust, by contrast, has broken down.1 How to breed resistance that lasts is the subject of the planned entry on breeding for disease resistance.

Key studies#

Guerreiro Filho, Ramalho and Andrade, 2018 – The Catuaí selection. Reanalysed the IAC records of the Catuaí selection, from the 1949 cross to the release, and four decades of later trials. It shows how decisions were made on few harvests in early generations, and what that means for breeding today.

Crop Breeding and Applied Biotechnology 18(4), 460–466 · doi:10.1590/1984-70332018v18n4p69

Alvarado-Alvarado, Posada-Suárez and Cortina-Guerrero, 2005 – Castillo. Cenicafé’s release bulletin for Castillo sets out the cross with the Timor Hybrid, the generations of selfing and selection, the five test sites and the composite make-up of the variety.

Avances Técnicos 337, Cenicafé · cenicafe.org

Sousa et al., 2019 – Genomic selection in arabica. The first genomic selection study in arabica, according to its authors, predicted traits of 195 plants from genome-wide markers. Prediction accuracy varied widely between traits, and the authors judged the method promising for shortening the selection cycle.

Frontiers in Plant Science 9, 1934 · doi:10.3389/fpls.2018.01934

Berny Mier y Teran et al., 2025 – Rust in the international trial. Analysed leaf rust in arabica varieties grown in many countries in the IMLVT. Resistance depended on the variety and on the site, and no variety was best everywhere.

Frontiers in Plant Science 16 · doi:10.3389/fpls.2025.1583595

Kawuki et al., 2026 – A century of robusta breeding. Robusta breeders and World Coffee Research staff compare methods, evaluation periods and results across Africa, Asia and the Americas. They name the many years needed to develop a variety as a main challenge and propose genomic selection and shorter evaluations.

Frontiers in Plant Science 16 · doi:10.3389/fpls.2025.1719980

Contested
  • Release year of Ruiru 11: the World Coffee Research catalogue and a 2012 study give 19857,20, and a 2009 study dates the release of resistant cultivars from 19858; the 2019 Starmaya paper says Ruiru 11 was released in 1986.6 The F1 hybrids entry lists the same two years.
Share or cite this entry

Change log

  • – Castillo: the 39 parents are no longer said to have been chosen in the F5 and F6; the progenies of 39 previously selected parents were evaluated in the F5 and F6 (brief, body, reference table). Reason: Alvarado-Alvarado et al. 2005: “En generaciones avanzadas de selección (F5 y F6) se evaluó la progenie de 39 progenitores previamente seleccionados”.
  • – Castillo: the “more than 22 years” now refers to what was learnt about rust over 22 years of cultivation, not to 22 years of selection; the five test locations are called sites, not Cenicafé stations. Reason: The bulletin ties the 22 years to the dynamics of rust in the producing regions and calls the five places “regiones o localidades”.
  • – Spread of breeding programmes after 1950: “most large programmes” replaced by comprehensive programmes in several other producing countries; the s1 passage, cut mid-word, replaced by the full sentence. Reason: van der Vossen et al. 2015 abstract: “comprehensive breeding programmes have been implemented in several other coffee producing countries”.
  • – Narrow genetic base: the arabica-wide statement now rests on Scalabrin et al. 2020; Setotaw et al. 2013 is cited for Brazil’s cultivars only. Reason: Setotaw et al. 2013 analyse Brazilian cultivars only (K2).
  • – Rume Sudan is no longer called a Sudanese landrace; the landrace bullet now gives the wild Ethiopian accessions used for F1 hybrids in Central America. Reason: The WCR Ruiru 11 page does not describe Rume Sudan as a landrace or give its origin; Georget et al. 2019 names wild Ethiopian accessions.
  • – C. racemosa: “in Brazil” removed for the 1950s cross, and the backcrosses described as natural backcrosses, as the source states. Reason: Dantas et al. 2021 calls the cross and the RC2 backcrosses natural and does not say where the cross was made.
  • – Backcrossing: MG Travessia (a cross within arabica) no longer supports a statement about genes from other species; that statement now cites Dantas et al. 2021 and Alkimim et al. 2017. India: “spontaneous interspecific hybrids” under “From robusta” now “spontaneous tetraploid hybrids between arabica and diploid species”, and the bullet heading widened. Reason: Carvalho et al. 2017 backcrosses Catuaí Amarelo and Mundo Novo; Prakash et al. 2016 abstract names hybrids with “other diploid species”.
  • – Robusta: “breeders multiply single trees as clones, from cuttings” replaced by the release of clonal (vegetatively propagated) and seed-propagated synthetic varieties; figure legend 9 changed accordingly; brief 2 changed. Reason: Kawuki et al. 2026 states that both synthetic and clonal varieties have been released and does not describe cuttings.
  • – F1 hybrids: the definition and the cultivated × wild Ethiopian parents now cite Georget et al. 2019 instead of the 2015 abstract; figure legend 7 no longer says the hybrid is cloned, since Ruiru 11 and Starmaya are produced from seed. Reason: The 2015 abstract does not define F1 hybrids; Georget et al. 2019 and the WCR Ruiru 11 page describe seed production by hand pollination and male sterility.
  • – ECOM is no longer described as the company that marketed Starmaya (body and tierNote); it supplied plant material and field support. Reason: Georget et al. 2019 does not say who markets Starmaya; it thanks ECOM (Exportadora Atlantic) for plant material and technical support.
  • – Innovea: launch year (2022), nine countries, over 200 member companies, crosses every six years and releases from 2033 removed; Innovea is now described from the 2026 review (A) as a WCR-led pre-competitive arabica breeding network. Reason: These numbers rested only on WCR’s news page (Tier B); number rule (one Tier A or two Tier B sources).
  • – Genomic selection: the 2019 arabica study is “the first” only according to its authors (brief, quick facts, study card). Public instead of national research institutes (IAC, Incaper and EPAMIG are state institutes). Reason: Sousa et al. 2019 write that “no GS work has addressed this species”; Setotaw et al. 2013, Kawuki et al. 2026.
  • – Timelines caption now states that the end events also differ (release article, release, first commercial seed expected in 2019); Brazil’s cultivar counts by institution (EPAMIG/UFV 12, IAPAR 11, Fundação Procafé 8) added to the reference table to match the chart. Reason: C5: every value in a chart must appear in the text or tables with its source; Georget et al. 2019 gives 2019 as the expected first commercial seed.
  • – Quotes corrected to the sources’ exact text: s5 parents (comma, not full stop, in the source), s15 mating system and genomic selection (a citation follows in the source), s18 curly apostrophe; Kawuki et al. 2026 now described as covering eight programmes, not written by breeders of eight programmes. Reason: Checked by hand against the sources (mechanical quote checker blocked).

Sources

  1. 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. doi:10.1007/s10681-015-1398-z · other copiesQuoted passage: a description; Review of arabica breeding by CIRAD-linked authors (abstract only): goals since the 1920s, kinds of resistant cultivars, the case for international networking and genomics-assisted selection, breeding kept national or regional..
  2. 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 copiesQuoted passage: a description; Genome study of 736 arabica accessions; the sentence on the mating system is background in the introduction..
  3. AAlvarado-Alvarado, G.; Posada-Suárez, H. E.; Cortina-Guerrero, H. A. (2005). Castillo: nueva variedad de café con resistencia a la roya (Avances Técnicos 337). Cenicafé (Colombian Coffee Research Centre), Chinchiná.Quoted passage: the authors’ own result; Cenicafé release bulletin for Castillo (Caturra × Timor Hybrid CIFC 1343): generations, selection of 39 parents, five Colombian trial sites, composite cultivar, release date 2005..
  4. AGuerreiro Filho, O.; Ramalho, M. A. P.; Andrade, V. T. (2018). Alcides Carvalho and the selection of Catuaí cultivar: interpreting the past and drawing lessons for the future. Crop Breeding and Applied Biotechnology 18(4), 460–466. doi:10.1590/1984-70332018v18n4p69 · other copiesQuoted passage: the authors’ own result; Reanalysis of the IAC records of the Catuaí selection (Caturra Amarelo × Mundo Novo, 1949) and of later competition trials in Brazil, 1973–2017..
  5. ACarvalho, G. R.; Bartholo, G. F.; Pereira, A. A.; Rezende, J. C. de; Botelho, C. E.; Oliveira, A. C. B. de; Silva, F. L. da (2017). MG Travessia: a coffee arabica cultivar productive and responsive to pruning. Crop Breeding and Applied Biotechnology 17, 281–285. doi:10.1590/1984-70332017v17n3c43 · other copiesQuoted passage: the authors’ own result; Cultivar release paper (EPAMIG and partners, Minas Gerais, Brazil): backcross-derived progenies tested over six to eight harvests at four sites; registered as MGS Travessia..
  6. AGeorget, F.; Marie, L.; Alpizar, E.; Courtel, P.; Bordeaux, M.; Hidalgo, J. M.; Marraccini, P.; Breitler, J.-C.; Déchamp, E.; Poncon, C.; Etienne, H.; Bertrand, B. (2019). Starmaya: The First Arabica F1 Coffee Hybrid Produced Using Genetic Male Sterility. Frontiers in Plant Science 10, 1344. doi:10.3389/fpls.2019.01344 · other copiesQuoted passage: the authors’ own result; Development of the F1 hybrid Starmaya by CIRAD and ECOM in Nicaragua: discovery of a male-sterile tree (2001), seed gardens (from 2009), multi-site trials (2013–2018)..
  7. A/BWorld Coffee Research Ruiru 11. World Coffee Research, Arabica Coffee Varieties catalogue.Quoted passage: data; WCR catalogue entry for Ruiru 11 (Kenya, Coffee Research Foundation, now KALRO): lineage, seed production by hand pollination, history of the programme..
  8. Avan der Vossen, H. A. M.; Walyaro, D. J. (2009). Additional evidence for oligogenic inheritance of durable host resistance to coffee berry disease (Colletotrichum kahawae) in arabica coffee (Coffea arabica L.). Euphytica 165, 105–111. doi:10.1007/s10681-008-9769-3 · other copiesQuoted passage: the authors’ own result; Inheritance study of resistance to coffee berry disease in arabica (Kenya); abstract only. Programme history and release dates are background in the abstract..
  9. 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 copiesQuoted passage: data; Pedigree analysis of the 121 arabica cultivars released in Brazil, 1939–2009, with the releasing institution of each (Table 1, Table 3)..
  10. APrakash, N. S.; Devasia, J.; Raghuramulu, Y.; Aggarwal, R. K. (2016). Genetic Diversity and Coffee Improvement in India. In: Rajpal, V. R.; Rao, S. R.; Raina, S. N. (eds), Molecular Breeding for Sustainable Crop Improvement, Sustainable Development and Biodiversity 11, Springer, Cham, pp. 231–268. doi:10.1007/978-3-319-27090-6_11 · other copiesQuoted passage: a description; CCRI breeding chapter, abstract only: selections released in India, use of spontaneous interspecific hybrids, integration of DNA marker tools..
  11. ADantas, J.; Motta, I. O.; Vidal, L. A.; Nascimento, E. F. M. B.; Bilio, J.; Pupe, J. M.; Veiga, A.; Carvalho, C.; Lopes, R. B.; et al.; last author Albuquerque, É. V. S. (2021). A comprehensive review of the coffee leaf miner Leucoptera coffeella (Lepidoptera: Lyonetiidae) – a major pest for the coffee crop in Brazil and others Neotropical countries. Insects 12(12), 1130. doi:10.3390/insects12121130 · other copiesQuoted passage: a result cited from earlier work; Review of the coffee leaf miner; the breeding history of resistance from C. racemosa in Brazil is cited from earlier work..
  12. AAlkimim, E. R.; Caixeta, E. T.; Sousa, T. V.; Pereira, A. A.; Oliveira, A. C. B. de; Zambolim, L.; Sakiyama, N. S. (2017). Marker-assisted selection provides arabica coffee with genes from other Coffea species targeting on multiple resistance to rust and coffee berry disease. Molecular Breeding 37, 6. doi:10.1007/s11032-016-0609-1 · other copiesQuoted passage: the authors’ own result; Brazilian (UFV/Embrapa/EPAMIG) study using validated molecular markers to find plants carrying the rust gene SH3 and the CBD gene Ck-1 in a breeding progeny..
  13. ASousa, T. V.; Caixeta, E. T.; Alkimim, E. R.; Oliveira, A. C. B.; Pereira, A. A.; Sakiyama, N. S.; Zambolim, L.; Resende, M. D. V. (2019). Early Selection Enabled by the Implementation of Genomic Selection in Coffea arabica Breeding. Frontiers in Plant Science 9, 1934. doi:10.3389/fpls.2018.01934 · other copiesQuoted passage: the authors’ own result; Genomic prediction in 195 arabica plants of 13 families from the Brazilian (UFV/EPAMIG) programme; release time of 24 years is the authors' assumption for their efficiency comparison..
  14. ALeroy, T.; Montagnon, C.; Cilas, C.; Yapo, A.; Charmetant, P.; Eskes, A. B. (1997). Reciprocal recurrent selection applied to Coffea canephora Pierre. III. Genetic gains and results of first cycle intergroup crosses. Euphytica 95(3), 347–354. doi:10.1023/A:1003074716379 · other copiesQuoted passage: the authors’ own result; First-cycle intergroup (Guinean × Congolese) progenies in three trials planted 1985–1986 in Côte d’Ivoire, six years of individual-tree data..
  15. AKawuki, R.; Akpertey, A.; Barrera, S.; Berny Mier y Teran, J. C.; Devasia, J.; Kraft, K.; Akbar, M. R.; Shivalingu, B. R.; Sseremba, G.; Teixeira, A.; Rocha, R. B.; Phan, V. H.; Wibowo, A.; Humphrey, T.; et al. (2026). Genetic improvement of robusta coffee (Coffea canephora) in the 20th and 21st centuries: prospects for increasing future breeding effectiveness and impact. Frontiers in Plant Science 16. doi:10.3389/fpls.2025.1719980 · other copiesQuoted passage: a description; Review of robusta breeding programmes in Africa, Asia and the Americas (2026): methods, cycle lengths, evaluation periods, genomic selection in Brazil, institutions..
  16. ABerny Mier y Teran, J. C.; et al.; Montagnon, C. (2025). Global Coffea arabica variety trials reveal genotype-by-environment interactions in resistance to coffee leaf rust (Hemileia vastatrix). Frontiers in Plant Science 16. doi:10.3389/fpls.2025.1583595 · other copiesQuoted passage: the authors’ own result; Leaf rust scores of arabica varieties in the WCR International Multilocation Variety Trial; analysis used 29 varieties at 23 sites with at least two years of data..
  17. A/BWorld Coffee Research International Multilocation Variety Trial. World Coffee Research, programme page.Quoted passage: a description; WCR page describing its International Multilocation Variety Trial: varieties gathered in 2015, countries and plots, aims, extension from 2025..
  18. BWorld Coffee Research (2022). World Coffee Research launches Innovea Global Coffee Breeding Network. World Coffee Research, news, 10 November 2022.Quoted passage: a description; WCR announcement of the Innovea breeding network (10 Nov 2022): member countries, funding by WCR’s company members, goals and planned timeline..
  19. A/BWorld Coffee Research Marsellesa. World Coffee Research, Arabica Coffee Varieties catalogue.Quoted passage: data; WCR catalogue entry for Marsellesa, a Sarchimor selected by CIRAD and ECOM in Nicaragua..
  20. AGichuru, E. K.; Ithiru, J. M.; Silva, M. C.; Pereira, A. P.; Várzea, V. M. P. (2012). Additional physiological races of coffee leaf rust (Hemileia vastatrix) identified in Kenya. Tropical Plant Pathology (short communication). doi:10.1590/S1982-56762012000600008 · other copiesQuoted passage: a result cited from earlier work; Race typing of 24 leaf rust samples from Kenya; the history of Kenyan resistant cultivars (Ruiru 11, Batian) is cited background..

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”.

The literature for this entry was last searched on 17 Nov 2026. A statement that no source consulted for this entry shows something refers to the sources above; newer or missing work may exist.

Suggest a change

On How coffee is bred. Tell us what you would change and why. A source that supports it helps most.