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

Arabica and robusta compared

How the two commercial coffee species differ as plants, in their climate needs, in the shape of the bean and in what the seed holds, and how a laboratory tells them apart in green coffee, roast coffee and blends.

Across the chain: Science

Source-checkedby Coffearium verification pass· Last source-checked 22 Oct 2026· Prepared with AI assistance· 19 sources (16 tier A)· 3 contested· Updated 22 Oct 2026
Reading depth
Contents
  • *C. arabica* has 44 chromosomes and mainly fertilises itself; *C. canephora* (robusta) has 22 and is an outcrossing species.1,2
  • In East African flora descriptions, canephora has larger leaves and more flowers per leaf axil (8–30 against 2–20 in arabica).3,4
  • Robusta seeds hold more caffeine, arabica seeds more sucrose; whether robusta reliably holds more chlorogenic acids is disputed.11,12,13,19
  • The diterpene 16-O-methylcafestol (16-OMC) marks robusta; a 2024 European standard specifies how to measure it in green and roast coffee.14,17
  • DNA tests on chloroplast markers identify the species and estimate the robusta share in a blend.18

Two species behind one drink#

Almost all coffee comes from two species. Coffea arabica gives arabica coffee. Coffea canephora gives the coffee traded as robusta. This entry sets the two side by side; each species has its own entry with the detail.

The deepest difference is in the chromosomes. Arabica has 44, in four sets, because it arose as a hybrid of two species.1 Canephora has 22, in two sets, and it is one of arabica’s two parents.2 How that hybrid arose is told in Polyploidy and the arabica genome.

The two also reproduce differently. Arabica flowers mainly pollinate themselves; insects only occasionally bring pollen from other plants.1 Canephora is outcrossing.2 According to a coffee science textbook chapter, it is strictly self-incompatible, needing pollen from another plant, so breeders select clones or cross chosen parents.5

How the plants differ#

Botanists describe the two plants in the Flora of Tropical East Africa, quoted by Kew’s Plants of the World Online. Arabica is a bush or tree 2–7 m tall. Canephora is a tree or bush 3.5–8.5 m tall.3,4

The leaves tell them apart more clearly. Arabica leaf blades are 7–18 cm long and 3–7.5 cm wide.3 Canephora blades are 12–31.5 cm long, sometimes up to 40 cm, and 4.5–12 cm wide.4

Flowers grow in clusters where a leaf meets the branch, the leaf axil. The flora counts 2–20 flowers per axil in arabica. In canephora it counts 8–30, sometimes up to 48.3,4 Both have white flowers and red fruit; cultivated arabica can also bear yellow or purple fruit.3,4 How flowering and fruit set work is covered in Flowering and pollination.

The fruit takes longer to ripen in canephora, at least in the studies opened for this entry. A 2025 review gives an average of 220–243 days from flowering to ripe fruit for arabica.8 A 2025 Brazilian trial recorded 252–308 days from flowering to harvest for conilon, a form of canephora.9 The sites and methods differ, so the numbers are not a direct comparison.

Where each grows best#

Arabica is a highland plant and robusta a plant of warmer lowlands, in the classic view. A 2006 review gives an optimum mean annual temperature of 18–21 °C for arabica and 22–26 °C for robusta.6 The same review states that robusta tolerates cold much less well than arabica.6

That warm optimum for robusta has been questioned. A 2020 study of yields on 798 farms in South-East Asia found robusta did best below 20.5 °C.7 The two findings are set side by side in Contested.

Rain matters too. The 2006 review gives an optimum of 1,200–1,800 mm a year for arabica. It says robusta adapts better than arabica to rainfall above 2,000 mm.6 How growers match a species to a site is covered in Site selection and climate.

The seeds side by side#

The seeds, the coffee beans, differ in shape. One 2022 study measured roasted beans from 21 arabica and 14 robusta samples with a camera system, in dishes of 20 beans: 105 dishes of arabica and 70 of robusta.10 The arabica beans were longer on average, 10.95 mm against 10.46 mm. The robusta beans were slightly wider, 8.42 mm against 8.25 mm.10

So robusta beans look rounder and arabica beans more oval. The width-to-length ratio was 0.81 in robusta and 0.76 in arabica.10 The drawing shows both means at the same scale.

1234
  1. Arabica bean. Mean 10.95 × 8.25 mm; width-to-length ratio 0.76 (105 dishes of 20 roasted beans).10
  2. Robusta bean. Mean 10.46 × 8.42 mm; width-to-length ratio 0.81 (70 dishes of 20 roasted beans).10
  3. Length. The guides mark the ends of the arabica bean; the robusta bean is about 0.5 mm shorter.10
  4. Width. The robusta bean is slightly wider, by 0.17 mm on average.10
Mean bean outlines from one 2022 imaging study of roasted beans, drawn at the same scale. Shape varies within each species and with growing conditions. Schematic outline, not a photograph.10

The study did not test shape alone. Its 100% correct classification came from a model that combined spectral, shape and colour features and sorted dishes of 20 beans, not single beans.10 Bean shapes within one species, such as peaberries, are covered in Peaberries and bean shapes.

What the seeds hold#

The seeds contain the same classes of compounds in different amounts. Robusta holds more caffeine: a 2021 review gives 0.9–1.5% of dry weight for arabica and 1.2–2.4% for robusta (basis not stated).11 Arabica holds more sucrose. In a 2004 study of wild trees, arabica held 7.99–11.0% sucrose and canephora 5.94–6.47%, on a dry-matter basis.13

Lipids are higher in arabica, about 15% on average against about 10% in robusta (basis not stated).14 Chlorogenic acids, a family of phenolic compounds, are usually reported higher in robusta, but a meta-analysis disagrees (see Contested).12,19

0246810121416% of dry matterCaffeine, arabica(dry weight)0.9–1.511 Caffeine, robusta(basis not stated)1.2–2.411 Chlorogenic acids, arabica4–8.412 Chlorogenic acids, robusta7–14.412 Sucrose, arabica(four wild trees)7.99–11.013 Sucrose, canephora(four wild trees)5.94–6.4713 Lipids, arabica average(basis not stated)≈1514 Lipids, robusta average(basis not stated)≈1014
Data table
Point% of dry matterSource
Caffeine, arabica (dry weight)0.9–1.50.9–1.511
Caffeine, robusta (basis not stated)1.2–2.41.2–2.411
Chlorogenic acids, arabica4–8.44–8.412
Chlorogenic acids, robusta7–14.47–14.412
Sucrose, arabica (four wild trees)7.99–11.07.99–1113
Sucrose, canephora (four wild trees)5.94–6.475.94–6.4713
Lipids, arabica average (basis not stated)≈151514
Lipids, robusta average (basis not stated)≈101014
Reported ranges for four constituents of green coffee, by species. Caffeine and chlorogenic acids are review ranges; sucrose spans four wild trees per species; lipids are averages. Bases as the sources state them.11,12,13,14

These differences carry into the cup, according to a roasting chemistry chapter. It links arabica’s higher sucrose to more caramel-like and buttery aroma compounds after roasting.15 It describes robusta’s aroma profile as smoky, earthy, roasty and phenolic.15 The whole picture of green coffee chemistry is in Green coffee composition.

Comparison table#

Values as the sources report them. Where sources differ, both are given; plant measurements come from one flora and describe wild and cultivated plants in East Africa.

Trait C. arabica C. canephora (robusta)
Chromosomes 2n = 4x = 44, allopolyploid1 2n = 2x = 22, diploid2
Breeding system Primarily self-fertilisation; occasional outcrossing by pollinators1 Outcrossing, highly heterozygous2; strictly self-incompatible (textbook)5
Habit and height Bush or tree, 2–7 m3 Tree or bush, 3.5–8.5 m4
Leaf blade 7–18 × 3–7.5 cm3 12–31.5(–40) × 4.5–12(–22) cm4
Flowers per leaf axil 2–20, in 1–3(–15) clusters; flowers (4–)5(–6)-merous3 8–30(–48), in 1–4(–7) dense clusters4
Corolla lobes 0.9–1.6 cm long3 0.8–1.9 cm long4
Fruit Red (yellow or purple in cultivated forms), 1–2 cm long3 Red, 0.9–1.7 cm long4
Seed (flora) 6–7 mm wide3 7–13 mm long, 5.5–9.5 mm wide4
Bean, roasted, mean length × width 10.95 × 8.25 mm (105 dishes of 20 beans)10 10.46 × 8.42 mm (70 dishes of 20 beans)10
Bean width-to-length ratio 0.7610 0.8110
Flowering to ripe fruit 220–243 days on average (review)8 252–308 days (conilon, 647 m, Brazil); 216–238 days cited for early genotypes at low altitude9
Optimum mean annual temperature 18–21 °C6 22–26 °C (2006 review)6; below 20.5 °C (2020 yield study)7
Rainfall Optimum 1,200–1,800 mm a year6 Adapts better to more than 2,000 mm6
Cold More tolerant6 Much less adaptable to low temperatures6
Caffeine, green 0.9–1.5% of dry weight11 1.2–2.4% (basis not stated)11
Chlorogenic acids, green 4–8.4% of dry matter12 7–14.4% of dry matter12 (contested)
Sucrose, green 9.32% mean, 7.99–11.0% dry matter (four wild trees)13 6.10% mean, 5.94–6.47% dry matter (four wild trees)13
Trigonelline, green 1.13% mean, 0.95–1.29% dry matter13 0.82% mean, 0.52–1.06% dry matter13
Lipids, green (basis not stated) ≈15% (average)14 ≈10%14
16-O-methylcafestol Reported absent14; traces found in 201816 Present; 600–1,800 and 1,204–2,236 mg/kg in roast robusta, in two studies cited in 201816

The online text of the flora gives the arabica seed length as “9–1.2(–1.5) cm”, which looks like a typesetting slip; we do not use it.3 The canephora seed range in the flora (7–13 mm long) brackets the mean bean lengths of the 2022 study, which measured roasted beans of unstated origin.4,10 The longer conilon cycle was measured at 647 m, a transitional altitude; the trial cites 216–238 days for early genotypes at low altitude.9

Chemistry behind the differences#

Sucrose and trigonelline. The 2004 study measured 14 wild species and six further taxa, four trees each.13 Arabica had one of the highest sucrose contents in the set, 9.3% on a dry-matter basis. Canephora averaged 6.10%.13 Across all species, sucrose ranged from 3.8% to 10.7% and trigonelline from 0.39% to 1.77%.13 The authors call both compounds aroma precursors and suggest canephora could be improved by crosses with C. eugenioides.13

Caffeine and chlorogenic acids. The caffeine ranges of the two species overlap at 1.2–1.5%, so on these ranges caffeine alone would not separate every sample.11 The chlorogenic acid ranges of a 2006 review overlap between 7% and 8.4%.12 A 2021 meta-analysis in a master’s thesis found the chlorogenic acid profiles of the two species extremely similar; roast level affected the amounts more than species did, except for total feruloylquinic and dicaffeoylquinic acids.19 Chlorogenic acids treats the dispute in full.

Lipids and diterpenes. Arabica lipids average about 15%, robusta about 10%.14 The diterpene 16-O-methylcafestol (16-OMC) is the species-specific one: a 2006 review reports its absence from arabica as confirmed.14 The review calls it the ideal characteristic for detecting robusta in arabica blends.14 The diterpenes as a group are covered in Coffee oil.

Aroma. According to the roasting chemistry chapter, robusta’s higher amino acids lead to more pyrazines, and its tryptophan to animal-smelling skatole.15 The same chapter attributes arabica’s higher diketones, furfurals and furaneol to its higher sucrose.15 No sensory study opened for this entry compared the two species under one protocol; cup character here rests on that chapter alone.

Telling them apart in the lab#

A buyer may want to know whether green coffee, a roast coffee or a blend contains robusta. Three kinds of method appear in the sources opened for this entry.

16-OMC by chromatography. EN 18003:2024 specifies a high-performance liquid chromatography (HPLC) method for 16-OMC in green and roasted coffee.17 It is suitable for 40–1,600 mg/kg; its collaborative study showed that 20–40 mg/kg can also be analysed, depending on the laboratory equipment.17 The standard leaves compliance assessment out of scope.17 In Germany, DIN EN 18003:2025-01 replaced DIN 10779:2011-03, a method for roasted coffee only.17 A 2006 review states that the older DIN method detects robusta at less than 2% of a blend.14

16-OMC by NMR. A 2018 study used low-field nuclear magnetic resonance (NMR) on roast coffee.16 It found 16-OMC and 16-O-methylkahweol in arabica for the first time, at an estimated combined level of at least 10 mg/kg in one sample.16 It detected robusta in arabica at about 1–2% w/w. The trace amounts in arabica set a floor near 1%.16 For comparison, the same study cites 600–1,800 mg/kg and 1,204–2,236 mg/kg of 16-OMC in roast robustas from earlier work.16

DNA. A 2018 method from IRD, the French research institute for development, uses high-resolution melting analysis of DNA from green and roasted coffee.18 It targets differences in chloroplast DNA rather than nuclear DNA, to make detection more sensitive. It identifies the species and, in a blend, estimates the percentage of robusta.18

Imaging. The 2022 imaging study combined spectral, shape and colour features of roasted beans, averaged over dishes of 20 beans. Its model classified every dish in the test set correctly and predicted the robusta share in mixtures.10 The authors present it as a rapid screening tool.10

Method What it detects Working range or limit Source
HPLC, EN 18003:2024 16-OMC in green and roasted coffee 40–1,600 mg/kg; 20–40 mg/kg depending on equipment CEN17
HPLC, DIN 10779 (withdrawn) 16-OMC in roasted coffee Robusta below 2% of a blend Speer & Kölling-Speer14; status DIN17
Low-field NMR 16-OMC and 16-O-methylkahweol in roast coffee Robusta at about 1–2% w/w; not below about 1% Gunning et al.16
High-resolution melting of chloroplast DNA Species identity; robusta share in blends Range not stated in the abstract Combes et al.18
Multispectral imaging of roasted beans Species of a dish of 20 beans; share in mixtures 100% correct in the test set Mihailova et al.10

Open questions#

  • Sensory comparison under one protocol. No study opened for this entry compared arabica and robusta cups with one sensory method; the cup descriptions rest on a textbook chapter.15
  • Bean shape across origins. The bean measurements come from one study of roasted beans; how far shape separates the species across origins, grades and green coffee is not established in the sources opened here.10
  • Detection floor for robusta. If arabica holds traces of 16-OMC, chemical tests cannot detect robusta much below about 1%; DNA methods avoid that marker, but their limit was not given in the abstract read for this entry.16,18
  • Robusta’s temperature optimum outside South-East Asia. The 2020 optimum comes from farms in one region (see Contested).7

Key studies#

Campa et al., 2004 – Sucrose and trigonelline in wild coffee species. Measured sucrose and trigonelline in seeds of 14 wild species and six further taxa, four trees each. Arabica held more of both than canephora, with one of the highest sucrose contents of all species tested.

Food Chemistry 88(1), 39–43 · doi:10.1016/j.foodchem.2004.01.020

Gunning et al., 2018–16-OMC in arabica. Analysed arabica and robusta roast coffees with low-field NMR and other methods. Found 16-O-methylcafestol and 16-O-methylkahweol in arabica for the first time, which limits the detection of robusta in blends to about 1% and above.

Food Chemistry 248, 52–60 · doi:10.1016/j.foodchem.2017.12.034

Combes, Joët & Lashermes, 2018 – A DNA test for robusta in arabica. Developed a high-resolution melting method on chloroplast DNA for green and roasted coffee. It identifies the species and estimates the share of robusta in a blend.

Food Control 88, 198–206 · doi:10.1016/j.foodcont.2018.01.014

Mihailova et al., 2022 – Imaging roasted beans. Measured spectral, shape and colour features of 3,500 roasted beans from 21 arabica and 14 robusta samples. Robusta beans were shorter and rounder on average, and a model built on selected features classified every dish of 20 beans in the test set correctly.

Food Chemistry: X 14, 100325 · doi:10.1016/j.fochx.2022.100325

Kath et al., 2020 – Not so robust. Modelled yields from 798 robusta farms in South-East Asia against temperature. Found an optimum below 20.5 °C, lower than earlier estimates, which questions the view of robusta as a warm-lowland crop.

Global Change Biology 26(6), 3677–3688 · doi:10.1111/gcb.15097

Contested
  • Chlorogenic acids by species: a 2006 review gives 4–8.4% of dry matter for *C. arabica* and 7–14.4% for *C. canephora*12, and a roasting chemistry chapter calls robusta significantly richer15; a 2021 meta-analysis in a master’s thesis found the chlorogenic acid profiles of the two species extremely similar, with roast level mattering more than species except for total feruloylquinic and dicaffeoylquinic acids, which were highest in green robusta.19
  • 16-O-methylcafestol in arabica: a 2006 review reports that its absence from arabica beans was confirmed14; a 2018 study detected it in arabica roast coffee and concluded that robusta below about 1% of a blend cannot be detected this way.16
  • Optimum temperature of robusta: a 2006 review gives 22–26 °C as the optimum annual mean, or 24–30 °C after another author6; a 2020 study of farm yields in South-East Asia puts it below 20.5 °C.7
Share or cite this entry

Change log

  • – Corrected the 2022 imaging study throughout (body, figure legend, comparison table, methods table, study card) – the counts 105 and 70 are dishes of 20 roasted beans from 21 arabica and 14 robusta samples, not single beans, and the model classified dishes, not individual beans, using spectral, shape and colour features together. Reason: Table 1 and Methods of the full text (PMC9108882) state the sampling unit; the entry had described the means as single-bean measurements.
  • – Replaced the SCA News interview as the source for the meta-analysis view on chlorogenic acids with the author's master's thesis (UC Davis 2021), and restated the view from the thesis (profiles extremely similar; roast level matters more than species except for total feruloylquinic and dicaffeoylquinic acids). Reason: The interview is a news item (Tier C-like) and may not stand alone (owner ruling, 22 Oct 2026).

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. ADenoeud, F.; Carretero-Paulet, L.; Dereeper, A.; et al. (2014). The coffee genome provides insight into the convergent evolution of caffeine biosynthesis. Science 345(6201), 1181–1184. doi:10.1126/science.1255274 · other copies
  3. ARoyal Botanic Gardens, Kew (2026). Coffea arabica L.. Plants of the World Online (World Checklist of Vascular Plants), description from Flora of Tropical East Africa (Verdcourt 1976).
  4. ARoyal Botanic Gardens, Kew (2026). Coffea canephora Pierre ex A.Froehner. Plants of the World Online (World Checklist of Vascular Plants), description from Flora of Tropical East Africa (Verdcourt 1976).
  5. BHerrera, J. C.; Lambot, C. (2017). The Coffee Tree – Genetic Diversity and Origin. In: Folmer, B. (ed.), The Craft and Science of Coffee, Academic Press, ch. 1, pp. 1–16. doi:10.1016/B978-0-12-803520-7.00001-3 · other copies
  6. ADaMatta, F. M.; Ramalho, J. D. C. (2006). Impacts of drought and temperature stress on coffee physiology and production: a review. Brazilian Journal of Plant Physiology 18(1), 55–81. doi:10.1590/S1677-04202006000100006 · other copies
  7. AKath, J.; Byrareddy, V. M.; Craparo, A.; Nguyen-Huy, T.; Mushtaq, S.; Cao, L.; Bossolasco, L. (2020). Not so robust: robusta coffee production is highly sensitive to temperature. Global Change Biology 26(6), 3677–3688. doi:10.1111/gcb.15097 · other copies
  8. 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
  9. AColodetti, T. V.; Tomaz, M. A.; Sartori, L.; de Salles, R. A.; de Souza, I. V.; Senra, J. F. de B.; Rodrigues, W. N. (2025). From anthesis to harvest: evolution of the fructification of Conilon coffee genotypes grown at a transitional altitude. Bragantia 84, e20250028. doi:10.1590/1678-4499.20250028 · other copies
  10. AMihailova, A.; Liebisch, B.; Islam, M. D.; Carstensen, J. M.; Cannavan, A.; Kelly, S. D. (2022). The use of multispectral imaging for the discrimination of Arabica and Robusta coffee beans. Food Chemistry: X 14, 100325. doi:10.1016/j.fochx.2022.100325 · other copies
  11. AOlechno, E.; Puścion-Jakubik, A.; Zujko, M. E.; Socha, K. (2021). Influence of Various Factors on Caffeine Content in Coffee Brews. Foods 10(6), 1208. doi:10.3390/foods10061208 · other copies
  12. AFarah, A.; Donangelo, C. M. (2006). Phenolic compounds in coffee. Brazilian Journal of Plant Physiology 18(1), 23–36. doi:10.1590/S1677-04202006000100003 · other copies
  13. ACampa, C.; Ballester, J. F.; Doulbeau, S.; Dussert, S.; Hamon, S.; Noirot, M. (2004). Trigonelline and sucrose diversity in wild Coffea species. Food Chemistry 88(1), 39–43. doi:10.1016/j.foodchem.2004.01.020 · other copies
  14. ASpeer, K.; Kölling-Speer, I. (2006). The lipid fraction of the coffee bean. Brazilian Journal of Plant Physiology 18(1), 201–216. doi:10.1590/S1677-04202006000100014 · other copies
  15. BPoisson, L.; Blank, I.; Dunkel, A.; Hofmann, T. (2017). The Chemistry of Roasting – Decoding Flavor Formation. In: Folmer, B. (ed.), The Craft and Science of Coffee, Academic Press, ch. 12, pp. 273–309. doi:10.1016/B978-0-12-803520-7.00012-8 · other copies
  16. AGunning, Y.; Defernez, M.; Watson, A. D.; Beadman, N.; Colquhoun, I. J.; Le Gall, G.; Philo, M.; Garwood, H.; Williamson, D.; Davis, A. P.; Kemsley, E. K. (2018). 16-O-methylcafestol is present in ground roast Arabica coffees: Implications for authenticity testing. Food Chemistry 248, 52–60. doi:10.1016/j.foodchem.2017.12.034 · other copies
  17. AEuropean Committee for Standardization (CEN); published as BS EN 18003:2024 and DIN EN 18003:2025-01 (2024). EN 18003:2024 Food authenticity – Determination of 16-O-Methylcafestol content of green and roasted coffee – HPLC-method. BSI Knowledge (product page with scope).
  18. ACombes, M.-C.; Joët, T.; Lashermes, P. (2018). Development of a rapid and efficient DNA-based method to detect and quantify adulterations in coffee (Arabica versus Robusta). Food Control 88, 198–206. doi:10.1016/j.foodcont.2018.01.014 · other copies
  19. BYeager, S. E. (2021). A Meta-Analysis of Acids in Coffee and a Quantification of Coffee Beverage Color. University of California, Davis (MSc thesis; chapter 2 is the meta-analysis later published by Yeager, Batali, Guinard and Ristenpart in Critical Reviews in Food Science and Nutrition).

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 Arabica and robusta compared. Tell us what you would change and why. A source that supports it helps most.