Diversity in coffee assessed with SSR markers:structure of the genus Coffea and perspectivesfor breeding

Philippe Cubry, Pascal Musoli, Hyacinte Legnate, David Pot, Fabien de Bellis,Valerie Poncet, Francois Anthony, Magali Dufour, and Thierry Leroy

Abstract: The present study shows transferability of microsatellite markers developed in the two cultivated coffee species(Coffea arabica L. and C. canephora Pierre ex Froehn.) to 15 species representing the previously identified main groupsof the genus Coffea. Evaluation of the genetic diversity and available resources within Coffea and development of molecu-lar markers transferable across species are important steps for breeding of the two cultivated species. We worked on 15species with 60 microsatellite markers developed using different strategies (SSR-enriched libraries, BAC libraries, gene se-quences). We focused our analysis on 4 species used for commercial or breeding purposes. Our results establish the hightransferability of microsatellite markers within Coffea. We show the large amount of diversity available within wild spe-cies for breeding applications. Finally we discuss the consequences for future comparative mapping studies and breedingof the two cultivated species.

Key words: SSR markers, microsatellites, Coffea, transferability, cross-amplification, genetic diversity.

Resume : La presente etude montre la transferabilite de marqueurs microsatellites developpes sur les deux especes de ca-feiers cultivees (Coffea arabica L. et C. canephora Pierre ex Froehn.) a 15 especes representant les principaux groupesprecedemment identifies du genre Coffea. L’evaluation de la diversite et des ressources genetiques disponibles au sein dugenre Coffea et le developpement de marqueurs moleculaires transferables d’une espece a l’autre sont des etapes importan-tes pour l’amelioration de ces deux especes. Nous avons travaille sur 15 especes avec 60 marqueurs microsatellites deve-loppes suivant differentes methodologies (banques enrichies en microsatellites, banques BAC, sequences de genes). Nousavons plus particulierement analyse quatre especes d’interet en commerce ou en amelioration. Nos resultats etablissent queles microsatellites sont hautement transferables dans le genre Coffea. Nous mettons en evidence l’important reservoir dediversite pour l’amelioration que constituent les especes sauvages de ce genre. Enfin nous discutons des implications pourde futures etudes de cartographie comparee et l’amelioration des deux especes cultivees.

Mots-cles : marqueurs microsatellites, Coffea, transferabilite, amplification croisee, diversite genetique.

Introduction

The genus Coffea consists of 103 species (Davis and Stof-felen 2006) originated from intertropical regions of Africaand Madagascar. Only two species are cultivated: C. arabicaL., which represents 65% of the world’s coffee production,and C. canephora Pierre ex Froehn. Coffee species are dip-loid (2n = 2x = 22) except for C. arabica, which is tetra-ploid (2n = 4x = 44). Coffea arabica is self-compatible,like two diploid species, C. heterocalyx Stoff. and C. antho-nyi Stoff. & F.Anthony (Davis and Stoffelen 2006). Pre-

vious phylogenetic studies based on other markers such asrDNA (Lashermes et al. 1997) and cpDNA variation (Cros etal. 1998) have shown that the genus Coffea is organizedinto 4 groups with different geographical origins, i.e., Centraland West Africa (WC clade), East Africa (E clade), Cen-tral Africa (C clade), and Madagascar (M clade).

Microsatellite markers present different properties thanthe other markers previously used (such as RFLPs, iso-zymes, and cpDNA) and give a complementary view of thecoffee genus diversity. SSR (simple sequence repeat) ormicrosatellite markers are highly variable and codominant

Received 5 April 2007. Accepted 10 October 2007. Published on the NRC Research Press Web site at genome.nrc.ca on 18 December2007.

Corresponding Editor: F. Belzile.

P. Cubry,1 D. Pot, F. de Bellis, M. Dufour, and T. Leroy. CIRAD, UMR DAP, TA A-96/03, avenue Agropolis, 34398 MontpellierCEDEX 5, France.P. Musoli. Coffee Research Institute, P.O. Box 185, Mukono, Uganda.H. Legnate. CNRA, BP 808, DIVO, Republique de Cote d’Ivoire.V. Poncet. IRD, UMR DIA-PC, 911 avenue Agropolis, BP 64501, 34394 Montpellier CEDEX 5, France.F. Anthony. IRD, UMR RPB, 911 avenue Agropolis, BP 64501, 34394 Montpellier CEDEX 5, France.

1Corresponding author (e-mail: philippe.cubry@cirad.fr).

50

Genome 51: 50–63 (2008) doi:10.1139/G07-096 # 2007 NRC Canada

(Tautz and Renz 1984; Jarne and Lagoda 1996). They havealready been analysed for their transferability within the cof-fee genus for 6 species, C. canephora, C. eugenioidesS.Moore, C. heterocalyx, C. liberica Bull ex Hiern.,C. anthonyi, and C. pseudozanguebariae Bridson (Poncet etal. 2004), and compared with AFLPs (Prakash et al. 2005).SSR markers have also been used to assess genetic diversityin the two main cultivated species, C. arabica and C. cane-phora (Anthony et al. 2002a, 2002b; Moncada andMcCouch 2004; Cubry et al. 2005; Prakash et al. 2005).The present study gives cross-amplification results for a setof microsatellite markers in a larger sample of species andindividuals.

In addition to a large survey of the transferability of themarkers, we performed a detailed analysis of the two culti-vated species (C. arabica and C. canephora) and two relatedspecies used for both quality and productivity improvement(C. liberica and C. congensis). A crisis of low prices hasoccurred during past years, and farmers have to produce abetter quality coffee to maintain their incomes. Identifyingthe amount of genetic diversity available for improvementis especially important for C. arabica, which has been iden-tified as a species with a very narrow genetic base (Anthonyet al. 2002a). Since the genus Coffea diverged recently fromothers (5 to 25 million years ago; Lashermes et al. 1996),most of the species are genetically highly related and a lotof hybridizations are possible (Louarn 1992). Indeed, spon-taneous and viable crosses of C. canephora � C. congensis,C. arabica � C. liberica, and C. arabica � C. canephorahave been described (Cramer 1948; Prakash et al. 2002).These hybrids are widely used in breeding programs for re-sistance to pests and disease or for quality improvement.

In the present paper, we analyse the diversity of 15 Coffeaspecies belonging to the 4 previously identified geneticgroups using 60 microsatellite markers from different ori-gins and covering the whole genome. We also detail the re-lationships among 4 species, 2 cultivated and 2 related wildones. Finally, we discuss the consequences for breeding ofC. arabica and C. canephora.

Materials and methods

Plant materialWe used a total of 42 individuals from 15 Coffea species

in our study (Table 1). Four species of particular interestwere represented by more than 4 individuals to enable com-parison of several diversity variables. These 4 species wereC. canephora, C. arabica, C. congensis, and C. liberica.

For C. arabica, we studied both cultivated and wild ac-cessions, including commercial hybrids between the twomain cultivars, ‘Typica’ and ‘Bourbon’. For C. canephoraand C. liberica, we analysed, respectively, genotypes fromdifferent genetic groups (B, C, SG2, and Guinean) and vari-eties (liberica, dewevrei) chosen to represent the greatest di-versity (Louarn 1992; Anthony 1992; Montagnon 2000;Dussert et al. 2003). Coffea canephora accessions also in-cluded new material from Uganda (Musoli et al. 2006), in-cluding wild material surveyed in Itwara Forest (UW) andthe cultivar ‘Nganda’ (UN). Coffea congensis was repre-sented by accessions from different Central African regions.

Eleven other species from different geographic origins

covering the whole repartition of Coffea were included toprovide an overview of the global diversity, including atleast 2 species of each of the previously described diversityclades (i.e., C, WC, E, and M).

Coffea canephora genotypes were kindly provided by theCNRA (Centre National de Recherche Agronomique) fromfield collection in Divo (Republique de Cote d’Ivoire). WildC. canephora (UW) and ‘Nganda’ (UN) genotypes fromUganda were conveniently provided by the CORI (CoffeeResearch Institute) of Uganda. Coffea arabica, C. congensis,C. liberica, and C. sessiliflora Bridson genotypes came fromfield collections in French Guiana. One individual of each ofthese 4 species was kindly provided by the IRD (Institut deRecherche pour le Developpement) greenhouse collection inMontpellier, France. Material of 9 other species also camefrom the IRD collection: C. anthonyi, previously known asC. ‘sp. Moloundou’, C. bertrandii A.Chev., C. eugenioides,C. humilis A.Chev., C. millotii J.-F.Leroy, C. pseudozangue-bariae, C. racemosa Lour., C. salvatrix Swynn. & Philipson,and C. stenophylla G.Don.

DNA extractionGenomic DNA was extracted from ground leaves follow-

ing an extraction method using a MATAB buffer adaptedfrom Risterucci et al. (2000). A purification of the extractsusing products from the solution-based Wizard1 SV Ge-nomic DNA Purification System (Promega, Madison, Wis-consin, USA, Cat. No. A1125) was then performed.

Microsatellite markersIn this study, we used microsatellite markers obtained

from different origins (Table 2). DLxxx primers were previ-ously published and developed from a C. canephora BAClibrary (Leroy et al. 2005). A second set came from a micro-satellite motif–enriched library of C. canephora clone 126(Dufour et al. 2001) and from an enriched library ofC. arabica ‘Caturra’ (Rovelli et al. 2000). Primers for theenriched C. arabica library came from Poncet et al. (2004)and primers for the enriched C. canephora library were de-signed by Poncet et al. (2007) using Primer3 software(Rozen and Skaletski 2000). SSRxxx primers were designedfrom sequences of sucrose synthase (SuSy) genes (Geromelet al. 2006) using Primer3 (D. Pot, unpublished data). A to-tal of 60 loci were screened in this study and all of them,except SSRxxx loci, have been mapped on an intraspecificC. canephora genetic map (T. Leroy, unpublished data).

PCR and data acquisitionFor each reaction, 2.5 ng of DNA template was mixed

with 5 mL of PCR buffer (10 mmol/L Tris-HCl, 50 mmol/LKCl, 2 mmol/L MgCl2, 0.001% glycerol), 200 mmol/L dNTPs,0.10 mmol/L of reverse primer, 0.08 mmol/L of forwardprimer tailed with M13 sequence, 0.10 mmol/L of fluores-cently labelled M13 primer, and 0.1 U of Taq DNA poly-merase. PCR amplifications were performed in anEppendorf Mastercycler ep 384 (Eppendorf, Westbury,New York, USA). The amplification program consisted ofan initial denaturation cycle of 4 min at 94 8C followedby 9 cycles of ‘‘touch-down’’ PCR consisting of 45 s at94 8C, 1 min at 60 8C to 55 8C, decreasing by 0.5 8Ceach cycle, and 1 min 30 s at 72 8C. The next 26 cycles

Cubry et al. 51

# 2007 NRC Canada

consisted of 94 8C for 45 s, 55 8C for 1 min, and 72 8Cfor 1 min 30 s prior to a final elongation step at 72 8Cfor 5 min.

Fluorescently labelled PCR products were analysed byelectrophoresis on a 6.5% polyacrylamide gel using aLI-COR1 4300 automated sequencer (LI-COR Biosciences,Lincoln, Nebraska, USA). Gel images were retrieved andannotated with the manufacturer’s program SAGAGT. Weassigned allele sizes manually to each individual on thebasis of the automated analyses of SAGAGT. Previouslystudied individuals of C. canephora (Cubry et al. 2005)served as controls. The data matrix was exported as a text

file and formatted in Excel1 software for the different pro-grams used for the analysis.

Data analysisA dissimilarity matrix was computed from the data file

using the software DARwin 5 (Perrier et al. 2003). The dis-similarities were calculated using a simple matching dis-tance index. Since C. arabica exhibited a maximum of 2alleles per locus in our data, we decided to manage geno-types from this species as diploid genotypes. The dissimilar-ity matrix was used to infer a global diversity tree using theweighted neighbor-joining method (Saitou and Nei 1987) as

Table 1. List of plant material and providers.

Coffea species Working name Variety or diversity group Collection

Species of particular interest for commercial or breeding purposesC. arabica Arabica_1 ‘Caturra’ IRD, FranceC. arabica Arabica_2 ‘Red Catuaı 1’ CIRAD, French GuianaC. arabica Arabica_3 ‘Guinee pita 1’ CIRAD, French GuianaC. arabica Arabica_4 ‘Sidamo 1’ CIRAD, French GuianaC. arabica Arabica_5 ‘Mundo Novo’ CIRAD, French GuianaC. arabica Arabica_et1 Wild ethiopian CIRAD, French GuianaC. arabica Arabica_et2 Wild ethiopian CIRAD, French GuianaC. arabica Arabica_et3 Wild ethiopian CIRAD, French GuianaC. canephora Can_b1 Congolese group B CNRA, Republique de Cote d’IvoireC. canephora Can_c1 Congolese group C CNRA, Republique de Cote d’IvoireC. canephora Can_sg2_1 Congolese group SG2 CNRA, Republique de Cote d’IvoireC. canephora Can_g1 Guinean CNRA, Republique de Cote d’IvoireC. canephora Can_g2 Guinean CNRA, Republique de Cote d’IvoireC. canephora Can_u1 Uganda, ‘Nganda’ CORI, UgandaC. canephora Can_u2 Uganda, wild CORI, UgandaC. canephora Can_u3 Uganda, wild CORI, UgandaC. canephora Can_g3 Guinean CNRA, Republique de Cote d’IvoireC. congensis Congensis_1 IRD, FranceC. congensis Congensis_2 CIRAD, French GuianaC. congensis Congensis_3 CIRAD, French GuianaC. congensis Congensis_4 CIRAD, French GuianaC. congensis Congensis_5 CIRAD, French GuianaC. liberica Liberica_1 IRD, FranceC. liberica Liberica_2_l liberica CIRAD, French GuianaC. liberica Liberica_3_l liberica CIRAD, French GuianaC. liberica Liberica_4_l liberica CIRAD, French GuianaC. liberica Liberica_5_d dewevrei CIRAD, French GuianaC. liberica Liberica_6_d dewevrei CIRAD, French GuianaC. liberica Liberica_7_d dewevrei CIRAD, French Guiana

Other species included in this studyC. anthonyi Anthonyi IRD, FranceC. bertrandii Bertrandii IRD, FranceC. brevipes Brevipes IRD, FranceC. eugenioides Eugenioides IRD, FranceC. humilis Humilis IRD, FranceC. milloti Milloti IRD, FranceC. pseudozanguebariae Pseudozanguebariae IRD, FranceC. racemosa Racemosa IRD, FranceC. salvatrix Salvatrix IRD, FranceC. sessiliflora Sessiliflora_1 IRD, FranceC. sessiliflora Sessiliflora_2 CIRAD, French GuianaC. sessiliflora Sessiliflora_3 CIRAD, French GuianaC. stenophylla Stenophylla IRD

52 Genome Vol. 51, 2008

# 2007 NRC Canada

Tab

le2.

Lis

tof

the

60SS

Rm

arke

rsus

edin

the

stud

y.

EM

BL

acc.

No.

Mar

ker

nam

eR

epea

tty

peN

o.of

repe

ats

Prim

erse

quen

ces

(5’?

3’)

Sequ

ence

orig

inPr

imer

orig

inSp

ecie

sof

orig

inA

J250

257

257

CA

9F:

GA

CC

AT

TA

CA

TT

TC

AC

AC

AC

Com

bes

2000

Ponc

et20

04C

offe

aar

abic

a‘C

atur

ra’

R:

GC

AT

TT

TG

TT

GC

AC

AC

TG

TA

AM

2311

8630

5T

G8

F:A

AC

TT

CA

CT

AA

TC

TG

TT

GT

TG

CT

GD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

GC

AC

AT

CT

AT

CC

AT

CT

TT

TG

GA

M23

1546

327

CA

9F:

GG

CT

CA

AA

AT

CA

CC

CT

TT

GT

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:C

TA

GG

AT

CG

TG

GC

AG

AA

GA

AG

AM

2315

4732

9G

T10

F:A

CT

CA

GA

CA

AA

CC

CT

TC

AA

CD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

GA

TG

TT

TT

GC

AT

CT

AT

TT

GG

AM

2315

4833

4A

C8

F:T

AT

GC

CT

CA

GC

AC

CT

AT

CT

AD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

TA

CT

TC

CC

CT

GT

TC

CT

TA

TG

AM

2315

4934

1C

A,

TA

12,

5F:

CA

TT

GG

TG

TC

AA

GG

GT

CA

AG

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:A

AA

GT

AT

CA

GA

AG

GA

AA

AG

TC

TC

GT

AA

AM

2315

5035

0G

T8

F:T

CA

AA

AG

AG

GG

CA

CG

AA

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:A

CG

AC

AA

TA

AC

TT

TG

CA

TG

TC

TA

M23

1551

351

GT

13F:

AA

GG

AT

GG

CA

AG

TG

GA

TT

TC

TD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

GC

AG

CT

CT

TG

AT

TG

TA

GT

TT

CG

TA

M23

1552

355

TG

15F:

CT

AT

GA

TG

TC

TT

CC

AA

CC

TT

CT

AA

CD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

GG

TC

CA

AT

TC

TG

TT

TC

AA

TT

TC

AM

2315

5335

6T

G14

F:T

GA

AG

TC

AA

CC

TG

AA

TA

CC

AG

AD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

AC

GC

AC

GC

AC

GA

AT

GA

M23

1554

358

CA

11F:

CA

TG

CA

CT

AT

TA

TG

TT

TG

TG

TT

TT

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:T

CT

CG

TC

AT

AT

TT

AC

AG

GT

AG

GT

TA

M23

1555

360

CA

10F:

AC

AG

TA

GT

AT

TT

CA

TG

CC

AC

AT

CC

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:A

CA

TT

TG

AT

TG

CC

TC

TT

GA

CC

AM

2315

5636

4A

21F:

AG

AA

GA

AT

GA

AG

AC

GA

AA

CA

CA

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:T

AA

CG

CC

TG

CC

AT

CG

AM

2315

5736

7A

C12

F:T

CA

AT

CC

CT

GT

AT

TC

CT

GT

TT

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:C

TA

GG

CA

CT

TA

AA

AT

CT

CT

AT

AA

CG

AM

2315

5836

8T

G13

F:C

AC

AT

CT

CC

AT

CC

AT

AA

CC

AT

TT

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:T

CC

TA

CC

TA

CT

TG

CC

TG

TG

CT

AM

2315

5937

1C

A9

F:A

GA

CA

CA

CA

AG

GC

AA

TA

AT

CA

AA

CD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

TC

TT

GA

GC

AG

CA

TG

GG

AA

CA

M23

1560

384

AC

10F:

AC

GC

TA

TG

AC

AA

GG

CA

AT

GA

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:T

GC

AG

TA

GT

TT

CA

CC

CT

TT

AT

CC

AM

2315

6138

8C

A9

F:A

TG

AA

AC

GA

GA

AT

CC

AT

AC

CC

TA

CD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

AG

AG

GT

AA

AA

GG

AA

AA

TG

CT

AG

AC

CA

M23

1562

392

TC

16F:

AA

GG

TA

TT

GG

TC

TG

CC

TT

TG

TD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

CT

AA

CC

CT

AA

TC

CC

CA

GC

AA

M23

1563

394

TG

9F:

GC

CG

TC

TC

GT

AT

CC

CT

CA

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:G

AA

GC

CA

GA

AA

GT

CA

GT

CA

CA

TA

GA

M23

1564

395

GT

13F:

CA

TC

AT

TT

TG

TT

GG

CA

AA

GD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

TG

GT

TA

TT

TC

CT

TC

TT

TG

TA

TT

G

Cubry et al. 53

# 2007 NRC Canada

Tab

le2

(con

tinu

ed).

EM

BL

acc.

No.

Mar

ker

nam

eR

epea

tty

peN

o.of

repe

ats

Prim

erse

quen

ces

(5’?

3’)

Sequ

ence

orig

inPr

imer

orig

inSp

ecie

sof

orig

in

AM

2315

6542

9A

13F:

CA

TT

CG

AT

GC

CA

AC

AA

CC

TD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

GG

GT

CA

AC

GC

TT

CT

CC

TG

AM

2315

6644

2C

A19

F:C

GC

AA

AT

CT

GA

GT

AT

CC

CA

AC

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:T

GG

AT

CA

AC

AC

TG

CC

CT

TC

AM

2315

6744

5A

C10

F:C

CA

CA

GC

TT

GA

AT

GA

CC

AG

AD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

AA

TT

GA

CC

AA

GT

AA

TC

AC

CG

AC

TA

M23

1568

456

AC

14F:

TG

GT

TG

TT

TT

CT

TC

CA

TC

AA

TC

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:T

CC

AG

TT

TC

CC

AC

GC

TC

TA

M23

1569

460

CA

11F:

TG

CC

TT

CA

AA

AT

GC

TC

TA

TA

AC

CD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

GC

TG

AT

AT

TC

TT

GG

AT

GG

AG

TT

GA

M23

1570

461

AC

9F:

CG

GC

TG

TG

AC

TG

AT

GT

GD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

AA

TT

GC

TA

AG

GG

TC

GA

GA

AA

M23

1571

463

AC

8F:

CA

TT

CT

TC

CC

AC

GA

TT

CT

AT

CT

CD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

GT

GA

CT

TT

CG

GT

TG

AA

AT

AC

TG

GA

M23

1572

471

CT

12F:

TT

AC

CT

CC

CG

GC

CA

GA

CD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

CA

GG

AG

AC

CA

AG

AC

CT

TA

GC

AA

M23

1573

472

CA

,T

A8,

8F:

AA

TC

AT

GG

GG

AC

AG

GA

CA

AG

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:T

CT

GC

TA

GA

CT

TG

AC

AT

CT

TT

TG

GA

M23

1574

477

AC

16F:

CG

AG

GG

TT

GG

GA

AA

AG

GT

Duf

our

2001

Ponc

et20

07C

offe

aca

neph

ora,

clon

e12

6R

:A

CC

AC

CT

GA

TG

TT

CC

AT

TT

GT

AM

2315

7549

5A

C8

F:C

AT

GG

AT

GG

GA

AG

GC

AG

TD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

CT

TG

GA

AA

AC

TT

GC

TG

AA

TG

TG

AM

2315

7650

1T

G8

F:C

AC

CA

CC

AT

CT

AA

TG

CA

CC

TD

ufou

r20

01Po

ncet

2007

Cof

fea

cane

phor

a,cl

one

126

R:

CT

GC

AC

CA

GC

TA

AT

TC

AA

GC

AJ3

0875

375

3C

A15

F:G

GA

GA

CG

CA

GG

TG

GT

AG

AA

GR

ovel

li20

00Po

ncet

2004

Cof

fea

arab

ica

‘Cat

urra

’R

:T

CG

AG

AA

GT

CT

TG

GG

GT

GT

TA

J308

755

755

CA

20F:

CC

CT

CC

CT

CT

TT

CT

CC

TC

TC

Rov

elli

2000

Ponc

et20

04C

offe

aar

abic

a‘C

atur

ra’

R:

TC

TG

GG

TT

TT

CT

GT

GT

TC

TC

GA

J308

774

774

CT

,C

A5,

7F:

GC

CA

CA

AG

TT

TC

GT

GC

TT

TT

Rov

elli

2000

Ponc

et20

04C

offe

aar

abic

a‘C

atur

ra’

R:

GG

GT

GT

CG

GT

GT

AG

GT

GT

AT

GA

J308

779

779

TG

17F:

TC

CC

CC

AT

CT

TT

TT

CT

TT

CC

Rov

elli

2000

Ponc

et20

04C

offe

aar

abic

a‘C

atur

ra’

R:

GG

GA

GT

GT

TT

TT

GT

GT

TG

CT

TA

J308

782

782

GT

15F:

AA

AG

GA

AA

AT

TG

TT

GG

CT

CT

GA

Rov

elli

2000

Ponc

et20

04C

offe

aar

abic

a‘C

atur

ra’

R:

TC

CA

CA

TA

CA

TT

TC

CC

AG

CA

AJ3

0879

079

0G

T21

F:T

TT

TC

TG

GG

TT

TT

CT

GT

GT

TC

TC

Rov

elli

2000

Ponc

et20

04C

offe

aar

abic

a‘C

atur

ra’

R:

TA

AC

TC

TC

CA

TT

CC

CG

CA

TT

AJ3

0880

980

9T

GA

11F:

AG

CA

AG

TG

GA

GC

AG

AA

GA

AG

Rov

elli

2000

Ponc

et20

04C

offe

aar

abic

a‘C

atur

ra’

R:

CG

GT

GA

AT

AA

GT

CG

CA

GT

CA

J308

837

837

TG

,G

A16

,11

F:C

TC

GC

TT

TC

AC

GC

TC

TC

TC

TR

ovel

li20

00Po

ncet

2004

Cof

fea

arab

ica

‘Cat

urra

’R

:C

GG

TA

TG

TT

CC

TC

GT

TC

CT

CA

J308

838

838

AC

9F:

CC

CG

TT

GC

CA

TC

CT

TA

CT

TA

Rov

elli

2000

Ponc

et20

04C

offe

aar

abic

a‘C

atur

ra’

R:

AT

AC

CC

GA

TA

CA

TT

TG

GA

TA

CT

CG

54 Genome Vol. 51, 2008

# 2007 NRC Canada

Tab

le2

(con

clud

ed).

EM

BL

acc.

No.

Mar

ker

nam

eR

epea

tty

peN

o.of

repe

ats

Prim

erse

quen

ces

(5’?

3’)

Sequ

ence

orig

inPr

imer

orig

inSp

ecie

sof

orig

in

AJ8

7188

2D

L00

3C

AA

T5

F:T

AA

CA

GA

AG

CA

CC

AA

AA

CC

Ler

oy20

05L

eroy

2005

Cof

fea

cane

phor

a,cl

one

126

R:

TC

TA

AA

CC

CA

CC

TC

AC

AA

CA

J871

889

DL

010

A14

F:T

AG

TC

CC

TT

TT

CA

GT

GG

TL

eroy

2005

Ler

oy20

05C

offe

aca

neph

ora,

clon

e12

6R

:T

TT

CT

TT

GT

TA

CG

GA

GT

GA

J871

890

DL

011

GC

T,

CA

T4,

8F:

AT

AC

AT

AA

GC

AA

GC

AC

TG

AL

eroy

2005

Ler

oy20

05C

offe

aca

neph

ora,

clon

e12

6R

:C

AG

AA

CA

AA

TG

AA

AT

GG

AA

J871

892

DL

013

CA

,C

T6,

8F:

AG

AG

GG

AT

GT

CA

GC

AT

AA

Ler

oy20

05L

eroy

2005

Cof

fea

cane

phor

a,cl

one

126

R:

AT

TT

GT

GT

TT

GG

TA

GA

TG

TG

AJ8

7189

9D

L02

0T

23F:

TG

CT

CA

AA

CT

TC

TT

GC

TL

eroy

2005

Ler

oy20

05C

offe

aca

neph

ora,

clon

e12

6R

:C

GC

CA

AC

TC

TA

AT

GT

GT

AJ8

7190

4D

L02

5C

17F:

TT

GT

TG

AG

AG

TG

GA

GG

AL

eroy

2005

Ler

oy20

05C

offe

aca

neph

ora,

clon

e12

6R

:C

CA

AA

GA

CA

GT

GC

AG

TA

AA

J871

905

DL

026

A17

F:C

GA

GA

CG

AG

CA

TA

AG

AA

Ler

oy20

05L

eroy

2005

Cof

fea

cane

phor

a,cl

one

126

R:

GC

TG

GA

AT

GA

AG

AA

TG

TA

GA

J871

911

DL

032

TA

CG

3F:

TG

TT

GG

TG

AA

GA

AA

TC

CL

eroy

2005

Ler

oy20

05C

offe

aca

neph

ora,

clon

e12

6R

:A

TG

GA

GA

CA

GG

AA

AT

AA

AC

AM

2315

77SS

R00

1T

3F:

CA

AT

AC

GG

CA

TG

CA

TT

TG

AC

Ger

omel

2006

Pot

2006

Cof

fea

cane

phor

a,cl

one

126

R:

TG

TT

GA

AC

AC

GC

AA

TT

GA

CC

AM

2315

78SS

R00

3A

6F:

AT

TT

GC

GT

GC

TG

GA

TG

TT

TT

Ger

omel

2006

Pot

2006

Cof

fea

cane

phor

a,cl

one

126

R:

AC

CA

TG

TA

GG

AA

GG

CC

AC

AG

AM

2315

79SS

R00

4T

9F:

CC

AA

CC

CT

AA

GA

TG

AT

TT

TT

GT

Ger

omel

2006

Pot

2006

Cof

fea

cane

phor

a,cl

one

126

R:

AA

CC

CC

TC

TC

AA

AA

CC

CA

GT

AM

2315

82SS

R00

5G

AT

2F:

AT

GT

GG

TG

CT

GA

TG

TG

CA

GT

Ger

omel

2006

Pot

2006

Cof

fea

cane

phor

a,cl

one

126

R:

GT

CA

CG

TG

GG

AT

GA

TG

AG

AA

AM

2315

80SS

R00

9G

AA

AA

5F:

CA

AA

CA

AA

AC

AG

TA

CA

AT

TC

AA

TC

CG

erom

el20

06Po

t20

06C

offe

aca

neph

ora,

clon

e12

6R

:A

TC

CC

TG

CG

AG

AC

CT

GA

CT

AA

M23

1581

SSR

010

AT

T2

F:C

GA

AA

GG

AA

CA

CA

GG

AA

CC

AG

erom

el20

06Po

t20

06C

offe

aca

neph

ora,

clon

e12

6R

:C

AG

TG

GT

GA

AC

TT

AA

TC

GT

CC

AA

M23

1583

SSR

014

T14

F:G

GA

TC

TT

AT

CG

CA

AT

GA

AC

CA

Ger

omel

2006

Pot

2006

Cof

fea

cane

phor

a,cl

one

126

R:

CC

AA

CA

GT

GT

CC

TT

GC

TG

AA

AM

2315

84SS

R01

5T

12F:

TT

CT

TC

AC

AA

GA

AC

CA

AC

CC

TA

AG

erom

el20

06Po

t20

06C

offe

aca

neph

ora,

clon

e12

6R

:A

AC

CC

CT

CT

CA

AA

AC

CC

AA

TA

M23

1585

SSR

016

T13

F:T

GG

TC

AA

TT

TG

AA

GC

GA

CT

GG

erom

el20

06Po

t20

06C

offe

aca

neph

ora,

clon

e12

6R

:C

CT

CC

AT

CC

TT

TC

CC

TT

AC

CA

M23

1586

SSR

017

TA

7F:

TG

TT

CC

TC

TG

GC

TG

TT

GA

TG

Ger

omel

2006

Pot

2006

Cof

fea

cane

phor

a,cl

one

126

R:

CC

GG

TT

GA

AT

GA

GG

GT

AA

AG

.

Cubry et al. 55

# 2007 NRC Canada

implemented in DARwin. Five thousand bootstrap iterationswere calculated to test the robustness of the nodes. Consid-ering that some species were represented by more than oneindividual, we inferred another diversity tree with one ran-domly chosen individual per species. This tree allowed abetter understanding of the genetic relationships betweenspecies without the interference of sampling size per spe-cies. The same inference method used for the global treewas used for this second tree.

Several genetic variables (e.g., number of alleles, gene di-versity, and observed heterozygosity) were calculated usingPowerMarker software (Liu and Muse 2005) for the globalsample and for each of the 4 species of particular interest.We also computed the percentage of polymorphic loci byspecies. Ninety-five percent confidence intervals for eachvariable were estimated by performing 5000 bootstrap itera-tions across loci.

Results

Amplifications across the genusThe availability (percentage of amplification) per marker

ranged from 30.9% to 100% among the 42 analysed geno-types, with a mean of 81.5% calculated from the raw matrixof observations (see Table S12). Even if 3 markers appearedto be specific to the Central Africa clade, good transferabil-ity of microsatellites across Coffea species was observed.

The percentage of amplification per individual rangedfrom 51.7% for one C. liberica genotype (note that themean for all C. liberica species is about 72%) to 98.3% forone C. canephora genotype. Values obtained here are closeto those found by Poncet et al. (2004). For the 4 main spe-cies, amplification ranged from 72% for C. liberica to 89%for C. arabica and 90% for C. canephora. Amplification forC. congensis was intermediate (83%).

Genus diversity analysisFigure 1 presents the neighbor-joining tree for the 42 in-

dividuals of the study based on 60 microsatellite loci. Boot-strap values greater than 40 are shown; this threshold wasarbitrarily chosen for the readability of the figure. Ten diver-sity groups were discriminated by the analysis. The 4 ge-netic groups WC, C, E, and M, previously described byLashermes et al. (1997), are indicated on this figure.

Groups C, E, and M were discriminated by our study,whereas species of the WC clade were classified in 7 differ-ent groups. Coffea arabica and C. congensis constitutedoriginal groups, while C. canephora and C. liberica wereeach represented by two groups. These two groups corre-spond to different geographical origins (Central and WestAfrica), as previously described by Berthaud (1986). ForC. liberica, these two groups appear to be the varieties,C. liberica var. liberica and C. liberica var. dewevrei. ForC. canephora the two groups correspond to the Guinean (G)clade and the Congolese clade, including the B and SG2diversity groups. We observed strong relationships betweenB, SG2, and related Ugandan accessions (UW, UN), as pre-

viously described (Musoli et al. 2006). Coffea brevipes canbe grouped with the Central African (Congolese) clade ofC. canephora, while C. humilis and C. stenophylla appearto be grouped.

Within C. arabica, wild and cultivated materials were dif-ferentiated, as expected from previous studies of a smallnumber of SSR markers (Anthony et al. 2002a, 2002b). Thecultivated varieties represent a narrow genetic base, sincedissimilarity distances between those genotypes are theshortest of the dendrogram.

The second tree, considering only one individual per spe-cies, allows us to describe 5 different groups for oursampled species. Groups M, C, and E are still discriminated,while species from West Africa (WC clade) are separatedinto two groups: C. arabica, C. canephora, and the relatedspecies C. congensis and C. brevipes form one group, whileC. liberica, C. humilis, and C. stenophylla form anothergroup. Bootstrap values supporting these groups are quitehigh for microsatellite markers.

The global diversity is high, with a mean gene diversityof 0.72 ± 0.03 and a mean allele number of 10.8 (see Table 3for details). The number of alleles varies from 1 to 22according to the locus considered. Of the total number ofalleles (648), 304 (47%) are specific to one species. A com-plete table of private alleles is given as supplementary mate-rial (Table S22). The percentage of the total number ofprivate alleles for each species ranges from 0% for C. antho-nyi to 31.25% for C. canephora, with a mean of 6.45% (seeTable S32). These results show the great amount of interspe-cific diversity within the genus, even if some species arerepresented by only one individual.

Considering the global sample, 59 markers are polymor-phic. Only one, SSR016, which derived from a genic se-quence, exhibited no polymorphism. At the intraspecificlevel, 91.7%, 75%, 76.7%, and 65% of the markers are poly-morphic in C. canephora, C. congensis, C. liberica, andC. arabica, respectively. For the other species, polymorphisminformation should not be taken into consideration becauseonly one or a small number of individuals are available.

Diversity analysis of several speciesFour species of particular interest because of their eco-

nomic importance or breeding potential were more accu-rately analysed in our study. This subsample of 4 speciescontributed an important part of the global sample diversity,with a mean number of alleles of 8. On the species diversitydiagram (Fig. 2) they appear to be in 2 related clades. Ta-ble 3 presents the results for allele number, gene diversity,and observed heterozygosity for C. arabica, C. canephora,C. congensis, and C. liberica (Table S42 presents values cal-culated for all the species). Coffea arabica shows the lowestdiversity, with a mean number of alleles of 2.10. Moreover,it is the only species that shows gene diversity less than ob-served heterozygosity. The global amount of diversity inthese 4 species is important, with a mean gene diversityhigher than 0.35. Coffea canephora appears to be the most

2 Supplementary data for this article are available on the journal Web site (http://genome.nrc.ca) or may be purchased from the Depositoryof Unpublished Data, Document Delivery, CISTI, National Research Council Canada, Building M-55, 1200 Montreal Road, Ottawa, ONK1A 0R6, Canada. DUD 5250. For more information on obtaining material refer to http://cisti-icist.nrc-cnrc.gc.ca/irm/unpub_e.shtml.

56 Genome Vol. 51, 2008

# 2007 NRC Canada

diverse, with a gene diversity of 0.55 and a mean number ofalleles of 5.00.

Discussion

Coffea diversityThe global amount of diversity within Coffea appears to

be high. Considering the 4 previously described clades, weshow that 3 groups can be confirmed (i.e., groups C, M,and E), while the fourth (WC) appears divided in two(Fig. 2). This division can be imputed to the use of SSRs,which have different properties than the previously usedmarkers, and the high number of markers used in this studycompared with the previous studies. Indeed, the high rate ofmutation for microsatellite markers helps us to better inves-tigate structure within species and species complexes.

Moreover, microsatellites are valuable tools to assess ge-netic structure at the species level, as demonstrated by theglobal diversity diagram (Fig. 1). This figure shows the rela-tionships within 4 species of the WC clade, indicating struc-ture at the intraspecific level for C. liberica, C. canephora,and C. arabica. In contrast, C. congensis appears to be ho-mogeneous, at least for the genotypes studied.

Finally, we validated our sampling strategy, which con-sisted of analysing at least 2 species per previously knowndiversity clade for the whole genus to have an overview ofthe global genus diversity. We sampled more genotypes for4 species particularly well known and of important eco-nomic and breeding interest (Lashermes et al. 1997; An-thony 1992; Poncet et al. 2004).

Our results validate the microsatellite-based approach toquickly study Coffea species by covering the entire genome,while sequence-based studies are generally limited to smallnumbers of genomic regions.

Transferability of microsatellite markersWe have confirmed the transferability of SSR markers

across the genus Coffea for a larger sample of species thanpreviously described. SSRs are useful markers for compara-tive studies across genera (Casasoli 2004). Their transfer-ability over species across a genus has been shown forseveral genera including Lycopersicon (Alvarez et al. 2001),Oryza (Gao et al. 2005), Vigna (Yu et al. 1999), and Coffea(Combes et al. 2000; Poncet et al. 2004). Newly developedmicrosatellites based on C. canephora sequences exhibit thesame properties as those previously developed based on

Fig. 1. Neighbor-joining tree for the 42 individuals analyzed based on the dissimilarity matrix calculated by simple matching. Bootstrapvalues were calculated with 5000 repetitions; only values greater than or equal to 40 are shown.

Cubry et al. 57

# 2007 NRC Canada

Tab

le3.

Sum

mar

yst

atis

tics

calc

ulat

edfo

rth

e60

SSR

mar

kers

for

the

glob

alsa

mpl

e(a

ll15

spec

ies

stud

ied)

,th

e4

spec

ies

focu

sed

on,

and

each

ofth

e4

spec

ies

sepa

rate

ly.

15sp

ecie

sC

.ar

abic

aC

.ca

neph

ora

C.

cong

ensi

sC

.li

beri

ca4

spec

ies

Mar

ker

NG

DH

oN

GD

Ho

NG

DH

oN

GD

Ho

NG

DH

oN

GD

Ho

DL

003

60.

610.

372

0.23

0.29

20.

400.

141

0.00

0.00

30.

370.

503

0.57

0.25

DL

010

120.

770.

362

0.19

0.22

50.

570.

114

0.54

0.80

30.

561.

008

0.72

0.45

DL

011

70.

620.

171

0.00

0.00

40.

610.

223

0.29

0.20

30.

470.

146

0.63

0.16

DL

013

120.

860.

302

0.50

1.00

30.

450.

001

0.00

0.00

30.

510.

007

0.83

0.38

DL

020

130.

860.

434

0.56

0.89

60.

690.

502

0.26

0.00

50.

640.

4311

0.86

0.50

DL

025

70.

780.

343

0.54

1.00

30.

530.

113

0.29

0.20

30.

510.

006

0.74

0.37

DL

026

130.

810.

111

0.00

0.00

50.

680.

004

0.58

0.00

60.

510.

439

0.79

0.10

DL

032

70.

730.

272

0.50

1.00

20.

400.

001

0.00

0.00

40.

500.

405

0.59

0.36

SSR

016

10.

000.

001

0.00

0.00

10.

000.

001

0.00

0.00

10.

000.

001

0.00

0.00

SSR

014

130.

790.

261

0.00

0.00

40.

580.

115

0.51

0.40

60.

640.

439

0.77

0.19

SSR

015

40.

320.

222

0.50

1.00

10.

000.

001

0.00

0.00

10.

000.

002

0.27

0.32

SSR

017

90.

720.

091

0.00

0.00

10.

000.

002

0.16

0.20

60.

680.

147

0.66

0.08

SSR

001

20.

150.

001

0.00

0.00

20.

180.

001

0.00

0.00

0N

AN

A2

0.08

0.00

SSR

003

30.

250.

001

0.00

0.00

20.

410.

002

0.28

0.00

10.

000.

002

0.25

0.00

SSR

004

30.

110.

022

0.10

0.11

10.

000.

001

0.00

0.00

10.

000.

002

0.03

0.03

257

130.

650.

313

0.56

1.00

20.

410.

001

0.00

0.00

30.

420.

258

0.61

0.42

305

70.

570.

402

0.50

1.00

30.

420.

602

0.29

0.40

10.

000.

004

0.34

0.33

327

130.

820.

333

0.55

1.00

60.

680.

294

0.59

0.50

10.

000.

008

0.52

0.14

329

130.

850.

413

0.60

1.00

50.

590.

253

0.52

0.25

40.

480.

296

0.51

0.30

334

40.

580.

102

0.10

0.11

40.

470.

222

0.28

0.00

20.

370.

004

0.59

0.58

341

60.

680.

111

0.00

0.00

30.

470.

002

0.25

0.00

20.

190.

2511

0.82

0.50

350

100.

810.

434

0.69

0.78

30.

540.

295

0.63

0.60

30.

510.

0010

0.83

0.52

351

130.

830.

542

0.50

1.00

60.

590.

715

0.67

0.75

30.

380.

204

0.38

0.10

355

160.

880.

512

0.50

1.00

70.

720.

444

0.56

0.60

50.

660.

716

0.73

0.11

356

140.

810.

594

0.53

0.43

50.

690.

784

0.53

0.67

10.

000.

008

0.78

0.46

358

80.

710.

141

0.00

0.00

40.

590.

221

0.00

0.00

10.

000.

009

0.78

0.73

SSR

009

100.

660.

391

0.00

0.00

40.

620.

563

0.50

0.50

60.

690.

7513

0.87

0.71

SSR

010

40.

300.

291

0.00

0.00

40.

450.

500

NA

NA

0N

AN

A8

0.77

0.62

360

120.

830.

260

NA

NA

60.

690.

222

0.30

0.00

60.

690.

606

0.66

0.09

364

70.

480.

241

0.00

0.00

60.

690.

562

0.38

0.00

30.

480.

4014

0.87

0.33

367

110.

840.

492

0.50

1.00

60.

720.

444

0.56

0.25

40.

560.

337

0.61

0.25

368

210.

880.

291

0.00

0.00

100.

800.

335

0.61

0.25

40.

520.

2510

0.81

0.59

371

110.

780.

542

0.50

1.00

50.

550.

385

0.70

0.80

50.

590.

4313

0.82

0.20

384

90.

830.

212

0.10

0.11

40.

600.

112

0.16

0.20

50.

560.

4310

0.76

0.67

388

180.

870.

313

0.44

0.00

80.

780.

782

0.16

0.20

30.

521.

008

0.82

0.23

392

150.

830.

361

0.00

0.00

50.

620.

134

0.58

0.60

80.

800.

8611

0.82

0.38

394

140.

740.

321

0.00

0.00

50.

430.

444

0.56

0.40

80.

770.

6713

0.80

0.37

395

160.

870.

293

0.21

0.13

90.

780.

572

0.26

0.00

40.

480.

2010

0.64

0.37

429

200.

860.

271

0.00

0.00

80.

790.

563

0.47

0.00

50.

610.

2015

0.87

0.27

442

70.

590.

181

0.00

0.00

60.

690.

292

0.23

0.33

0N

AN

A13

0.81

0.22

445

90.

790.

422

0.50

1.00

30.

490.

333

0.36

0.50

20.

370.

008

0.66

0.17

58 Genome Vol. 51, 2008

# 2007 NRC Canada

Tab

le3

(con

clud

ed).

15sp

ecie

sC

.ar

abic

aC

.ca

neph

ora

C.

cong

ensi

sC

.li

beri

ca4

spec

ies

Mar

ker

NG

DH

oN

GD

Ho

NG

DH

oN

GD

Ho

NG

DH

oN

GD

Ho

456

110.

700.

221

0.00

0.00

110.

810.

440

NA

NA

0N

AN

A5

0.74

0.44

460

220.

880.

542

0.50

1.00

50.

600.

138

0.74

0.80

70.

730.

8011

0.74

0.28

461

130.

860.

414

0.47

0.56

70.

740.

333

0.38

0.20

50.

640.

5718

0.85

0.68

463

70.

730.

592

0.50

1.00

50.

710.

672

0.19

0.25

30.

500.

4014

0.89

0.45

471

110.

810.

261

0.00

0.00

50.

650.

294

0.56

0.25

50.

620.

675

0.65

0.65

472

150.

880.

456

0.69

1.00

60.

720.

254

0.52

0.25

30.

380.

339

0.77

0.32

477

160.

870.

382

0.50

1.00

50.

530.

332

0.26

0.00

30.

490.

0010

0.88

0.54

495

90.

750.

071

0.00

0.00

60.

690.

331

0.00

0.00

10.

000.

0011

0.82

0.43

SSR

005

110.

690.

102

0.10

0.11

30.

500.

003

0.29

0.20

30.

450.

207

0.66

0.10

501

160.

850.

472

0.49

0.89

90.

790.

561

0.00

0.00

50.

510.

5714

0.87

0.58

753

130.

820.

583

0.54

1.00

50.

660.

384

0.65

1.00

40.

530.

678

0.79

0.72

755

150.

870.

593

0.59

1.00

90.

760.

565

0.67

0.75

40.

600.

8013

0.89

0.76

774

80.

580.

122

0.10

0.11

30.

260.

111

0.00

0.00

10.

000.

006

0.53

0.10

779

90.

860.

582

0.50

1.00

70.

740.

384

0.58

0.60

50.

730.

869

0.85

0.71

782

90.

770.

195

0.68

0.80

10.

000.

005

0.64

0.20

40.

540.

206

0.73

0.25

790

160.

880.

563

0.60

1.00

90.

770.

675

0.66

0.60

40.

420.

4314

0.86

0.71

809

80.

710.

532

0.50

1.00

30.

350.

441

0.00

0.00

50.

721.

007

0.67

0.65

837

100.

820.

243

0.29

0.13

60.

700.

433

0.42

0.25

30.

480.

209

0.82

0.25

838

160.

890.

463

0.60

1.00

50.

650.

222

0.28

0.50

30.

450.

2010

0.87

0.50

Mea

n*11

0.72

0.32

20.

300.

495

0.55

0.29

30.

340.

274

0.44

0.34

80.

690.

37M

ean{

110.

720.

322

0.30

0.48

50.

550.

303

0.35

0.27

40.

450.

358

0.69

0.37

SD1

0.03

0.02

00.

030.

060

0.03

0.03

00.

030.

040

0.03

0.04

00.

030.

032.

5%l.b

.10

0.66

0.27

20.

230.

374

0.49

0.24

20.

290.

203

0.39

0.27

70.

630.

3197

.5%

u.b.

120.

760.

362

0.36

0.61

50.

600.

353

0.41

0.34

40.

510.

439

0.74

0.42

Not

e:N

,nu

mbe

rof

alle

les;

GD

,ge

nedi

vers

ity;

Ho,

obse

rved

hete

rozy

gosi

ty;

NA

,no

tav

aila

ble

(mis

sing

data

);SD

,st

anda

rdde

viat

ion;

2.5%

l.b.

and

97.5

%u.

b.,

low

eran

dup

per

boun

dari

esof

the

95%

conf

iden

cein

terv

al.

*M

ean

valu

esba

sed

only

onm

arke

rsw

ithno

mis

sing

data

for

the

cons

ider

edsp

ecie

s.{ M

ean

valu

esca

lcul

ated

over

5000

boot

stra

pite

rati

ons

and

base

don

lyon

mar

kers

with

nom

issi

ngda

tafo

rth

eco

nsid

ered

spec

ies.

Cubry et al. 59

# 2007 NRC Canada

C. arabica sequences, since mean percentages of amplifica-tion are the same. This result will be used for developmentof comparative mapping, utilization of new markers, andknowledge transfer from one species to another.

SSRs described in genes involved in sucrose metabolismappear to have some specific behaviour, since they exhibitvery low diversity (1–4 alleles in the global sample) orintermediate diversity (9–13 alleles). These results willallow us to use these markers to study gene regions impli-cated in sucrose metabolism.

In our work, using new markers, we validate the relationbetween C. anthonyi and C. eugenioides, which was previ-ously described by Lashermes et al. (1997). These two spe-cies show high similarity based on both morphological andmolecular data. However, C. anthonyi originated from Ca-meroon, while C. eugenioides is native to East Africa. Noother coffee species belonging to the same clade (C) hasbeen observed between these two distant geographic areas,and there is no clear explanation for the discontinuous distri-bution of these coffee trees (Anthony 1992).

We can use these two species to improve C. arabica vari-eties, considering their genetic relationships and the originalself-compatible system of C. anthonyi (Anthony et al. 2006).These two species show some of the lowest concentrationsof caffeine (0.6%) of the genus Coffea and exhibit high con-centrations of trigonelline (1.6% for C. anthonyi, 1.3% forC. eugenioides; F. Anthony, personal communication), analkaloid compound. These two characters have always inter-ested breeders in coffee improvement. Meanwhile, since few

genotypes are in collection worldwide, these two specieshave not been agronomically well characterized and experi-ments are necessary to assess potential resistances to bioticand abiotic stresses usable for improvement.

On the other hand, part of the C. arabica genome hasbeen shown to originate from an ancestral species geneti-cally close to C. eugenioides or C. anthonyi (Lashermes etal. 1999). These relationships can be used to better under-stand the elaboration and functioning of the allotetraploidgenome of C. arabica, in particular comportment of homeol-ogous chromosomes during meiosis.

Diversity and genetic properties of cultivated and relatedwild species

The diversity and genetic relationships of C. arabica,C. canephora, and related species are examined in ourwork. Coffea arabica has been treated as a diploid speciesbecause of the presence of only 2 alleles on all the loci.This is not surprising considering the allotetraploid originand amphidiploid nature of C. arabica and its autogamy.Coffea arabica is the only species that exhibited an expectedheterozygosity lower than the observed heterozygosity. Thisresult is consistent with other studies (Lashermes et al.1999; Aggarwal et al. 2007). It could result from the fixedheterozygosity (Lashermes et al. 1999) during the speciationprocess including two different ancestral genomes. Data de-rived from SNP analysis (Pot et al. 2006) confirm this hy-pothesis with the construction of two haplotypes based onsequences. One is close to C. canephora and related species,

Fig. 2. Neighbor-joining tree for 15 individuals (one per species) based on the dissimilarity matrix calculated by simple matching. Bootstrapvalues were calculated with 5000 repetitions.

60 Genome Vol. 51, 2008

# 2007 NRC Canada

while the other exhibits strong relationships with C. euge-nioides. However, heterozygosity within the two ancestralgenomes appears to have been lost, since only one allelefrom each genome remains in C. arabica. This result indi-cates a possible lack of recombination between the ancestralgenomes, while recombination within each genome occursnormally.

We included the two varieties of C. liberica, i.e., C. liber-ica var. liberica and C. liberica var. dewevrei. These twovarieties were genetically well differentiated in previouswork (N’Diaye et al. 2005). In our study, the differentiationbetween these two varieties and their divergence from otherspecies was confirmed.

Coffea congensis, which is considered an ecotype ofC. canephora (Prakash et al. 2005), is differentiated fromC. canephora, but both species are grouped in the samecluster in Fig. 2. Our study also points out the relatednessof C. canephora and C. brevipes. Coffea brevipes originatedfrom Cameroon and Gabon (Chevalier 1947; Anthony 1992;Stoffelen 1998). This species has been described, likeC. congensis (Sybenga 1960; Anthony 1992; Prakash et al.2005), as an ecotype of C. canephora (Chevalier 1947; An-thony 1992; Stoffelen 1998). Our work provides evidence toconfirm the hypothesis that C. brevipes is a dwarf form ofC. canephora, since this species appears to be related to theCentral African genotypes of C. canephora (Fig. 1). Fieldstudies should be performed to validate this point of view.

Coffea canephora is the most diverse species, with 95 pri-vate alleles, i.e., 31.25% of the total number of private al-leles and 14.66% of the total number of alleles. Our results(Fig. 1) confirm the division of this species into at least twogroups, i.e., a Congolese group from Central Africa and aGuinean group from West Africa. In contrast, C. libericaand C. congensis exhibit, respectively, 52 and 27 private al-leles, while C. arabica presents 20 private alleles. Theglobal amount of diversity for C. canephora, C. congensis,and C. liberica is very high compared with that for C. arab-ica, which has the lowest diversity even if wild individualsof this species are more diverse than cultivated ones. Theseresults are in accordance with previous studies (Anthony etal. 2002a; Moncada and McCouch 2004) and corroboratethe very narrow genetic base of C. arabica, suggesting asmall number of founders for this species.

Conclusion and consequences for breedingOur work shows the transferability of SSR markers over

the genus Coffea. We point out the potential usefulness ofrelated wild species in breeding strategies for C. arabicaand C. canephora to provide new variability. These resultsincrease the importance of genus diversity studies. Our re-sults, as well as previous analyses using ITS and RFLPmarkers (Lashermes et al. 1997, 1999), lead us to considerthat a high potentiality for breeding has not yet been ex-ploited using species of these two clades.

We propose working on two axes. First, since C. liberica,C. congensis, and the cultivated species are all grouped inrelated clades, the potentialities of crosses between thesespecies are high and the resulting hybrids would have an im-portant level of fertility (Louarn 1992). Variability observedwithin these species can be used for improvement of bever-age and bean quality, productivity, and resistance to biotic

and abiotic stresses in the cultivated species. Second, breed-ing potentialities with species from other diversity groupsare important to assess, since interesting characters havebeen described. For example, C. racemosa (E clade accord-ing to Cros et al. 1998) has been used for coffee leaf minerresistance (Guerreiro et al. 1999; Mondego et al. 2005) andC. anthonyi (C clade) could be used for self-compatibility.

Breeding C. arabica will have to take into account its al-lopolyploid origin. Considering the low rate of recombina-tion between the two ancestral genomes, the introduction ofrecessive alleles coding for traits of interest will be difficult.

Comparative genetic mapping and association mappingwill be developed for future breeding programs. Relation-ships between C. canephora, C. eugenioides, C. arabica,and related species will be analysed to assess valuable traitsfor both quality and resistance improvement throughout thegenus.

AcknowledgementsTechnical help was provided by the Montpellier Languedoc-

Roussillon Genopole genotyping platform. The authorsthank the NARO-CORI (Uganda), the CNRA (Republiquede Cote d’Ivoire), and the IRD (France) for providing plantmaterial. P. Cubry is supported by a grant of the Frenchministry of research. The authors are grateful to J.L. Noyerfor discussions and advice on an early version of themanuscript. We also thank an anonymous reviewer forcomments and advice on this paper.

ReferencesAggarwal, R.K., Hendre, P.S., Varshney, R.K., Bhat, P.R.,

Krishnakumar, V., and Singh, L. 2007. Identification, character-ization and utilization of EST-derived genic microsatellite mar-kers for genome analyses of coffee and related species. Theor.Appl. Genet. 114: 359–372. PMID:17115127.

Alvarez, A.E., van de Wiel, C.C.M., Smulders, M.J.M., and Vosman,B. 2001. Use of microsatellites to evaluate genetic diversityand species relationships in the genus Lycopersicon. Theor.Appl. Genet. 103: 1283–1292. doi:10.1007/s001220100662.

Anthony, F. 1992. Les ressources genetiques des cafeiers: collecte,gestion d’un conservatoire et evaluation de la diversitegenetique. Collection Travaux et Documents Microfiches n8 81,ORSTOM (now IRD), Paris.

Anthony, F., Combes, C., Astorga, C., Bertrand, B., Graziosi, G.,and Lashermes, P. 2002a. The origin of cultivated Coffea ara-bica L. varieties revealed by AFLP and SSR markers. Theor.Appl. Genet. 104: 894–900. PMID:12582651.

Anthony, F., Quiros, O., Topart, P., Bertrand, B., and Lashermes,P. 2002b. Detection by simple sequence repeat markers of intro-gression from Coffea canephora in Coffea arabica cultivars.Plant Breed. 121: 542–544. doi:10.1046/j.1439-0523.2002.00748.x.

Anthony, F., Noirot, M., Couturon, E., and Stoffelen, P. 2006. Newcoffee (Coffea L.) species from Cameroon bring original charac-ters for breeding [CD-ROM]. In 21st International Conferenceon Coffee Science, Montpellier, 11–15 September 2006. Editedby ASIC. Paris, France.

Berthaud, J. 1986. Les ressources genetiques pour l’ameliorationdes cafeiers africains diploıdes. Doctoral thesis, Universite deParis-Sud, Orsay, France.

Casasoli, M. 2004. Cartographie genetique comparee chez les faga-

Cubry et al. 61

# 2007 NRC Canada

cees. Doctoral thesis, Universite de Bordeaux 1, Bordeaux,France.

Chevalier, A. 1947. Les cafeiers du globe, fascicule III, Systema-tique des cafeiers et faux cafeiers. Paris.

Combes, M.C., Andrzejewski, S., Anthony, F., Bertrand, B., Rovelli,P., Graziosi, G., and Lashermes, P. 2000. Characterization ofmicrosatellite loci in Coffea arabica and related coffee species.Mol. Ecol. 9: 1178–1180. doi:10.1046/j.1365-294x.2000.00954-5.x. PMID:10964241.

Cramer, P.J.S. 1948. Les cafeiers hybrides du groupe Congusta.Bull. Agric. du Congo Belge, 34: 29–48.

Cros, J., Combes, M.C., Trouslot, P., Anthony, F., Hamon, S.,Charrier, A., and Lashermes, P. 1998. Phylogenetic analysis ofchloroplast DNA variation in Coffea L. Mol. Phylogenet. Evol.9: 109–117. doi:10.1006/mpev.1997.0453. PMID:9479700.

Cubry, P., De Bellis, F., Pot, D., Musoli, P., Legnate, H., Leroy, T.,and Dufour, M. 2005. Genetic diversity analyses and linkagedisequilibrium evaluation in some natural and cultivated popula-tions of Coffea canephora. In Proceedings of the 4th PlantGenomics European Meeting, Amsterdam, 20–23 September2005.

Davis, A., and Stoffelen, P. 2006. An annotated taxonomic con-spectus of the genus Coffea (Rubiaceae). Bot. J. Linn. Soc. 152:465–512. doi:10.1111/j.1095-8339.2006.00584.x.

Dufour, M., Hamon, P., Noirot, M., Risterucci, A.M., and Leroy, T.2001. Potential use of SSR markers for Coffea spp. genetic map-ping [CD-ROM]. In 19th International Scientific Colloquium onCoffee, Trieste, 2001. Edited by ASIC. Paris, France.

Dussert, S., Lashermes, P., Anthony, F., Montagnon, C., Trouslot,P., Combes, M.-C., et al. 2003. Coffee (Coffea canephora). InGenetic diversity of cultivated tropical plants. Edited by P. Ha-mon, M. Seguin, X. Perrier, and J.C. Glaszmann. Science Pub-lishers, Inc., Enfield, N.H. pp. 239–258.

Gao, L.Z., Zhang, C.H., and Jia, J.Z. 2005. Cross-species transfer-ability of rice microsatellites in its wild relatives and the poten-tial for conservation genetic studies. Genet. Resour. Crop Evol.52: 931–940. doi:10.1007/s10722-003-6124-3.

Geromel, C., Ferreira, L.P., Cavalari, A.A., Pereira, L.F.P.,Guerreiro, S.M.C., Vieira, L.G.E., et al. 2006. Biochemical andgenomic analysis of sucrose metabolism during coffee (Coffeaarabica) fruit development. J. Exp. Bot. 57: 3243–3258. doi:10.1093/jxb/erl084. PMID:16926239.

Guerreiro, O., Silvarolla, M.B., and Eskes, A.B. 1999. Expressionand mode of inheritance of resistance in coffee to leaf minerPerileucoptera coffeella. Euphytica, 105: 7–15.

The International Plant Names Index. 2007. Available from http://www.ipni.org [accessed 10 December 2007].

Jarne, P., and Lagoda, P.J. 1996. Microsatellites, from molecules topopulations and back. Trends Ecol. Evol. 11: 424–429. doi:10.1016/0169-5347(96)10049-5.

Lashermes, P., Cros, J., Combes, M.C., Trouslot, P., Anthony, F.,Hamon, S., and Charrier, A. 1996. Inheritance and restrictionfragment length polymorphism of chloroplast DNA in the genusCoffea L. Theor. Appl. Genet. 93: 626–632.

Lashermes, P., Combes, M.C., Trouslot, P., and Charrier, A. 1997.Phylogenetic relationships of coffee-tree species (Coffea L.) asinferred from ITS sequences of nuclear ribosomal DNA. Theor.Appl. Genet. 94: 947–953. doi:10.1007/s001220050500.

Lashermes, P., Combes, M.C., Robert, J., Trouslot, P., D’Hont, A.,Anthony, F., and Charrier, A. 1999. Molecular characterizationand origin of the Coffea arabica L. genome. Mol. Gen. Genet.261: 259–266. PMID:10102360.

Leroy, T., Marraccini, P., Dufour, M., Montagnon, C., Lashermes,P., Sabau, X., et al. 2005. Construction and characterization of a

Coffea canephora BAC library to study the organization of su-crose biosynthesis genes. Theor. Appl. Genet. 111: 1032–1041.doi:10.1007/s00122-005-0018-z. PMID:16133319.

Liu, K., and Muse, S.V. 2005. PowerMarker: integrated analysisenvironment for genetic marker data. Bioinformatics, 21: 2128–2129. doi:10.1093/bioinformatics/bti282. PMID:15705655.

Louarn, J. 1992. La fertilite des hybrides interspecifiques et les re-lations genomiques entre cafeiers diploıdes d’origine africaine(genre Coffea L., sous-genre Coffea). Doctoral thesis, Universitede Paris-Sud, Orsay, France.

Moncada, P., and McCouch, S. 2004. Simple sequence repeat di-versity in diploid and tetraploid Coffea species. Genome, 47:501–509. doi:10.1139/g03-129. PMID:15190367.

Mondego, J.M.C., Guerreiro-Filho, O., Bengtson, M.H.,Drummond, R.D., Felix, J.M., Duarte, M.P., et al. 2005. Isolationand characterization of Coffea genes induced during coffeeleaf miner (Leucoptera coffeella) infestation. Plant Sci. 69:351–360.

Montagnon, C. 2000. Optimization des gains genetiques dans leschema de selection recurrente reciproque de Coffea canephoraPierre. Doctoral thesis, Ecole Nationale Superieure Agronomi-que de Montpellier, Montpellier, France.

Musoli, P., Aluka, P., Cubry, P., Dufour, M., De Bellis, F.,Ogwang, J., et al. 2006. Fighting against coffee wilt disease:Uganda wild canephora genetic diversity and usefulness. In 21stInternational Conference on Coffee Science, Montpellier, 11–15 September 2006. Edited by ASIC. Paris, France.

N’Diaye, A., Poncet, V., Louarn, J., Hamon, S., and Noirot, M.2005. Genetic differentiation between Coffea liberica var.liberica and C. liberica var. dewevrei and comparison withC. canephora. Plant Syst. Evol. 253: 95–104. doi:10.1007/s00606-005-0300-1.

Perrier, X., Flori, A., and Bonnot, F. 2003. Data analysis methods.In Genetic diversity of cultivated tropical plants. Science Pub-lishers, Inc., Enfield, N.H. pp. 43–76.

Poncet, V., Hamon, P., Minier, J., Carasco, C., Hamon, S., andNoirot, M. 2004. SSR cross-amplification and variation withincoffee trees (Coffea spp.). Genome, 47: 1071–1081. doi:10.1139/g04-064. PMID:15644965.

Poncet, V., Dufour, M., Hamon, P., Hamon, S., de Kochko, A., andLeroy, T. 2007. Development of genomic microsatellite markersin Coffea canephora and their transferability to other coffee spe-cies. Genome, 50(12):1156–1161. doi:10.1139/G07-073.

Pot, D., Bouchet, S., Cubry, P., Dufour, M., De Bellis, F., Jourdan,I., et al. 2006. Nucleotide diversity of genes involved in sucrosemetabolism. Towards the identification of candidate genes con-troling sucrose variability in Coffea spp. In 21st InternationalConference on Coffee Science, Montpellier, 11–15 September2006. Edited by ASIC. Paris, France.

Prakash, N.S., Combes, M.C., Somanna, N., and Lashermes, P.2002. AFLP analysis of introgression in coffee cultivars (Coffeaarabica L.) derived from a natural interspecific hybrid. Euphy-tica, 124: 265–271. doi:10.1023/A:1015736220358.

Prakash, N.S., Combes, M.C., Dussert, S., Naveen, S., and La-shermes, P. 2005. Analysis of genetic diversity in Indian robustacoffee genepool (Coffea canephora) in comparison with a repre-sentative core collection using SSRs and AFLPs. Genet. Resour.Crop Evol. 52: 333–343. doi:10.1007/s10722-003-2125-5.

Risterucci, A.M., Grivet, L., N’Goran, J.A.K., Pieretti, I., Flament,M.H., and Lanaud, C. 2000. A high-density linkage map ofTheobroma cacao L. Theor. Appl. Genet. 101: 948–955. doi:10.1007/s001220051566.

Rovelli, P., Mettulio, R., Anthony, F., Anzueto, F., and Lashermes,P. 2000. Microsatellites in Coffea arabica L. In Coffee biotech-

62 Genome Vol. 51, 2008

# 2007 NRC Canada

nology and quality. Edited by T. Sera, C.R. Soccol, A. Pandey,and S. Roussos. Kluwer Academic Publishers, the Netherlands.pp. 123–133.

Rozen, S., and Skaletski, H.J. 2000. Primer 3. Version 0.2 [com-puter program]. Available from http://primer3.sourceforge.net/.

Saitou, N., and Nei, M. 1987. The neighbor-joining method: a newmethod for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406–425. PMID:3447015.

Stoffelen, P. 1998. Coffea and Psilanthus (Rubiaceae) in tropicalAfrica: a systematic and palynological study, including a revi-

sion of the West and Central African species. Doctoral thesis,Katholieke Universiteit Leuven, Leuven, Belgium.

Sybenga, J. 1960. Genetics and cytology of coffee. A literature re-view. Bibliographica Genet. 19: 217–316.

Tautz, D., and Renz, M. 1984. Simple sequences are ubiquitous re-petitive components of eukaryotic genomes. Nucleic Acids Res.12: 4127–4138. doi:10.1093/nar/12.10.4127. PMID:6328411.

Yu, K., Park, S.J., and Poysa, V. 1999. Abundance and variation ofmicrosatellite DNA sequences in beans (Phaseolus and Vigna).Genome, 42: 27–34. doi:10.1139/gen-42-1-27.

Cubry et al. 63

# 2007 NRC Canada