Content uploaded by Alexandre De Kochko
Author content
All content in this area was uploaded by Alexandre De Kochko
Content may be subject to copyright.
ORIGINAL PAPER
Improving the quality of African robustas: QTLs
for yield- and quality-related traits in Coffea canephora
Thierry Leroy &Fabien De Bellis &Hyacinthe Legnate &Edmund Kananura &
Gustavo Gonzales &Luiz Felipe Pereira &Alan Carvalho Andrade &
Pierre Charmetant &Christophe Montagnon &Philippe Cubry &Pierre Marraccini &
David Pot &Alexandre de Kochko
Received: 29 October 2010 / Revised: 27 January 2011 / Accepted: 7 February 2011
#Springer-Verlag 2011
Abstract Coffea canephora breeding requires combining
sustainable productivity with improved technological and
cup quality characteristics. Beverage quality is a complex
and subjective trait, and breeding for this trait is time
consuming and depends on knowledge of the genetics of its
components. A highly variable C. canephora progeny
resulting from an intraspecific cross was assessed for 63
traits over 5 years. To identify quantitative trait loci (QTLs)
controlling agronomic, technological, and quality-related
traits, a genetic map comprising 236 molecular markers was
constructed, and composite interval mapping was per-
formed. Beverage quality was evaluated in relation to
biochemical and cup tasting traits. QTLs were identified for
almost half of the traits evaluated, with effects ranging from
6% to 80% of phenotypic variation. Most of them present a
consistent detection over years. The strongest QTLs
explained a high percentage of the variation for yield in
2006 (34% to 57%), bean size (25% to 35%), content of
chlorogenic acids (22% to 35%), sucrose and trigonelline
content (29% to 81%), and acidity and bitterness of coffee
beverages (30% to 55%). Regions of the C. canephora
genome influencing beverage quality were identified. Five
QTL zones were co-localized with candidate genes related
to the biosynthesis of the analyzed traits: two genes coding
for caffeine biosynthesis, one gene implicated in the
biosynthesis of chlorogenic acids, and two genes implicated
in sugar metabolism. This is one of the first studies on the
identification of QTLs combining agronomic and quality
Communicated by D. Grattapaglia
Electronic supplementary material The online version of this article
(doi:10.1007/s11295-011-0374-6) contains supplementary material,
which is available to authorized users.
T. Leroy (*):F. De Bellis :P. Charmetant :P. Cubry :
P. Marraccini :D. Pot
UMR AGAP, CIRAD,
34398 Montpellier, France
e-mail: Thierry.leroy@cirad.fr
C. Montagnon
UMR RPB, CIRAD,
34398 Montpellier, France
H. Legnate
Centre National de Recherche Agronomique (CNRA),
BP 808,
Divo, Ivory Coast
E. Kananura
National Agricultural Research Organisation (NARO),
P.O. Box 185, Kituza, Mukono, Uganda
G. Gonzales
Department of Analytical chemistry, University of Seville,
41012 Seville, Spain
L. F. Pereira
Instituto Agronomico do Paraná (IAPAR), LBI-AMG,
CP 481,
86001–970 Londrina, Paraná, Brazil
A. C. Andrade
Laboratory of Molecular Genetics (LGM-NTBio),
Embrapa Genetic Resources and Biotechnology,
CP 02372,
70770–900 Brazilian Federal District, Brazil
A. de Kochko
UMR DIADE, IRD,
34394 Montpellier, France
Tree Genetics & Genomes
DOI 10.1007/s11295-011-0374-6
traits in coffee. The high variability of quality traits within
C. canephora and the presence of consistent QTLs offer
breeders a promising tool to improve coffee cup quality.
Keywords Coffee .QTL .Beverage quality .Yield .
Cup tasting .Biochemical traits
Abbreviations
QTL Quantitative trait loci
MAS Marker-assisted selection
Y_200XYield for the year 200X
CY_200X/YCumulated yield from year 200Xto 200Y
BS Bean size
PB Rate of pea berries
CA Caffeine content
TR Trigonelline content
SU Sucrose content
3C 3-Caffeoylquinic acid content (3-CQA)
4C 4-Caffeoylquinic acid content (4-CQA)
5C 5-Caffeoylquinic acid content (5-CQA)
5F 5-Feruloylquinic acid content (5-FQA)
34dC 3,4-Dicaffeoylquinic acid (3,4di-CQA)
35dC 3,5-Dicaffeoylquinic acid (3,5di-CQA)
45dC 4,5-Dicaffeoylquinic acid (4,5di-CQA)
FR Fragrance
AR Aroma
BO Body
FL Flavor
AC Acidity
BI Bitterness
AF Aftertaste
GL Global note
LG Linkage group
Introduction
Two coffee species have worldwide economic importance,
namely Coffea arabica L. and Coffea canephora Pierre
known commercially as Arabica and Robusta, respectively
(Wintgens 2004). Large variations within and between
cultivated species are observed (Anthony et al. 1993,2001;
Gomez et al. 2009; Cubry et al. submitted for publication).
Arabica represents approximately 65% of worldwide coffee
production and presents higher quality than Robusta due to its
lower bitterness and caffeine content and its more appreciated
flavor (for a review, see Leroy et al. 2006). To date, quality
has not often been considered before the last steps of the
breeding process for C. canephora (Charrier and Berthaud
1988). Two main genetic and geographic C. canephora
groups were identified: the Guinean group from western
Africa and the Congolese group from central Africa (Berthaud
1986). Further studies also divided the Congolese group into
four subgroups (Montagnon et al. 1992;Dussertetal.1999).
A recurrent selection program has been built on C. canephora
breedinginCôted’Ivoire since 1984. This recurrent selection
breeding is based on the improvement of complementary
Guinean and Congolese populations (Leroy et al. 1993).
The main traits commonly taken into account for
C. canephora coffee quality improvement are as follows:
bean size and extractable soluble solids with respect to
technological qualities; aroma precursors, such as sugars,
caffeine, trigonelline, lipids, and chlorogenic acids, as
biochemical traits; and organoleptic traits assessed by cup
tasting. Ky et al. (2001a) described the diversity observed
in a number of quality traits, including caffeine, trigonel-
line, chlorogenic acids, and sucrose in Robusta and Arabica
coffees. For most compounds, they indicated that the
geographical origin of the plants within the genetic groups
was the main factor contributing to variability. Regarding
cup quality, Moschetto et al. (1996) reported differences
between genetic groups related to preference, aroma,
acidity, body, and bitterness. Additionally, the same study
concluded that Guinean genotypes were on average inferior
to Congolese genotypes for preference and aroma. They
also indicated good linear correlation coefficients between
preference and certain factors, such as acidity and aroma.
Montagnon et al. (1998) observed that the variations in
yield and quality traits were independent within C. canephora,
meaning that quality could be improved without lowering
yield. In the same report, the authors showed that narrow
sense heritability was high for caffeine content (h
2
=0.80) and
bean weight (0.73), intermediate for chlorogenic acids (0.36),
and low for sucrose content (0.11). For cumulative yield over
four harvests, narrow sense heritability values were estimated
as intermediate with values rangingfrom0.3to0.4(Leroyet
al. 1994; Montagnon et al. 2003).
Other results for interspecific hybrids suggested a high
value (0.71) for the heritability of trigonelline content
(Ky et al. 2001b), with a maternal mode of inheritance.
For sucrose content, while Montagnon et al. (1998)
indicated that this trait could be difficult to improve
because h
2
was low, Ky et al. (2000a) found additive
transmission among their interspecific hybrids, presenting
the possibility of choosing parents for its improvement.
Relationships between different chlorogenic acid mono-
mers or dimers were also studied by Ky et al. (1999)using
an interspecific cross between Coffea liberica and Coffea
pseudozanguebariae, indicating a linear relationship be-
tween caffeoylquinic and dicaffeoylquinic contents. In the
same interspecific population, Barre et al. (1998) indicated
that caffeine was under polygenic control, with strong
genetic effects.
Unraveling the genetic basis of complex traits, such as
yield potential and stability and coffee quality, can be
Tree Genetics & Genomes
undertaken through the construction of a genetic linkage
map followed by quantitative trait loci (QTL) identification.
The first C. canephora map was constructed with
doubled haploids using restriction fragment length poly-
morphism (RFLP) markers (Paillard et al. 1996). Another
map has been developed using doubled haploids for the
analysis of segregation distortion (Lashermes et al. 2001).
Recently, a map has been developed (Lefebvre-Pautigny
et al. 2010) from a segregating population of 93
individuals resulting from a cross between heterozygous
genotypes, using mainly RFLP, simple sequence repeat
(SSR), and expressed sequence tag (EST)–SSR markers.
Maps were also developed for interspecific crosses
(Akaffou et al. 2003; Coulibaly et al. 2003;Kyetal.
2000b;N’Diaye et al. 2007) to identify QTLs involved in
contrasting traits between wild species using mainly
AFLP markers.
To date, few QTL studies have been performed on
coffee, and no QTL report is available related to quality
traits for coffee. QTLs have been identified for the
incompatibility S locus (Lashermes et al. 1996), pollen
viability restoration (Coulibaly et al. 2003), fructification
time (Akaffou et al. 2003), morphological traits (N’Diaye et
al. 2007), and for somatic embryogenesis capacity (Priyono
et al. 2010). However, the genetic diversity from the
Guinean pool has not been explored in these QTL studies,
and the main C. canephora breeding program is based on
reciprocal recurrent selection between the Congolese and
Guinean pool (Leroy et al. 1993).
The aims of the present study were: (1) to construct an
intraspecific genetic linkage map for C. canephora using a
pseudo-backcross progeny between Guinean and Congo-
lese genotypes, (2) to identify QTLs for yield and quality-
related traits, and (3) to draw conclusions related to yield
and quality breeding of C. canephora.
Materials and methods
Plant material
Genetic mapping was performed on an intraspecific popula-
tion of 273 pseudo-backcross individuals of C. canephora.
This progeny resulted from a cross between an intergroup
progeny genotype (Guinean 410×Congolese A03)usedasthe
female and a Guinean genotype used as the male (02183). A
trial was set up in 2000 on Divo station at the Centre
National de Recherche Agronomique (CNRA) in Republic
of Côte d’Ivoire (RCI) (5°46′04.07″N, 5°13′22.09″W,
altitude 200 masl). The experimental design was a fully
randomized single tree plot. Trees were planted at a high
density (2.5×1.5 m, approximately 3,000 plants per hectare).
The grandparents 410 and A03 and the parent 02183 were
planted in the same plot as the progeny. Data were collected
between 2002 and 2006.
During the trial (2000–2006), the pattern of rainfall was
quite normal, with two rainy seasons (May to June and
September to November). Eighteen trees from our progeny
died during the first dry season; the trees were not irrigated
and they were fertilized twice a year with 20 g of 23-10-5
(NPK)+4 Mg. Treatment against weeds was performed six
times a year. Coffee leaf rust appeared on the trees, but no
treatment was performed for this because this disease had
no significant effect on yield.
Cherries from the entire trial were sun-dried. Samples for
technological, chemical, and organoleptic analyses were
wet processed. After pulping, beans were fermented and
then dried without direct sun exposure.
Evaluation of phenotypic traits
We analyzed each trait separately within each year to take
into account environmental annual effects. A total of 21
phenotypic traits were recorded for variable number of
years, totalling 57 yearly phenotypic traits. Cumulative
yields were analyzed for the 2002 to 2003, 2002 to 2004,
2002 to 2005, 2002 to 2006, 2003 to 2006, and 2004 to
2006 harvests. A total of 63 traits were thus evaluated for
this trial. These traits can be split into four classes: yield,
technological, biochemical, and organoleptic.
Yield traits
Yield was calculated by summing the weights of fresh
berries (in grams) harvested during the productive period
(from the second fortnight of September until the end of
December). Fresh red fruits were harvested by hand each
month and weighed immediately for each tree. Observa-
tions were recorded from the second until the sixth year
after planting (i.e., 2002 to 2006) for 248 trees. It was
observed that 25 of the 273 trees never produced any fruit,
including the 18 trees that died during the first dry season.
Technological traits
Samples of 250 g with 12% moisture were used to estimate
bean size (BS) as the weight of 100 kernels (in grams) and
the percentage of round beans, or pea berries (PB, in
percentage). Both traits were recorded during 4 years
(2003–2006) in samples of 155, 227, 216, and 191 trees,
respectively, depending on the availability of fruits.
Biochemical traits
Biochemical traits were analyzed for the 2003 (103
samples) and 2005 (204 samples) harvests. Caffeine,
Tree Genetics & Genomes
trigonelline, and chlorogenic acid content were analyzed by
reversed-phase high-performance liquid chromatography
(HPLC) with spectrophotometric detection. A 5-μm particle
size C18 150×4.6 mm column was used. Mobile phase
consisted of a mixture of methanol and an aqueous solution
of 5 mM triethylamine and acetic acid (pH 3.0, 30:70 v/v).
Elution was performed in isocratic mode at 1 ml min
−1
.
Detection was conducted with a diode array detector.
Caffeine and trigonelline were quantified using the peak
area (measuring the absorbance at 273 nm for caffeine and
265 nm for trigonelline) by external calibration using
standards. Prior to HPLC determination, the powdered
assay portion was extracted using an aqueous suspension of
MgO at 105°C for 20 min. Quantification of chlorogenic
acids was achieved by peak area measurement. Seven
chlorogenic acids were measured: three caffeoylquinic
acids, 3-CQA (3C), 4-CQA (4C), and 5-CQA (5C); the
feruloylquinic acid 5-FQA (5F) and three dicaffeoylquinic
acids, 3,4di-CQA (34dC), 3,5di-CQA (35dC), and 4,5di-
CQA (45dC). These are the main chlorogenic acids present
in C. canephora (Ky et al. 2001a).
For chlorogenic acid determinations, the methodology
used was reversed-phase HPLC using mobile phases A
(2 mM H
3
PO
4
in methanol 5% (v/v) pH 2.7) and B (2 mM
H
3
PO
4
in methanol 5% (v/v) pH 3.9). A gradient program
was used for 45 min with different volumes of phases A
and B. Chromatograms were recorded at 325 nm. The flow
rate was 0.8 ml min
−1
. Sucrose (SU) was quantified by
enzymatic spectrophotometric determination on green cof-
fee, as described by Alcazar et al. (2004).
Organoleptic traits
The liquoring method used was adapted from CIRAD
protocol (Ribeyre, personal communication) and performed
at NARO facilities (Uganda). Samples were received from
the experimental field (CNRA, RCI) and stored until
testing. A maximum of four samples per day were roasted
to medium roast using a PROBERT roasting machine. Each
sample was tested in triplicate using 10 g of roasted beans
ground to medium size and 200 ml of controlled nearly
boiled water (95°C, pH ~7.0 and dry residues <100 mg) for
each replicate. After mixing and prior to analysis, time was
allowed for coffee powder to settle and for the temperature
to cool down to 50°C. The same panel of five well-trained
tasters conducted the whole experiment. A sample score
using a six-class scale (from 0 = not detected to 5 =fully
detected) was performed on eight coffee quality descriptors:
fragrance, aroma, body, flavor, acidity, bitterness, and
aftertaste. A global note was attributed by each taster based
on a global preference of the beverage. Organoleptic
analyses were carried out on the three last harvests (from
2004 to 2006), with an unequal number of samples (trees)
analyzed each year: 62 trees in 2004, 112 trees in 2005, and
180 trees in 2006.
Data analysis
The mean, standard deviation and variance were calculated
for all traits. Histograms were built, and normality was
checked using a Shapiro and Wilk test; an appropriate
transformation was performed to reach a normal distribu-
tion whenever needed.
Genotypic analyses and map construction
DNA extraction
Genomic DNA was extracted from ground leaves following
an extraction procedure using MATAB buffer adapted from
Risterucci et al. (2000). Purification of the extracts was then
performed using an anion exchange resin column (Nucleo-
Bond AX 20 from Macherey-Nagel, Düren, Germany).
Molecular markers
A set of 147 genomic microsatellites (SSRs) was mapped.
They were identified in enriched libraries (Poncet et al.
2004,2007; Combes et al. 2000; Baruah et al. 2003;
Moncada and Mac Couch 2004). We also mapped 69 SSRs
derived from EST sequences, genes or BAC end sequences
(Leroy et al. 2005; Lin et al. 2005; Geromel et al. 2006).
Twenty fragments of genes were mapped for sugar, lipid,
caffeine, and chlorogenic acid metabolism (Bouchet et al.
2005; Geromel et al. 2006). Fragments of genes related to
drought tolerance in coffee were also mapped (Freire et al.
2010). The genes mapped are presented in Table 1.
Genotyping
For the genes to be mapped, primers were defined at
different locations within genes in promoters, exons, or 3’
or 5’UTR regions. Then, PCR amplification was per-
formed for parents, indicating whether a size polymorphism
(insertion/deletion) was present. If a size polymorphism
was present, the genes were mapped using the same
genotyping method as was used for SSRs.
For each PCR reaction, 5 ng of DNA in 5 μl of water
was used as a template and mixed in a final reaction volume
of 10 μl with 1× buffer (10 mM Tris–HCl, 50 mM KCl,
2 mM MgCl
2
, 0.001% glycerol), 200 μM dNTPs, 0.10 μM
of reverse primer, 0.08 μM of forward primer tailed with
M13 sequence, 0.10 μM of fluorescently labeled M13
primer, and 0.1 U of Taq polymerase. PCR amplifications
were performed in an Eppendorf Mastercycler ep 384
(Eppendorf, Westbury, NY, USA). The amplification
Tree Genetics & Genomes
Table 1 List of genes mapped
Marker Function GenBank
Coffea
a
BLASTx
b
Species E
value
Primer F Primer R
G_4CL_C2 4-Coumarate CoA ligase –XP_002267459 Vitis vinifera 9.00E-
06
AAATCCAAAGGCGAATTGTG ACAAGGTCGGGCATGATTAC
G_4CL_C5 4-Coumarate CoA ligase –XP_002307770 Populus
trichocarpa
1.00E-
26
CCGTACAAGCTCGCTCTATG AGACACGTGGAGACGGATTC
G_4CL_C6 4-Coumarate CoA ligase –XP_002514904 Ricinus
communis
1.00E-
42
TTGCCCCAGAAATTCTCAAC GGCGGTTTCATGTTCATTCT
G_C18332cenar1 No hit –No hit ––ATGGTGGACATCCTGGTGAG GCCAGCAAGTACATGGAGTG
GC03_CCoAMT_intron1 Caffeoyl-CoA-O-
methyltransferase
ABO77959 GCCATAAAAGCCTTCTGCAA GGCTCTGGCTCTCTTGGATA
G_CFS_1D SAM dependent carboxyl
methyltransferase
BAC43760 CATATGAATGGAGGCGAAGG CAATGTCCCGAACTGTTGAA
G_CWI_SSR01 Cell wall invertase AM231577 CAATACGGCATGCATTTGAC TGTTGAACACGCAATTGACC
G_CWI_SSR05 Cell wall invertase ABI17893 ATGTGGTGCTGATGTGCAGT GTCACGTGGGATGATGAGAA
GC17_CP26_intron1 Chlorophyll a/b binding
protein
–Q9XF89 Arabidopsis
thaliana
6.00E-
23
ACGACGAGCTCGCCAAGT GCAGGAAAATTCTTCTGTCAGG
G_KO_C1_2 Ent-kaurene oxydase FJ409844 GCCTCGACCACATCTTTGTT GGCAGGAGAACAATTCAAGC
G_KO_sing2_1 Ent-kaurene oxydase ACQ99375 CTATGTTGATCGCGTGCATT ATGGAGCTCAAGAAGCTGGA
GC19_MYB61_5′UTR R2R3 Myb TF –P81393 Antirrhinum
majus
1.00E-
74
TCAGCCTGTCCTGCATATTG TGAGCTTTCTCACAGCAAGG
G_NMT_A SAM dependent carboxyl
methyltransferase
BAC43756 ACCGCAAACTCGAGAAAGAA ATCCCCAATTCAATCACCAA
G_promSUSY_SSR08 Sucrose synthase AM231581 CGATTTTACACAAGCGTGACA TCTTTTCTTTTCTTCCGGATTG
G_promSUSY_SSR09 Sucrose synthase AM231580 CAAACAAAACAGTACAATTCAATCC ATCCCTGCGAGACCTGACTA
GC32_RBCS_intron1 Ribulose-l,5-
bisphosphatecarboxylase
CAD11990 TTACTTCCCTTGCCACCAAC AAGTCTCGTTCTTCAACTTTCCA
GC25_SDD_3′UTR_1 Serine-type endopeptidase –O65351 Arabidopsis
thaliana
2.00E-
65
CAGCGATACACGGTCACATT AGAGCCCCGATTGATCTTCT
G_SUSY_SSR12 Sucrose synthase AM231583 CAAACAAACAGTACAATTCAATCC ACCCCTGTTTTGCTGTTCAC
G_SUSY_SSR14 Sucrose synthase CAI56307 GGATCTTATCGCAATGAACCA CCAACAGTGTCCTTGCTGAA
G_SUSY2_c6_d Sucrose synthase CAJ32597 TGGCTGGAGTTTATGGCTTC CCTTATTATTATTGAGGAGCAACC
The name of the markers on the map, their function, the mnemonic number in databases, and the primers used for these markers are indicated
a
Coffea spp. sequences. Nucleotide sequence mnemonics are shown in italics and protein sequence mnemonics in bold
b
Best match when performing a BLASTx search in the NCBI nr database (BLASTx 2.2.24)
Tree Genetics & Genomes
program consisted of an initial denaturation cycle of 5 min
at 94°C, followed by ten cycles of “touch-down”PCR
consisting of 45 s at 94°C, 1 min at 60°C to 55°C,
decreasing by 0.5°C each cycle, and 1 min at 72°C. The
next 20 cycles consisted of 94°C for 45 s, 55°C for 1 min,
and 72°C for 1 min, prior to a final elongation step at 72°C
for 5 min.
Fluorescently labeled PCR products were analyzed by
electrophoresis on a 6.5% polyacrylamide gel using a LI-
COR 4300 automated sequencer (LI-COR Biosciences,
Lincoln, NE, USA). Gel images were retrieved, and gels
were scored manually. Individuals were scored according to
the parental segregation types.
Statistical analysis
Statistical analysis was performed to determine variation in
the data and to calculate correlation coefficients. Means
over years and coefficients of variation were computed for
each trait. These coefficients of variation were calculated
considering data throughout harvest years.
Pearson phenotypic correlation coefficients between
traits were evaluated, and the significance was tested by a
Pvalue test. These correlations were calculated for
individual annual data for all traits, as well as for mean
phenotypic data over years.
Map construction
Due to the high heterozygosity of both parents, a pseudo-
testcross strategy was used (Grattapaglia and Sederoff
1994). After building independent datasets with segregating
markers for each parent, parental maps were constructed
using JoinMap 4 software (Van Ooijen 2006)with
Kosambi’s mapping function and default parameters. A
consensus map was then built for a log-of-odds (LOD)
score higher than 4.4 and visualized using Spidermap
software (Rami, unpublished).
QTL analysis
A MapQTL 5 software (Van Ooijen 2004) was used for the
detection of QTLs on the consensus genetic map for the 63
traits observed, with between 62 and 246 individuals scored
for each trait. The data for all 63 traits were used for QTL
detection.
Interval mapping was used to detect QTLs. An LOD
threshold was experimentally determined for a given trait
using the permutation test of MapQTL 5 with 500 iterations.
Specific thresholds were determined for the 1% confidence
level for each trait in each linkage group (LG). At the genome-
wide level, thresholds were determined for 10%, 5%, and 1%
confidence levels. In a second step, composite interval
mapping referred as MQM in MapQTL (Multiple QTL
models, Jansen and Stam 1994) was used after an automatic
cofactor selection allowing the removal of some loci flanking
the most important QTLs and then localizing QTLs with
smaller effects more precisely. The confidence interval was
determined by the LOD-1 method for each QTL.
Allelic effects of the QTLs were estimated as A
f
=[(μ
ac
+
μ
ad
)−(μ
bc
+μ
bd
)]/4 for female additivity, A
m
=[(μ
ac
+
μ
bc
)−(μ
ad
+μ
bd
)]/4 for male additivity, and D=[(μ
ac
+
μ
bd
)−(μ
ad
+μ
bc
)]/4 for dominance, where μ
ac
,μ
ad
,μ
bc
,and
μ
bd
are the estimated phenotypic means associated with each
of the four possible genotypic classes, ac, bc, ad, and bd,
derived from a <ab × cd> cross (Segura et al. 2009). The
genetic LGs carrying QTLs were presented using MapChart
software (Voorrips 2002).
Results
Quantitative trait analysis
A total of 63 quantitative traits were analyzed separately.
The mean, SD, min, and max values were calculated for the
progeny. Mean values were calculated for three parental
genotypes: 410 and A03 as grandparents, and 02183 as the
male parent of the progeny.
Yield traits
Individual yields for each of the 248 trees of the progeny
were recorded for five consecutive years (2002 to 2006)
along with the calculation of cumulative yields. These five
harvests constitute the first cycle of production for young
trees. After this fifth harvest, trees were cut for a new
production cycle. The 2004 to 2006 harvests were the most
productive. For further analyses, all yield data were log
transformed (log (yield in grams+ 1)) to fit data normality.
Summary data for parents and progeny are presented in
Table S1. The trees were planted in 2000; yields increased
yearly until 2004 and remained stable from 2004 to 2006,
with a slight decrease occurring in 2005 due to alternation
in production. Yield presented a high variability, with a
coefficient of variation reaching 62.2% (Table 2). This high
variability is always observed for yield in coffee, as strong
environmental and ontogenic effects are observed for young
plants.
Correlations were estimated for yearly and cumulative
yields after log transformation (see supplementary Table S2).
All of the yearly and cumulative data were moderately to
highly correlated (0.21 to 0.55), except for yield in 2002 and
2006. The cumulative 5-year yield was highly correlated
with yearly yields from 2004 to 2006, with correlation
coefficients higher than 0.6.
Tree Genetics & Genomes
Technological traits
BS and the rate of PB were measured for harvests from
2003 to 2006. The results for the progeny and for parental
genotypes are presented in Table S1. Bean size remained
low for the progeny and the rate of pea berries was high,
always being higher than 30%. The values observed for the
progeny were within the values observed for the parents
and grandparents (Table S1). Technological traits exhibited
medium coefficients of variation (19% for BS and 34% for
PB, Table 2). The phenotypic correlations were highly
significant for each trait considering successive years, with
values above 0.5 in most cases (Table S2). Significant
positive correlations were also observed between PB and
BS, except in 2005.
Biochemical traits
Biochemical traits were estimated for the 2003 and 2005
harvests. A high variability was observed every year within
the progeny, and values were quite stable from 1 year to
another. For all compounds, the values observed for the
progeny were included between the values measured for the
parents 410 and A03 (Table S1). Some data were modified
to fit normality; for example, data on chlorogenic acids
were transformed using the ArcSIN ((x/100)
0.5
) formula.
The concentration of biochemical compounds in the green
beans presented a coefficient of variation ranging from
11.16% (5-CQA) to 58.19% (3,5di-CQA); see Table 2.
Phenotypic variability was moderate for caffeine, sucrose,
and trigonelline content (19% to 22%). Variability for the
different chlorogenic acids was highly variable, with a
positive correlation with the compound’s mean value. The
lowest variability was observed for 5-CQA (11.16%, for a
mean of 55.06), and the highest was observed for 3,5di-
CQA (58.19%, for a mean of 3.27).
Correlations for biochemical traits are presented in
supplementary Table S2. For sucrose, caffeine and 3-
CQA, 4-CQA and 5-FQA chlorogenic acid content, a high
positive correlation was observed between values in 2003
and 2005. A highly significant positive correlation was
observed between 3-CQA and 4-CQA content (0.99 in
2003 and 0.96 in 2005), whereas highly significant negative
correlations were observed between these two chlorogenic
acids and the other chlorogenic acids, including 5-CQA and
5-FQA.
For the dicaffeoylquinic acids, highly significant positive
correlations were observed between the three compounds
analyzed in both years of the analysis. Correlations between
the contents of all caffeoylquinic acids (except 5-FQA) and
the contents of dicaffeoylquinic acids were negative in
2003 (−0.12 to −0.66) and were significantly negative with
the three di-CQAs in 2005 (−0.34 to −0.43).
Additionally, no significant correlation was observed
between sucrose, caffeine or trigonelline content, and the
content of other compounds, except for caffeine content in
2003 and dicaffeoylquinic acid contents in 2005.
Organoleptic traits
Each coffee sample produced from each plant was
individually tested for its organoleptic characteristics for
three successive years from 2004 to 2006. Small differences
were observed for all traits between genotypes and years.
The Guinean parents of the progeny (410 and 02183)
presented much higher bitterness than the Congolese
grandparent A03, but the differences were quite small for
acidity. Within the progeny, the bitterness was quite high
but decreased over the years. Meanwhile, acidity remained
low for all years, with values also decreasing from year to
year. The global preference notation given by the judges
decreased in 2006, along with decreasing aroma and body
(Table S1). The variability of organoleptic traits (Table 2)
was generally low to medium (13% to 23%), except for
acidity and bitterness, which showed high levels of
variability (99% and 40%, respectively) with very low
mean values. The high level of variability found for acidity
Table 2 Mean and coefficient of variation for all traits, calculated
from data throughout the 5 years of the experiment
Type of trait Trait Mean Coefficient of variation in percent
Yield Y 1.82 62.23
Technological PB 36.28 34.12
BS 9.91 19.33
Biochemical SU 5.27 22.63
CA 2.56 19.56
TR 0.84 22.94
3C 9.62 20.36
5C 55.06 11.16
4C 13.62 17.38
5F 11.48 26.55
34dC 3.26 36.78
35dC 3.27 58.19
45dC 4.39 53.47
Organoleptic FR 2.03 13.29
AR 1.97 14.63
BO 1.98 16.48
FL 1.85 20.16
AC 0.13 99.24
BI 1.33 39.79
AF 1.72 22.09
GL 1.86 23.60
CA caffeine, FR fragrance, AR aroma, BO body, FL flavor, AC acidity,
BI bitterness, AF aftertaste, GL global note
Tree Genetics & Genomes
was due to its very low content for the beans tested, and
thus, this trait was very difficult for the tasters to estimate.
Significant correlations between organoleptic traits were
observed each year, but not between years, indicating the
independence of successive years (Table S2). As expected,
acidity was highly positively correlated with the global
preference note, with values ranging from 0.49 to 0.60, but it
was negatively correlated with bitterness (−0.33 to −0.37).
Phenotypic correlations among all mean trait values over
the years were evaluated (Table 3), including cumulative
yield over five years. Significant but low positive correla-
tions were observed between cumulative yield (CY_2002/
6), bean size (0.21), and sucrose content (0.15). Cumulative
yield was also significantly correlated with favorable
organoleptic traits, such as acidity, flavor, and the global
note (0.19 to 0.29). Technological traits (bean size and rate
of pea berries) presented highly significant correlations
with 3-CQA (0.15 to 0.17) and with 5-FQA (negative
correlations from −0.17 to −0.22). Finally, the correlations
between flavor, acidity, and the global note, with respect to
organoleptic traits, and sucrose and 5-CQA content, among
biochemical traits, were highly and positively significant. A
negative correlation was observed between flavor and
acidity with trigonelline and 5-FQA content. Bitterness
was highly negatively correlated with sucrose (−0.25) and
was highly positively correlated with caffeine content
(0.20).
LG analysis
A total of 248 markers were initially used for genotyping
the population of 248 individuals. Markers or genotypes
with more than 10% of missing data were eliminated.
Finally, 238 markers were mapped using 184 individual
trees. Eleven LGs were constructed, corresponding to the
11 gametic chromosomes of C. canephora. The total length
of our consensus map was 1,290 cM, with an average
distance of 5.5 cM between markers and a maximum
distance of 37 cM between markers. The length of the LGs
was variable, ranging from 57.8 cM for LG I to 243.3 cM
for LG B. Segregation distortion was observed for all
markers in LG I, with χ
2
test values found to be highly
significant (p<0.05 to 0.0001). The consensus map is
presented in Fig. 1. We built both parental maps, but as the
map for the male Guinean parent 02183 was sparse, we
only used the female map to confirm the position of
markers on the consensus map.
Sixteen genes were mapped to nine LGs. These genes
were identified in EST libraries in the framework of our
collaboration with Brazilian teams from IAPAR and
EMBRAPA/CENARGEN (Geromel et al. 2006; Freire et
al. 2010). Two genes involved in caffeine metabolism were
mapped to LGs A and I. One gene (two markers) related to
lipid metabolism has been mapped to LG J. Four genes
involved in the biosynthesis of chlorogenic acids were
mapped to LGs B, F, G, and H. For CGA biosynthesis, we
mapped one CCoAOMT gene that was previously mapped
by Campa et al. (2003) and three 4CL genes. This last
enzyme acts at different levels in the CGA metabolic chain
pathway (Lepelley et al. 2007; Joët et al. 2009). Five
putative candidate genes implicated in drought tolerance
were mapped on to the B, C, F, G, and J LGs. Finally, four
genes for sucrose metabolism were mapped to LGs A, D,
and F (five markers from two sucrose synthase genes, SUS1
and SUS2,andtwoCWI genes encoding cell wall
invertases). With respect to previous work on the charac-
terization of a C. canephora BAC library using sucrose
synthase genes (Leroy et al. 2005), the copy number of
these genes were confirmed, with one copy found for SUS2
and three copies found for SUS1.
QTL detection
QTL detection was performed from a synthetic map with
209 markers. Markers that were very close to each other
were eliminated for QTL detection. Phenotypic data were
available for all of the trees, and these trees were integrated
in the QTL study independently of their elimination in map
construction. The QTL detection was, thus, performed
using all of the 248 trees of the progeny. Figure 2presents
the main QTLs identified per LG for all data and Table 4
presents all of the QTLs and their characteristics.
Seven QTLs were identified for yield (Table 4). One
main QTL zone was located in LG K for the 2006 yield and
the cumulative yield for 3 to 5 years of harvest. Two QTLs
that explained 34% to 42% of the phenotypic variations for
the 2006 yield were identified in LG I, and one QTL was
identified in the D LG for the 2003 yield. Dominance
effects were predominant for cumulative yield.
For the rate of pea berries, six QTLs were identified in
LGs F, J, and K. A female additive effect was found to be
the most important effect for this trait. Eleven QTLs for
bean size were found in LGs A, B, D, F, and J, with little
consistency throughout the years, and quite surprisingly,
dominance effects were the most important for this
heritable trait. One QTL in LG A explained 35% of the
variation in this trait in 2005, while two QTL zones in LG
B explained up to 25% of the trait variation.
Co-localization between agronomic and technological
traits was limited to LG F, where a QTL for bean size and
the rate of pea berries co-localized. With respect to
biochemical traits, 27 QTL zones were implicated in this
study (Table 4). Female additive effects were quite important
for chlorogenic acids and caffeine. For sucrose and trigonel-
line content, male additive and dominant effects were also
found. QTLs were identified for all traits analyzed. For these
Tree Genetics & Genomes
Table 3 Phenotypic correlations among mean values of the traits over the years (Pearson’s correlation test)
BS CY_2002/6 FR AR BO FL AC BI AF GL SU CA TR 3C 5C 4C 5F 34dC 35dC 45dC
PB 0.19** ns 0.22** ns ns ns ns ns ns ns ns ns ns 0.17** ns 0.16* −0.17** ns ns ns
BS 0.21** ns ns 0.14 * ns 0.15* ns ns ns ns 0.16* 0.17* 0.15* ns ns −0.22** ns 0.14* ns
CY_2002/6 ns ns 0.19** 0.19** 0.18** 0.18* 0.29** 0.20** 0.15* ns ns ns ns ns ns ns ns ns
FR 0.29** 0.18** ns ns ns 0.18* ns ns ns ns 0.16* ns ns ns ns ns ns
AR 0.16* 0.19** ns ns 0.15* ns ns ns ns ns ns ns ns ns ns ns
BO 0.42** 0.29** ns 0.45** 0.44** ns ns ns ns ns ns ns ns ns 0.17*
FL 0.59** −0.26** 0.53** 0.74** 0.23** ns −0.18* ns 0.26** ns −0.15* ns ns ns
AC −0.19** 0.50** 0.66** 0.30** ns −0.24** ns 0.18* ns −0.18** ns ns ns
BI ns −0.23** −0.25** 0.20** ns ns ns ns ns ns ns ns
AF 0.59** 0.14* ns ns ns 0.18* ns −0.16* ns ns ns
GL 0.25** ns ns ns 0.29** ns ns ns ns ns
SU ns ns −0.15* 0.25** −0.15* −0.16* ns ns ns
CA 0.21** −0.15* ns −0.13* ns 0.15* 0.22** 0.20**
TR ns ns ns 0.26** ns ns ns
3°C −0.27** 0.96** −0.39** −0.36** −0.43** −0.43**
5°C −0.22** −0.19** −0.44** −0.35** −0.44**
4°C −0.41** −0.41** −0.47** −0.47**
5°F −0.17* ns ns
34dC 0.82** 0.88**
35dC 0.78**
ns not significant, FR fragrance, AR aroma, BO body, FL flavor, AC acidity, BI bitterness, AF aftertaste, GL global note
*P<0.05; **P< 0.01; only significant correlations are shown
Tree Genetics & Genomes
three traits, no consistency was observed throughout the
years, but some QTLs in LGs I and K for trigonelline content
explained a large part of the variation of the trait: 81.2% and
41.8% in 2003 and 2005, respectively.
Eight QTLs were identified for 3-CQA and 4-CQA
contents in LGs A, B, and I, with a good consistency
throughout years. Both QTLs for 4-CQA content in 2003
and 2005 in LG B represented 28.8% and 40.6% of the
variation in this trait, respectively (26.7% and 16% for 3-
CQA). For 5-FQA, six QTLs were identified in LGs A, B,
D, F, and I, with a common QTL found in LG I in 2003 and
2005. The QTLs for this trait in LG F explained up to 35%
of its variation in 2003.
In LG A, a unique QTL zone was related to both 4-CQA
and 5-FQA content, which are negatively correlated in
2005. In LG B, a large QTL zone included QTLs for 3-
CQA, 4-CQA, 5-CQA, and 5-FQA content. In LG D, a
small zone included QTLs for 5-FQA and 3,5di-CQA.
Another zone in the same LG includes QTLs for trigonel-
line, 5-FQA, and 4-CQA. In LG I, the QTL zone included
QTLs for caffeine (2005), CGA monomers and sucrose in
2005. In LG J, a small zone included QTLs for 3,4 and
3,5di-CQA. In LG K, one QTL zone was identified for
caffeine and trigonelline in 2005, and another was found for
5-CQA in 2003.
Six QTLs were identified for organoleptic traits (Table 4).
One QTL was identified for the global note in LG H (not
presented in Fig. 2). For bitterness (2004), one QTL was
identified in LG D. In LG I, we observed a QTL zone for
acidity in 2006 and bitterness in 2005 and 2006. This result
is consistent with the negative phenotypic correlations
observed between these traits. For bitterness, female
additive effects were predominant, while for acidity, only
a male additive effect was identified.
All of the QTLs found for organoleptic properties
explained more than 15% of the trait variability and up to
54.8% for acidity in LG I. The major co-localizations of
QTL for traits from different sources of data were as follows:
–In LG B, bean size in 2004 and 2005 with caffeoyl-
quinic acids in 2005
–In LG I, co-localization between bitterness, acidity (in
2006), sucrose content (2005), content of 3-CQA, 4-
CQA and 5-FQA (2003 and 2005), caffeine content
(2005), and yield (2006) was found.
–In LG J, bean size (2006) with dicaffeoylquinic acids
(2005)
Some QTL zones co-localized with genes implicated in
different metabolic pathways related to coffee quality. In
LG A, a QTL for caffeine content co-localized with a
caffeine synthase gene implicated in the last steps of
caffeine biosynthesis (Misako and Kouichi 2004). In LG
B, a composite QTL zone, including all caffeoylquinic
acids and bean size was localized near a 4CL gene
implicated in the biosynthesis of chlorogenic acids and
near genes encoding two invertase genes (mapped from
BAC ends sequences). In LG D, QTLs for bean size (2006)
were located in the same zone as a gene encoding a cell
wall invertase CWI and a copy from a SUS1 gene mapped
from BAC end sequences (Leroy et al. 2005). In LG F,
QTLs for bean size and the rate of pea berries were located
near a SUS1 sucrose synthase gene. In LG I, a gene implied
in caffeine metabolism, one N-methyltransferase, was
found close to QTLs for caffeine (2005), monomer CGAs
and sucrose (2005), and acidity and bitterness (2006).
Discussion
Our population of C. canephora possesses a large amount
of phenotypic variability for yield and for traits related to
quality. This outcome was expected for yield. For sucrose
and caffeine content, it presents a higher variability than
that observed in C. canephora accessions by Ky et al.
(2001a). As it is known that sucrose content has additive
transmission, selection for this trait will be possible. For
biochemical traits, a good correlation was observed
between the years, which is consistent with the results
obtained by Ky et al. (2000a).
CGA biosynthesis in coffee seeds has been analyzed by
Joët et al. (2010). These authors defined “control boxes”for
CGA biosynthesis and showed relationships between
mono- and di-CQA chlorogenic acids and the influence of
abiotic factors such as low temperatures. Linear relation-
ships between di-CQA have also been established (Bertrand
et al. 2003b). Our results confirm linear relationships
between CQA and di-CQA, all di-CQA being derived from
5-CQA (Ky et al. 1999).
The genetic map
As expected for C. canephora, a genetic map of 11 linkage
groups was produced using 238 markers, with an average
of 5.5 cM between markers. As previously reported for
intra- and interspecific maps in coffee (Paillard et al. 1996;
Lashermes et al. 2001; Ky et al. 2000b), segregation
distortion was observed for LG I. The distorted markers
have not been discarded during map construction as they
have for other species (Venkateswarlu et al. 2006) because
Fig. 1 Genetic linkage map of C. canephora based on SSR markers
and genes, constructed with 238 markers and 184 plants with a
minimum LOD score of 4.4. Distortion of markers in linkage groups
is indicated by stars corresponding to the significance of the χ
2
test
(double asterisk,p=0.05 to septuple asterisk,p=0.0001). Candidate
genes are indicated in bold
Tree Genetics & Genomes
A BCDE F
GH I J K
Tree Genetics & Genomes
it can be considered that these distortions have a biological
basis. As has been noted for other plants (Billote et al.
2005; Lu et al. 2002), the presence of lethal genes or
reproduction regulating genes could explain these distor-
tions (Zamir and Tadmor 1986). In C. canephora, the self-
incompatibility gene S should be located in the LG
designated LG 9 by Lashermes et al. (1996), and this LG
has been then confirmed to be highly distorted (Lashermes
et al. 2001), suggesting a link between segregation
distortion and the S locus. Our LG I corresponds to LG 9
of Lashermes. In our study, if we consider that this simple
gametophytic factor is responsible of the distortion ob-
served, estimates of the recombination frequency should
not be affected (Lu et al. 2002). In the other LGs, only
small clusters of markers presented segregation distortion
(Fig. 1).
QTL identification and gene discovery
We were able to identify QTLs for most traits with large
effects and consistent detection over the years, especially for
yield, technological traits, and chlorogenic acid content, thus
indicating ontogenic stability. These QTLs, also referred to as
stable QTLs, are important because they can be excellent
candidates for MAS work (Kenis et al. 2008). For other traits,
such as organoleptic traits and other biochemical compounds
related to quality, the consistency over the years is less
obvious. LG I is highly distorted, and it could, thus, affect
the quality of QTL identification. Recent reports (Xu 2008;
Zhang et al. 2010) have noted that distortion can decrease the
possibility of QTL detection, but it should not increase the
rate of false positives. Zhang et al. (2010) also noted the
importance of the population size for avoiding modifications
of QTLs due to distortion. Our coffee population of more
than 240 trees can be considered as a large population, thus
limiting the decrease in the power of QTL detection.
However, Xu (2008) suggested that the power of QTL
mapping could be artificially increased when a map is sparse,
as in our study. In conclusion, the QTLs identified in our
distorted LG I should not prevent further investigation on
these regions of the genome.
In this study, several zones of interest were identified. One
zone in LG K (ranging from 45 to 60 cM) is of great interest
for cumulative yield, and the QTL on this LG co-localized
with trigonelline content in the 2003 sample. As previously
reported for the organoleptic quality of fresh tomato fruits
(Causse et al. 2000;Salibaetal.2001), we identified large
clusters of QTLs related to quality in LGs I and B. The main
QTL zone for quality in LG I is associated with correlated
traits like acidity and bitterness, caffeine and chlorogenic
acids A key gene involved in caffeine biosynthesis is co-
localized with these QTLs. It was demonstrated here for the
first time that the genetic variability in chemical compounds,
such caffeine and chlorogenic acids, is related to the genetic
variability of beverage quality (acidity and bitterness).The
establishment of this clear relationship can be considered as
the main result of this work. We also observed co-
localization between bean size and quality traits in LG B.
Finally, we identified a co-localization between a QTL for
bean size, a SUS1 gene and a CWI gene coding for the cell
wall invertase in LG D. This result is of extreme importance,
as several invertase- and sucrose synthase-encoding genes
have been mapped along with a QTL related to fruit size in
tomatoes (Fridman et al. 2000; Causse et al. 2004). These
genes are involved in the control of tomato fruit size (Klann
et al. 1996; D´Aoust et al. 1999). In shaded coffee beans,
Geromel et al. (2008) also observed higher invertase activity
during the latest stages of perisperm development that could
be related to the larger size of beans in comparison to those
grown under full sun conditions.
Uncovering co-localization of QTLs with mapped
candidate genes from various biological pathways and
subsequent use of these genes in association mapping will
present possibilities for determining the causes of variation
of quality in coffee (Henery et al. 2007). We can explain the
co-localization between QTLs for organoleptic traits and
genes involved in caffeine or CGA biosynthesis based on
the fact that both CGA and caffeine are involved in
conferring bitterness to the beverage.
Our study is the first to report the identification of QTLs
involved in quality traits and yield in coffee. We identified
major QTLs for yield and quality-related traits. We confirmed
the relationships between some traits and their involvement in
determining quality. Investigation of chemical and biochem-
ical traits related to quality presents an alternative approach to
improve quality, as few QTLs were discovered for direct
beverage quality traits, except for acidity and bitterness. These
types of indirect traits, such as caffeine, chlorogenic acid, and
sugar content, are important for breeders because they are
relatively easy to quantify and because their variability is high,
and they presented significant correlations with beverage
quality traits. We also mapped genes involved in the
biosynthesis of crucial compounds, including caffeine, sugars
and chlorogenic acids; they co-localize with QTLs. Our
results constitute a first step toward the detection of favorable
alleles related to beverage quality in coffee. Finally, we
identified several regions of interest that will be the target for
future association mapping studies. The regions of the genome
connected to yield determination and to quality are different,
Fig. 2 Main QTL localizations in the linkage groups for the
agronomical, technological, biochemical, and organoleptic traits. For
each trait, the QTLs are represented by boxes for a confidence interval
of LOD
max
±1. QTLs were confident at 10% (normal), 5% (italics), or
1% (bold) at the genome-wide level. The names of the QTLs are
presented in Table 4
Tree Genetics & Genomes
Me_ssr120074
0
G_CFS_1D
14
Mg_M273
46
Mg_M361
67
Mg_M850
74
Me_ssr121086
92
Me_ssr125629
103
Me_BE_SPS_M04
104
Me_ssr120037
107
G_SUSY2_c6_d
116
SU_2003_A
CA_2003_A 55FF__22000055__AA22
4C_2005_A
BBSS__22000055__AA
A
Mg_M518
0
Me_ssr121876
35
G_4CL_C2
45
Me_BE_CWI_DL031
51
Me_BE_INV3_DL030
52
Mg_CMA0044
62
Mg_M468
77
Mg_M774
91
G_MYB61_5'UTR
96
Me_ssr120291
109
Mg_M398
114
Mg_M795
122
Me_ssr131684
130
Me_ssr122056
143
Mg_M509
166
Mg_M783
178
Mg_CMA0031
179
Mg_M784
185
Me_ssr120263
201
Me_ssr124612
208
Mg_M384
235
Mg_M495
243
55FF__22000055__BB
44CC__22000033__BB
33CC__22000033__BB
33CC__22000055__BB
44CC__22000055__BB
55CC__22000055__BB
BS_2005_B
BBSS__22000044__BB11BS_2004_B2
BBSS__22000066__BB
AACC__22000066__BB
B
Mg_M742
0
Me_BE_SUSY_DL025
3
Me_BE_CWI_DL013
18
G_CWI_SSR05
21
Mg_M429
22
Me_BE_SUSY_DL028
31
Mg_M392
45
Mg_M461
59
Me_ssr131583
63
Mg_M790
71
Mg_M355
85
Mg_M471
100
BBII__22000044__DD
BS_2006_D1
BS_2006_D2
BS_2006_D3
35dC_2003_D
5F_2005_D
Y_2003_D
D
Mg_M383
0
Me_BE_gA71_DL003
13
G_CCoAMT_intron1
23
Mg_ssrR127
35
Mg_M416
43
G_SUSY_SSR14
59
G_promSUSY_SSR09
65
G_SDD_3'UTR_1
70
Mg_M331
80
G_CWI_SSR01
93
Mg_M328
114
Mg_M353
125
55FF__22000033__FFPPBB__22000044__FF
BS_2006_F
PPBB__22000066__FFBS_2004_F
F
Mg_M441
0
Mg_M445
4
Me_ssr122793
11
Mg_M300
26
G_NMT_A
36
Mg_M277
46
Me_ssr124358
52
Mg_ssrR313
57
BI_2005_I
BBII__22000066__II
AACC__22000066__II
3C_2003_I
4C_2003_I
33CC__22000055__II
CA_2005_I
55FF__22000055__II
55FF__22000033__II
SU_2005_I
TR_2005_I
YY__22000066__II11
YY__22000066__II22
I
Mg_CMA0041
0
G_KO_sing2_1
16
Mg_M337
25
Mg_ssrR290
41
Mg_M378
49
Me_ssr119463
58
Me_BE_62K8_M05
76
GC17_CP26_intron1
91
Mg_M501
103
35dC_2005_J
34dC_2005_J
PPBB__22000055__JJ11PPBB__22000055__JJ22
BS_2006_J
J
Mg_M773
0
Mg_M395
11
Mg_M313
29
Mg_M448
48
Mg_ssrR111
61
Me_ssr131504
75
Mg_M260
83
5C_2003_K CA_2005_K
CY_2002/6_K
CY_2003/6_K
CY_2004/6_K
YY__22000066__KK
TR_2003_K
PPBB__22000055__KK
PB_2006_K
K
Tree Genetics & Genomes
Table 4 List of QTLs identified by MQM for agronomic, technological, biochemical, and organoleptic traits
Traits LG QTL name Cofactors
a
LOD
max
LOD
max
position
b
R
2c
Af Am D
Cumulative yield,
2002–2006
K CY_2002/6_K –3.96 49.2 17.4 0.06 0.06 −0.09
Cumulative yield,
2003–2006
K CY_2003/6_K –4.07 50.2 16.7 0.06 0.06 −0.09
Cumulative yield,
2004–2006
K CY_2004/6_K –3.98 50.2 16.2 0.07 0.07 −0.09
Yield, 2003 D Y_2003_D –4.07 66.1 8.1 0.26 −0.02 −0.18
Yield, 2006 I Y_2006_I1 G_NMT_A, Me_ssr122793, 6.33 43.8 34.0 −0.34 −0.41 −0.29
Mg_M445, Mg_M313
IY_2006_I2 G_NMT_A, Me_ssr122793, 9.46 57.9 42.4 0.03 0.57 −0.03
Mg_M445, Mg_M313
KY_2006_K G_NMT_A, Me_ssr122793, 9.93 62.4 56.8 0.24 0.25 −0.25
Mg_M445, Mg_M313
Bean size, 2004 B BS_2004_B1 Me_ssr120037, Mg_M328 4.64 71.9 10.5 0.38 −0.30 −0.20
BBS_2004_B2 Me_ssr120037, Mg_M328 4.48 130.8 8.1 0.42 0.07 0.18
FBS_2004_F Me_ssr120037, Mg_M328 4.26 113.2 7.9 0.24 0.42 0.01
Bean size, 2005 A BS_2005_A Me_ssr125629,
Me_BE_SPS_M04,
16.44 103.7 35.3 −0.63 0.00 0.01
Mg_M328, Mg_M521
d
BBS_2005_B Me_ssr125629,
Me_BE_SPS_M04,
4.24 64.3 6.4 −0.05 −0.25 −0.49
Mg_M328, Mg_M521
d
Bean size, 2006 B BS_2006_B Mg_M783, Mg_CMA0031, 10.30 145.8 25.3 0.50 −0.01 −0.15
Mg_BAC19_M40
d
,
Me_ssr119463
DBS_2006_D1 Mg_M783, Mg_CMA0031, 4.27 0 6.1 −0.12 0.00 0.45
Mg_BAC19_M40
d
,
Me_ssr119463
DBS_2006_D2 Mg_M783, Mg_CMA0031, 4.92 15.3 10.9 −0.09 0.00 0.60
Mg_BAC19_M40
d
,
Me_ssr119463
DBS_2006_D3 Mg_M783,Mg_CMA0031, 5.12 33.3 14.1 0.04 0.15 0.66
Mg_BAC19_M40
d
,
Me_ssr119463
FBS_2006_F Mg_M783,Mg_CMA0031, 4.49 52.8 11.0 −0.33 0.36 −0.37
Mg_BAC19_M40
d
,
Me_ssr119463
JBS_2006_J Mg_M783, Mg_CMA0031, 5.00 57 7.2 −0.46 −0.19 −0.01
Mg_BAC19_M40
d
,
Me_ssr119463
Pea berry, 2004 F PB_2004_F –6.46 49.8 19.4 −6.05 0.91 −0.18
Pea berry, 2005 J PB_2005_J1 –3.89 39.6 11.5 3.45 0.12 1.50
JPB_2005_J2 –3.94 92.8 8.1 3.03 1.23 0.59
KPB_2005_K
d
–4.49 83.3 16.6 −3.71 1.48 −3.12
Pea berry, 2006 F PB_2006_F Mg_ssrR127,
G_SSR09_promSUSY
5.21 61.5 10.5 −4.33 0.27 1.15
KPB_2006_K
d
Mg_ssrR127,
G_SSR09_promSUSY
4.76 85.5 13.5 −3.83 2.00 −2.60
34dCQA, 2005 J 34dC_2005_J –4.15 53 13.1 0.02 −0.08 −0.40
35dCQA, 2003 D 35dC_2003_D –2.52 45.2 10.7 -0.70 −0.20 0.52
35dCQA, 2005 J 35dC_2005_J –4.10 52 13.0 0.00 −0.13 −0.43
3CQA, 2003 B 3C_2003_B –6.58 57.3 26.7 1.02 −0.04 0.06
I 3C_2003_I –4.02 15.6 21.1 0.83 −0.46 -0.49
3CQA, 2005 B 3C_2005_B Mg_M774, Mg_M398 9.18 63.3 16.0 0.97 −0.33 -0.06
Tree Genetics & Genomes
indicating that simultaneous selection can be conducted on
both types of traits, as previously suggested by Montagnon et
al. (1998). The ongoing work on coffee genome sequencing
will allow for more accurately defining the location of genes
of interest in relation to the QTLs identified in this study.
Consequences for breeding
The perspectives arising from the results of our study are
important in relation with the improvement of the quality of
C. canephora. This study was based on a “backcross”
Table 4 (continued)
Traits LG QTL name Cofactors
a
LOD
max
LOD
max
position
b
R
2c
Af Am D
I3C_2005_I Mg_M774, Mg_M398 5.31 20.4 6.5 0.47 −0.26 -0.05
4CQA, 2003 B 4C_2003_B –6.10 53.1 28.8 1.18 −0.10 0.32
I 4C_2003_I –4.04 15.6 21.6 1.00 −0.55 −0.50
4CQA, 2005 A 4C_2005_A Mg_M774 3.96 86.5 8.3 −0.64 0.33 −0.17
B4C_2005_B Mg_M774 19.62 64.3 40.6 1.41 −0.25 −0.05
5CQA, 2003 K 5C_2003_K
d
–4.31 0 18.5 1.89 0.90 1.45
5CQA, 2005 B 5C_2005_B –4.26 65.3 10.6 −1.83 0.86 -0.43
5FQA, 2003 F 5F_2003_F –5.21 0 35.3 0.92 0.59 1.36
I5F_2003_I –5.12 27.8 23.4 −1.47 −0.43 0.04
5FQA, 2005 A 5F_2005_A2 –5.47 82.5 13.7 0.88 −0.65 −0.35
B5F_2005_B –5.89 41.1 15.2 −1.10 0.46 −0.11
D5F_2005_D –4.38 48.2 10.8 −0.66 0.53 −0.49
I5F_2005_I –7.08 24.8 16.7 −1.26 0.12 −0.27
Caffeine, 2003 A CA_2003_A Me_BE_gA71_DL003,
Mg_300
3.78 22.9 16.2 −0.18 −0.09 0.11
Caffeine, 2005 C CA_2005_C
d
Me_ssr121336
d
,
G_NMT_A,
5.08 2 13.2 0.15 0.11 0.01
Mg_M448
I CA_2005_I Me_ssr121336
d
,
G_NMT_A,
4.05 22.8 7.4 −0.14 0.07 0.08
Mg_M448
K CA_2005_K Me_ssr121336
d
,
G_NMT_A,
3.99 38.2 7.2 −0.10 0.07 −0.07
Mg_M448
Sucrose, 2003 A SU_2003_A Me_ssr120074 4.42 0 28.8 −0.40 −0.58 0.15
Sucrose, 2005 I SU_2005_I –4.39 40.8 12.9 0.40 0.10 0.02
Trigonelline, 2003 K TR_2003_K Mg_M313 11.73 64.4 81.2 0.16 0.15 0.11
Trigonelline, 2005 I TR_2005_I Mg_M441 4.07 60.9 41.8 −0.04 0.04 −0.08
Acidity, 2006 B AC_2006_B Mg_M300, Me_ssr122793 11.91 173.2 29.8 0.00 −0.05 0.00
IAC_2006_I Mg_M300, Me_ssr122793 8.22 42.8 54.8 0.05 0.04 0.05
Bitterness, 2004 D BI_2004_D Mg_M429 5.07 7.7 31.7 0.36 −0.16 −0.03
Bitterness, 2005 I BI_2005_I –4.37 0 16.8 −0.23 −0.06 0.00
Bitterness, 2006 I BI_2006_I –5.24 28.8 15.3 −0.17 −0.02 −0.02
Global, 2005 H GL_2005_H
d
–4.03 73.3 18.6 −0.03 0.09 −0.14
The trait, the linkage group where the QTL is located, the QTL name, the markers used as eventual cofactors in the MQM analyzed, the maximum
LOD and its position, the percentage of phenotypic variation explained for the trait (R
2
), allelic female and male additivity and dominance are
indicated. All of these QTLs present a confidence under 1% at the LG level and 10% at the genome-wide (GW) level. QTLs that present a confidence under
5% GW at the genome level are indicated in italics, and for less than 1% confidence, they are shown in bold
MQM multiple QTL mapping, TR trigonelline, LG linkage group, Af female additive effects, Am male additive effects, Ddominance effects
a
Cofactors were selected using the “Automatic Cofactor Selection”procedure in MapQTL 4.0 software
b
Position of the LOD maximum expressed in cM on the reference map
c
Percentage of phenotypic variation explained by the QTL
d
Not displayed in Fig. 2
e
As presented in “Materials and methods”
Tree Genetics & Genomes
progeny between the Congolese and the Guinean groups on
which the selection program in Côte d’Ivoire is established.
These QTL studies are specific to the pedigree, site, and
ontogenic stage of the individuals investigated. Thus, we
will have to validate the stability of the QTLs obtained here
in other locations and in different progenies, particularly in
the offspring of crosses generated between the diversity
groups used in the selection scheme. For the offspring of
crosses generated within the diversity groups, we will have
to perform complementary QTL studies. Association
mapping studies within genetic groups or on large coffee
collections will also allow precisely determining the zones
of the genome and the genes implicated in the establish-
ment of quality. The results presented here and future
studies to be implemented in other populations will
contribute developing marker-assisted selection for quality
improvement, considering the favorable alleles of the
markers involved in the QTLs for quality. A marker-
assisted selection could be applied directly to our progeny
for the selection of varieties presenting good yield and
quality.
Finally, our work is of interest for assisting breeders
attempting to control the introgression of resistance genes
from C. canephora to C. arabica without lowering quality
(Bertrand et al. 2003a).
Acknowledgments This work was supported by EU grant ICA4-
CT-2001-10068. The University of Trieste (Italy) kindly provided 16
SSR markers.
References
Akaffou DS, Ky CL, Barre P, Hamon S, Louarn J, Noirot M (2003)
Identification and mapping of a major gene (Ft1) involved in
fructification time in the interspecific cross Coffea pseudozan-
guebariae xC. liberica var. Dewevrei: impact on caffeine content
and seed weight. Theor Appl Genet 106:1486–1490
Alcazar A, Jurad JM, Martin MJ, Pablos F, Gonzales AG (2004)
Enzymatic-spectrophotometric determination of sucrose in coffee
beans. Talanta 67:760–766
Anthony F, Clifford MN, Noirot M (1993) Biochemical diversity in
the genus Coffea L.: chlorogenic acids, caffeine and mozambio-
side contents. Genet Res Crop Evol 40:61–70
Anthony F, Bertrand B, Quiros O, Wilches A, Lashermes P, Berthaud
J, Charrier A (2001) Genetic diversity of wild coffee (Coffea
arabica L.) using molecular markers. Euphytica 118:53–65
Barre P, Akaffou S, Louarn J, Charrier A, Hamon S, Noirot M (1998)
Inheritance of caffeine and heteroside contents in an interspecific
cross between a cultivated coffee species Coffea liberica var
dewevrei and a wild species caffeine-free C. pseudozanguebar-
iae. Theor Appl Genet 96:306–311
Baruah A, Naik V, Hendre PS, Rajkumar R, Rajendrakumar P,
Aggarwal K (2003) Isolation and characterization of nine
microsatellite markers from Coffea arabica L. showing wide
cross-species amplifications. Mol Ecol Notes 3:647–650
Berthaud J (1986) Les ressources génétiques pour l'amélioration des
caféiers africains diploïdes. Evaluation de la richesse génétique
des populations sylvestres et de ses mécanismes organisateurs.
Conséquences pour l'application. ORSTOM, Paris
Bertrand B, Guyot B, Anthony F, Lashermes P (2003a) Impact of the
Coffea canephora gene introgression on beverage quality of C.
Arabica. Theor Appl Genet 107:387–394
Bertrand C, Noirot M, Doulbeau S, de Kochko A, Hamon S, Campa C
(2003b) Chlorogenic acid content swap during fruit maturation in
Coffea pseudozanguebariae. Quantitative comparison with
leaves. Plant Sci 165:1355–1361
Billote N, Marseillac N, Risterucci AM et al (2005) Microsatellite-
based high density linkage map in oil palm (Elaeis guineensis
Jacq.). Theor Appl Genet 110:754–765
Bouchet S, Marraccini P, Jourdan I, Leroy T, Vieira LGE, Ferreira LP,
Musoli P, Pot D (2005) Nucleotide diversity and molecular
evolution of five genes involved in the sucrose biosynthesis pathway
of Coffea canephora. In: Proceedings of the 4th Plant Genomics
European Meeting, Amsterdam, 20–23 Sept 2005
Campa C, Noirot M, Bourgeois M, Pervent M, Ky CL, Chrestin H,
Hamon S, de Kochko A (2003) Genetic mapping of a caffeoyl-
coenzyme A 3-0-methltransferase gene in coffee trees. Impact on
chlorogenic acid content. Theor Appl Genet 107:751–756
Causse M, Saliba V, Lesschaeve I, Buret M (2000) Genetic analysis of
organoleptic quality in fresh-market tomato. 2. Mapping QTLs
for sensory attributes. Theor Appl Genet 102:257–272
Causse M, Duffe P, Gomez MC, Buret M, Damidaux R, Zamir D, Gur
A, Chevalier C, Lemaire-Chamley M, Rothan C (2004) A genetic
map of candidate genes and QTLs involved in tomato fruit size
and composition. J Exp Bot 55:1671–1685
Charrier A, Berthaud J (1988) Principles and methods in coffee plant
breeding Coffea canephora Pierre. In: Clarke RJ, Macrae R (eds)
Coffee, vol 4, Agronomy. Elsevier Applied Science, London, pp
167–197
Combes MC, Andrzejewski S, Anthony F, Bertrand B, Rovelli P,
Grasiozi G, Lashermes P (2000) Characterization of micro-
satellite loci in Coffea arabica and related coffee species. Mol
Ecol Notes 9:1171–1193
Coulibaly I, Revol B, Noirot M, Poncet V, Lorieux M, Carasco-Lacombe
C, Minier J, Hamon P (2003) AFLP and SSR polymorphism in a
Coffea interspecific backcross progeny ((C. heterocalyx xC.
canephora)xC. canephora). Theor Appl Genet 107:1148–1155
D’Aoust MA, Yelle S, Nguyen-Quoc B (1999) Antisense inhibition of
tomato fruit sucrose synthase decreases fruit setting and the sucrose
unloading capacity of young fruit. Plant Cell 11:2407–2418
Dussert D, Lashermes P, Anthony F, Montagnon C, Trouslot P, Combes
MC, Berthaud J, Noirot M, Hamon S (1999) Le caféier, Coffea
canephora. In: Hamon P, Seguin M, Perrier X, Glaszmann JC
(eds) Diversité génétique des plantes tropicales cultivées. CIRAD,
Montpellier, pp 175–194
Freire LP, Vieira NG, Vinecky F, Alvas GSC, Leroy T, Pot D, Elbelt
S, Marques T, Rodrigues GC, Marraccini P, Andrade AC (2010).
Expression analysis and nucleic polymorphism of candidate
genes for drought tolerance in coffee (in press)
Fridman E, Pleban T, Zamir D (2000) A recombination hotspot delimits a
wild-species quantitative trait locus for tomato sugar content to 484 bp
within an invertase gene. Proc Natl Acad Sci USA 97:4718–4723
GeromelC,FerreiraLP,GuerreiroSMC,CavalariAA,PotD,PereiraLFP,
Leroy T, Vieira LGE, Mazzafera P, Marraccini P (2006) Biochemical
and genomic analysis of sucrose metabolism during coffee (Coffea
arabica) fruit development. J Exp Bot 57:3243–3258
Geromel C, Ferreira LP, Davrieux F, Guyot B, Ribeyre F, dos Santos
Scholz MB, Pereira LFP, Vaast P, Pot D, Leroy T, Androcioli
Filho A, Vieira LGE, Mazzafera P, Marraccini P (2008) Effects
of shade on the development and sugar metabolism of coffee
(Coffea arabica L.) fruits. Plant Physiol Biochem 46:569–579
Tree Genetics & Genomes
Gomez C, Dussert S, Hamon P, Hamon S, de Kochko A, Poncet V
(2009) Current genetic differentiation of Coffea canephora Pierre
ex A. Froehn in the Guineo-Congolian African zone: cumulative
impact of ancient climatic changes and recent human activities.
BMC Evol Biol 9:167
Grattapaglia D, Sederoff R (1994) Genetic maps of Eucalyptus
grandis and Eucalyptus urophylla using a pseudo-test cross
mapping strategy and RAPD markers. Genetics 137:1121–1137
Henery ML, Moran GF, Wallis IR, Foley WJ (2007) Identification of
quantitative trait loci influencing foliar concentrations of terpenes
and formylated phloroglucinol compounds in Eucalyptus nitens.
New Phytol 176:82–95
Jansen RC, Stam P (1994) High resolution of quantitative traits into
multiple loci via interval mapping. Genetics 136:1447–1455
Joët T, Laffargue A, Salmona J, Doulbeau S, Descroix F, Bertrand B,
de Kochko A, Dussert S (2009) Metabolic pathways in tropical
dicotyledonous albuminous seeds: Coffea arabica as a case
study. New Phytol 182:146–162
Joët T, Salmona J, Laffargue A, Descroix F, Dussert S (2010) Use of
the growing environment as a source of variation to identify the
quantitative trait transcripts and modules of co-expressed genes
that determine chlorogenic acid accumulation. Plant Cell Environ
33:1220–1233
Kenis K, Keulemans J, Davey MW (2008) Identification and stability
of QTLs for fruit quality traits in apple. Tree Genet Gen 4:647–
661
Klann EM, Hall B, Bennett AB (1996) Antisense acid invertase
(TIV1) gene alters soluble sugar composition and size in
transgenic tomato fruit. Plant Physiol 112:1321–1330
Ky CL, Louarn J, Guyot B, Charrier A, Hamon S, Noirot M
(1999) Relations between and inheritance of chlorogenic acid
contents in an interspecific cross between Coffea pseudozan-
guebariae and Coffea liberica var ‘dewevrei’. Theor Appl
Genet 98:628–637
Ky CL, Doulbeau S, Guyot B, Akaffou S, Charrier A, Hamon S,
Louarn J, Noirot M (2000a) Inheritance of coffee bean sucrose
content in the interspecific cross Coffea pseudozanguebariae x
Coffea liberica ‘dewevrei’. Plant Breed 119:165–168
Ky CL, Barre P, Lorieux M, Trouslot P, Akaffou S, Louarn J, Charrier
A, Hamon S, Noirot M (2000b) Interspecific genetic linkage
map, segregation distortion and genetic conversion in coffee
(Coffea sp.). Theor Appl Genet 101:669–676
Ky CL, Louarn J, Dussert S, Guyot B, Hamon S, Noirot M (2001a)
Caffeine, trigonelline, chlorogenic acids and sucrose diversity in
wild Coffea arabica L. and C. canephora P. accessions. Food
Chem 75:223–230
Ky CL, Guyot B, Louarn J, Hamon S, Noirot M (2001b) Trigonelline
inheritance in the interspecific Coffea pseudozanguebariae xC.
liberica var. dewevrei cross. Theor Appl Genet 102:630–634
Lashermes P, Couturon E, Moreau N, Paillard M, Louarn J (1996)
Inheritance and genetic mapping of self -incompatibility in
Coffea canephora Pierre. Theor Appl Genet 93:458–462
Lashermes P, Combes MC, Prakash NS, Trouslot P, Lorieux M,
Charrier A (2001) Genetic linkage map of Coffea canephora:
effect of segregation distortion and analysis of recombination rate
in male and female meioses. Genome 44:589–596
Lefebvre-Pautigny F, Wu F, Philippot M, Rigoreau M, Priyono ZM,
Frasse P, Bouzayen M, Broun P, Pétiard V, Tanksley SD,
Crouzillat D (2010) High resolution syntheny maps allowing
direct comparisons between the coffee and tomato genome. Tree
Genet Gen 6:565–577
Lepelley M, Cheminade G, Tremillon N, Simkin A, Caillet V,
McCarthy J (2007) Chlorogenic acid syntheses in coffee: An
analysis of CGA content and real-time RT-PCR expression of
HCT, HQT, C3H1, and CCoAOMT1 genes during grain
development in C. canephora. Plant Sci 172:978–996
Leroy T, Montagnon C, Charrier A, Eskes AB (1993) Reciprocal
recurrent selection applied to Coffea canephora Pierre I.
Characterization and evaluation of breeding populations and
value of intergroup hybrids. Euphytica 67:113–125
Leroy T, Montagnon C, Cilas C, Charrier A, Eskes AB (1994)
Reciprocal recurrent selection applied to Coffea canephora Pierre
II. Evaluation of genetic parameters. Euphytica 74:121–128
Leroy T, Marraccini P, Dufour M, Montagnon C, Lashermes P, Sabau
X, Ferreira LP, Jourdan I, Pot D, Andrade AC, Glaszmann JC,
Vieira LGE, Piffanelli P (2005) Construction and characterization
of a Coffea canephora BAC library to study the organization of
sucrose biosynthesis genes. Theor Appl Genet 111:1032–1041
Leroy T, Ribeyre F, Bertrand B, Charmetant P, Dufour M, Montagnon
C, Marraccini P, Pot D (2006) Genetics of coffee quality. Braz J
Plant Physiol 18:229–242
Lin C, Mueller LA, Mc Carthy J, Crouzillat D, Pétiard V, Tanksley S
(2005) Coffee and tomato share common gene repertoires as
revealed by deep sequencing of seed and cherry transcripts.
Theor Appl Genet 112:114–130
Lu H, Romero-Severson J, Bernardo R (2002) Chromosomal regions
associated with segregation distortion in maize. Theor Appl
Genet 105:622–628
Misako K, Kouichi M (2004) Caffeine synthase and related
methyltransferases in plants. Front Biosci 9:1833–1842
Moncada P, McCouch S (2004) Simple sequences repeat diversity in
diploid and tetraploid Coffea species. Genome 47:501–509
Montagnon C, Leroy T, Yapo A (1992) Diversité génotypique et
phénotypique de quelques groupes de caféiers (Coffea canephora
Pierre) en collection. Conséquences sur leur utilisation en
sélection. Café Cacao Thé 36:187–198
Montagnon C, Guyot B, Cilas C, Leroy T (1998) Genetic parameters
of several biochemical compounds from green coffee, Coffea
canephora. Plant Breed 117:576–578
Montagnon C, Leroy T, Cilas C, Charrier A (2003) Heritability of
Coffea canephora yield estimated from several mating designs.
Euphytica 133:209–218
Moschetto D, Montagnon C, Guyot B, Perriot JJ, Leroy T, Eskes AB
(1996) Studies on the effect of genotype on cup quality of Coffea
canephora. Trop Sci 36:18–31
N’Diaye A, Noirot M, Hamon S, Poncet V (2007) Genetic basis of
species differentiation between Coffea liberica Hiern and C.
canephora Pierre: analysis of an interspecific cross. Genet
Resour Crop Evol 54:1011–1021
Paillard M, Lashermes P, Pétiard V (1996) Construction of a
molecular linkage map in coffee. TheorAppl Genet 93:41–47
Poncet V, Hamon P, Minier J, Carasco C, Hamon S, Noirot M (2004)
SSR cross-amplification and variation within coffee trees (Coffea
spp.). Genome 47:1071–1081
Poncet V, Dufour M, Hamon P, Hamon S, de Kochko A, Leroy T
(2007) Development of genomic microsatellite markers in Coffea
canephora and their transferability to other coffee species.
Genome 50:1156–1161
Priyono FB, Rigoreau M, Ducos JP, Sumirat U, Mawardi S, Lambot
C, Broun P, Pétiard V, Wahyudi T, Crouzillat D (2010) Somatic
embryogenesis and vegetative cutting capacity are under distinct
control in Coffea canephora Pierre. Plant Cell Rep 29:343–357
Risterucci AM, Grivet L, N’Goran JAK et al (2000) A high-density
linkage map of Theobroma cacao L. Theor Appl Genet 101:948–
955
Saliba V, Causse M, Langlois D, Philouze J, Buret M (2001) Genetic
analysis of organoleptic quality in fresh-market tomato. 1.
Mapping QTLs for physical and chemical traits. Theor Appl
Genet 102:273–283
Segura V, Durel CE, Costes E (2009) Dissecting apple architecture
into genetic, ontogenetic and environmental effects: QTL
mapping. Tree Genet Gen 5:165–179
Tree Genetics & Genomes
Van Ooijen JW (2004) MapQTL 5, software for the mapping of
quantitative trait loci in experimental populations. Kyazma BV,
Wageningen
Van Ooijen JW (2006) JoinMap 4 software for the calculation of
genetic linkage maps. Kyazma BV, Wageningen
Venkateswarlu M, Urs SR, Nath BS, Shashidar HE, Maheswaran M,
Veeraiah TM, Sabitha MG (2006) A first genetic linkage map of
mulberry (Morus spp.) using RAPD, ISSR and SRR markers and
pseudo testcross mapping strategy. Tree Genet Gen 3:15–24
Voorrips RE (2002) Mapchart: software for the graphical presentation
of linkage maps and QTLs. J Hered 93:77–78
Wintgens JN (2004) Coffee: growing, processing, sustainable production
A guidebook for growers, processors, traders, and researchers.
Wiley-VCG Verlag GmbH & Co., Weinheim
Xu S (2008) QTL mapping can benefit from segregation distortion.
Genetics 180:2201–2208
Zamir D, Tadmor Y (1986) Unequal segregation of nuclear genes in
plants. Bot Gaz 147:355–358
Zhang L, Wang S, Li H, Deng Q, Zheng A, Li S,Li P, Li Z, Wang J (2010)
Effects of missing markers and segregation distortion on QTL
mapping in F2 populations. Theor Appl Genet. doi:10.1007/s00122-
010-1372-z
Tree Genetics & Genomes
