CROP BREEDING, GENETICS & CYTOLOGY
SSR and Pedigree Analyses of Genetic Diversity among CIMMYT Wheat Lines
Targeted to Different Megaenvironments
S. Dreisigacker, P. Zhang, M. L. Warburton, M. Van Ginkel, D. Hoisington, M. Bohn, and A. E. Melchinger*
ABSTRACT
duced the concept of breeding for different MEs. A ME
is defined as a large, not necessarily contiguous area,
Improved bread wheat (Triticum aestivum L.) cultivars for diverse
which usually encompasses more than one country and
agroecological environments are important for success in the effort
to increase food production. In the 1980s, CIMMYT introduced the is frequently transcontinental. It is characterized by sim-
megaenvironment (ME) concept to breed wheats specifically adapted
ilar biotic and abiotic stress conditions, cropping sys-
to different areas. Our objective was to analyze the genetic diversity
tems, and consumer demands (Rajaram et al., 1994).
among 68 advanced CIMMYT wheat lines targeted to different MEs
Twelve MEs have been classified, six of which are fo-
by using 99 simple sequence repeats (SSRs) and the coefficient of
cused on efficient selection of better-adapted spring
parentage (COP). The average number of alleles detected was higher
bread wheat, the dominant type of wheat in developing
for the 47 genomic SSRs (5.4) than for the 52 SSRs derived from
countries. The concept has permitted expanding breed-
expressed sequence tags (EST) (3.3), but gene diversity between MEs
ing efforts relevant within each ME. In breeding for
was similar for both types of markers. No significant differences among
enhanced adaptation, adequate genetic diversity is a
the five MEs were observed for the means of SSR-based genetic
prerequisite for any crop improvement program. The
similarities (GS), calculated as 1 ⫺ Rogers’ distance, and COP values.
Both measures showed a low correlation (r ⫽ 0.43). High levels of genetic progress through selection is directly related to
genetic diversity were found within the germplasm targeted to each
the variability present in the gene pool, and the quality
ME. However, principle coordinate analysis based on modified Rog-
of the genes contributed by the parents.
ers’ distances did not separate the genotypes according to their tar-
The COP is an indirect measure of genetic diversity
geted MEs. We conclude that presence of a single core germplasm
among genotypes based on the probability that alleles
can reflect large phenotypic differences. A sufficient number of diverse
at a certain locus are identical by descent. Calculation
breeding lines for each ME is required because MEs generally com-
of COP values rests on simplifying assumptions regard-
bine various production areas. SSRs represent a powerful tool to
ing the relatedness of ancestors, parental contribution
quantify genetic diversity in wheat, but genotypic differentiation for
to the offspring, and absence of selection and genetic
adaptation to specific MEs in the CIMMYT program could not be
drift, which are not met under breeding conditions (Cox
proven.
et al., 1985; Cowen and Frey, 1987). In contrast, molecu-
lar markers measure diversity directly at the DNA level.
In studies of autogamous crops with low levels of appar-
W
heat, together with maize (Zea mays L.) and rice
ent genetic variability such as wheat, soybean [Glycine
(Oryza sativa L.), is one of the three major food
max (L.) Merr.], and rice, SSRs proved to be a suitable
crops in the world. It is grown in a variety of environ-
marker system. They are generally genome specific,
ments, ranging from fully irrigated (e.g., northern India,
abundant, codominant in nature, and show a fairly uni-
Egypt), to high rainfall (e.g., northwestern Europe, east-
form distribution over the genome. SSRs have been
ern Africa, southern zone of Latin America), and
applied in many aspects of genetic diversity analyses
drought-prone regions (e.g., U.S. Great Plains, most of
such as genetic differentiation caused by selection (Sta-
Australia, parts of Argentina). In these areas wheat
chel et al., 2000), fingerprinting of genotypes to analyze
production experiences a range of biotic and abiotic
the structure of germplasm collections (Parker et al.,
stresses and crop improvement requires precise focusing
2002; Huang et al., 2002), and the analysis of temporal
on the needs of the crop in each area, the producers,
changes in diversity (Donini et al., 2000; Christiansen
the processing industry, and the consumers (Lantican
et al., 2002).
et al., 2002).
Traditional methods to develop SSRs are based on
More than one half of the wheat production environ-
isolating and sequencing genomic libraries, which con-
ments are located in developing countries, which fall within
tain putative SSR tracts (Adams et al., 1992). A novel
the mandate of CIMMYT. In the 1980s, CIMMYT intro-
source for generating SSRs is provided by screening
EST databases available online (Kota et al., 2001). This
S. Dreisigacker and A.E. Melchinger, Inst. of Plant Breeding, Seed
recent approach allows researchers to shift from the
Science, and Population Genetics, Univ. of Hohenheim, 70593 Stutt-
use of anonymous markers with unknown effect on the
gart, Germany; M. Bohn, Dep. of Crop Science, Univ. of Illinois,
phenotype to markers physically associated with coding
Urbana, IL 61801; P. Zhang, M. van Ginkel, M.L. Warburton, and
D. Hoisington, CIMMYT, Mexico D.F., Mexico. Received 17 March
2003. *Corresponding author (melchinger@pz.uni-hohenheim.de).
Abbreviations: AMOVA, analysis of molecular variance; CIMMYT,
International Maize and Wheat Improvement Center; COP, coeffi-Published in Crop Sci. 44:381–388 (2004).
Crop Science Society of America cient of parentage; EST, expressed sequence tag; GS, genetic similar-
ity; ME, megaenvironment; SSR, simple sequence repeat.677 S. Segoe Rd., Madison, WI 53711 USA
381
382 CROP SCIENCE, VOL. 44, MARCH–APRIL 2004
cent dyes 6-carboxyfluoresein, tetrachloro-6-carboxyfluores-
regions, which may more accurately reflect the effects
ein, or hexachloro-6- carboxyfluoresein. PCR was performed
of selection, both natural and artificial.
with the following standard temperature profile: 29 cycles with
The objectives of the present study were to (i) evalu-
a 1 min denaturing step at 94⬚C, 2 min annealing temperatures
ate the use of genomic and EST-derived SSRs for de-
between 50 and 64⬚C depending on the different primer combi-
termining the genetic diversity among advanced spring
nations, and 2 min extension at 72⬚C. The 1-min time spread
bread wheat lines from the CIMMYT breeding pro-
of the standard profile cycle was modified in some cases to
gram, (ii) compare genetic distances based on SSRs
fully optimize amplification conditions.
with the COP estimates of these wheat lines, and (iii)
Amplification products were separated on an ABI
TM
377
determine the diversity for SSRs within and among sets
Sequencer (Perkin Elmer/Applied Biosystems, Foster City,
CA) using 4.5% (w/v) polyacrylamide denaturing gels (acryl-
of lines targeted to different MEs.
amide:bisacrylamide 29:1). Running conditions were 2400 V,
40 mA, 120 W electrophoresis power and 40 mW laser power.
MATERIALS AND METHODS
Products from up to five SSRs could be distinguished simulta-
neously because of the three different fluorescent dyes and
Plant Materials
migration distance differences. Fragment sizes were calculated
A total of 68 CIMMYT advanced spring bread wheat lines
semiautomatically by the computer software GeneScan 3.1
from crosses made during 1989 to 1996 were chosen for this
(Perkin Elmer/Applied Biosystems) by comparing fragments
study (Table 1). Most of the lines were bred by a “modified
with an internal size standard (GeneScan 350 or 500) labeled
bulk” procedure described by Van Ginkel et al. (2002). Seed
with N,N,N,N,-tetramethyl-6-carboxyrhodamine. GeneScan
of outstanding F
7
lines was harvested in bulks for subsequent
fragments were assigned to alleles by the category function
yield trials. These yield trials were grown in replicated and
of the software Genotyper 2.1 (Perkin Elmer/Applied Biosys-
latinized ␣-lattice designs at Cd. Obregon (Sonora, Mexico)
tems). Sixty-four genotypes were run on each gel plus two
or Toluca (Mexico State, Mexico) in 2000 and 2001 under
wheat lines, Opata and Synthetic, as controls.
conditions simulating the different MEs for which they are
We could not optimize the amplification profile of nine
being bred (e.g., full irrigation, reduced irrigation, drought,
SSRs for the scoring with Genotyper. These markers were
and heat stress, etc.). On the basis of their performance, 8 to
optimized to run on small (16 by 20 cm) 6% (w/v), 19:1 acryl-
15 advanced lines were selected from yield trials representative
amide:bis-acrylamide denaturing gels (ATTO
8
AE-6220). The
of the first five spring bread wheat MEs (Table 2). The lines
gels were run for 2 h at about 350 V, with a 100-bp ladder
were chosen as candidates for further evaluation at interna-
as a standard. For fragment visualization, silver staining was
tional testing sites (Van Ginkel et al., 2002). Progenies from
applied according to Applied Biotechnology Center’s Manual
three crosses (Alucan/Duluca, PF869107/CEP8825//Milan and
of Laboratory Protocols. The fragment length of each SSR
Babax/Amadina//Babax) were identified and selected for
was determined with the scientific image system Kodak ID
more than one ME.
2.02 (Kodak, New Haven, CT).
SSR Analyses
Statistical Analyses
DNA extraction was performed with the CTAB method of
Reproducibility of SSR amplification and scoring was deter-
Saghai-Maroof et al. (1984) modified according to CIMMYT
mined on the basis of the percentage of disagreements in the
Applied Biotechnology Center’s Manual of Laboratory Proto-
fragment size of the two standard lines, Opata and Synthetic,
cols (Hoisington et al., 1994). Twenty seeds per advanced line
for gels loaded with the same markers. Allele frequencies at
were grown in the greenhouse and after 2 wk young leaves
the 99 loci, total gene diversity (H
T
), gene diversity within
were harvested from 5 to 10 plants per line. Leaves were
MEs (H
S
), and the proportion of diversity resulting from gene
bulked for DNA extraction to assess the genetic variability
differentiation between MEs (G
ST
) were calculated according
within each line as described by Gilbert et al. (1999). Quality
to Nei (1987). The measures were considered separately for
and quantity of the isolated DNA was determined on 1% (w/v)
the two SSR sources to examine the influence of the genome
agarose gels by comparing bands to known concentrations of
location of the markers (genomic and EST-SSRs).
DNA.
The COP for all pairwise combinations of wheat lines was
SSR information was obtained from two different sources:
calculated on the basis of fully expanded genealogical informa-
46 SSRs were collected from a conventional genomic library
tion extracted from CIMMYT’s International Wheat Informa-
(genomic SSRs) developed at IPK Gatersleben by Ro
¨
der et al.
tion System (Payne et al., 2002). Calculations of COP were
(1998 and unpublished data) and 51 SSRs derived from ESTs
based on the assumptions described by St. Martin (1982),
(EST-SSR) with the prefix “DuPw” were kindly provided by
except for sister lines, which were assigned a COP of 0.56
DuPont, Wilmington, DE (Dupont, unpublished data; Eujayl
instead of 1.0 following Cox et al. (1985). Genetic similarity
et al., 2002). The SSRs from both sources were distributed
(GS) assigned as 1 ⫺ Rogers’ distance (Rogers, 1972) was
equally over the genome. In addition, the 1BS/1R translocation
estimated to compare SSR-based and COP estimates. Pear-
EST-SSR marker Taglgap (Devos et al., 1995) and the SSR
son’s correlation coefficient (r ) between GS and COP values
marker WMC56 developed by the Wheat Microsatellite Consor-
was calculated for related pairs of lines (COP ⱖ 0.05). Stan-
tium (Agrogene, France) were used. Details for each of the 99
dard errors of genetic similarity estimates were obtained by
SSRs can be found online (http://www.cimmyt.org/english/webp/
a bootstrap procedure with resampling over markers (Weir,
support/publications/support_materials/ssr_mw1.htm; verified
1996). Furthermore, modified Rogers’ distance (Wright, 1978)
23 September 2003).
was calculated among all possible pairs of lines as a basis for
PCR reactions were performed in a model PTC225 ther-
the application of multivariate methods, because it represents
mocycler (MJ Research, Inc., Waltham, MA). Each 20-L
a Euclidean distance.
reaction mixture contained 25 ng template DNA, 150 nM of
An analysis of molecular variance (AMOVA) on the basis
each primer, 250 M dNTPs, 200 M MgCl
2
,1⫻ PCR buffer
of SSR data was computed to test the differentiation of the 68
and 2.5 U of Taq-polymerase. Forward primers were labeled
at the 5⬘ end with either one of three phosphoramidite fluores- genotypes according to the five MEs. The hierarchical analysis
DREISIGACKER ET AL.: GENETIC DIVERSITY AMONG CIMMYT WHEATS 383
Table 1. Pedigrees of the 68 CIMMYT spring bread wheat lines classified by megaenvironments (ME).
No. Pedigree†
ME1IR (irrigated zones)‡
1 KAUZ//ALTAR 84/AOS/3/KAUZ
2 ATTILA/3/HUI/CARC//CHEN/CHTO/4/ATTILA
3 PRINIA/WEAVER//STAR/3/WEAVER
4 OASIS/4*BORL95
5 WEAVER/WL3926//SW89.3064
6 RABE/2*MO88
7 CNDO/R143//ENTE/MEXI_2/3/AEGILOPS SQUARROSA (TAUS)/4/WEAVER/5/2*KAUZ
8 CHEN/AEGILOPS SQUARROSA (TAUS)//FCT/3/2*WEAVER
9 CHEN/AEGILOPS SQUARROSA (TAUS)//FCT/3/STAR
10 CHUM18/5*BCN
11 P1.861/RDWG
12 CMH80A.542/CNO79
13 SUPER SERI #2
14 PVN//CAR422/ANA/5/BOW/CROW//BUC/PVN/3/YR/4/TRAP#1
15 BABAX/AMADINA//BABAX (WEEBILL1)
ME1HT (irrigated hot zones)
16 VEE/PJN//2*TUI
17 PFAU/WEAVER
18 CAZO/KAUZ//KAUZ
19 MNCH/3*BCN
20 W462//VEE/KOEL/3/PEG//MRL/BUC
21 XIANG82.2661/2*KAUZ
22 SW89-5124*2/FASAN
23 KEA/TAN/4/TSH/3/KAL/BB//TQFN/5/PAVON/6/SW89.3064
24 LAJ3302/2*MO88
25 CROC_1/AE.SQUARROSA (205)//2*BCN
26 PICUS/4/CS5A/5RL-1//BUC/BJY/3/ALD/PVN/5/LAJ3302
27 PFAU/MILAN
28 TAM200/TUI
29 SABUF/7/ALTAR 84/AE.SQUARROSA (224)//YACO/6/CROC_1/AE.SQUARROSA
(205)/5/BR12*3/4/IAS55*4/CI14123/3/IAS55*4/EG,AUS//IAS55*4/ALD
30 [KASYON]//PVN/SPRW
ME2 (high rainfall zones)
31 ALUCAN/DUCULA
32 IAS58/4/KAL/BB//CJ71/3/ALD/5/CNR/6/THB/CEP7780
33 R37/GHL121//KAL/BB/3/JUP/MUS/4/2*YMI #6/5/CBRD
34 TNMU/6/CEP80111/CEP81165/5/MRNG/4/YKT406/3/AG/ASN//ATR
35 TNMU/MILAN
36 TNMU/6/PEL74144/4/KVZ//ANE/MY64/3/PF70354/5/BR14/7/BR35
37 DUCULA/TNMU
38 TRAP#1/BOW//VEE#5/SARA/3/ZHE JIANG 4/4/DUCULA
39 PASTOR//MUNIA/ALTAR 84
40 OR791432/VEE#3.2//MILAN
41 MUNIA/ALTAR 84//AMSEL
42 TNMU/MUNIA
43 TNMU/ATTILA
44 HXL8088/DUCULA
45 PF869107/CEP8825//MILAN
ME3 (high rainfall, acid soil zones)
46 ALUCAN/DUCULA
47 TNMU/TUI
48 TNMU/OCEP17
49 OCEP15/KAUZ//TNMU
50 BR14*2/SUM3//TNMU
51 TNMU/BR35//THB/CEP7780
52 KVZ/3/TOB/CTFN//BB/4/BLO/5/TAN/6/PRL/7/MILAN
53 PF869107/CEP8825//MILAN
ME4 (semiarid zones)
54 DUCULA//VEE/MYNA
55 SRMA/TUI
56 CROC_1/AE.SQUARROSA (224)//OPATA
57 PIOS/DUCULA
58 LAJ3302/3/GZ156/NAC//PSN/URES/4/WEAVER
59 3VASKAR/G303.1M.1.3.2.2.2//KAUZ/3/SKAUZ/4/KAUZ
60 PASA/SAET
61 NL456/VEE#5//CHIL/3/MUNIA
62 TZPP*2/ANE//INIA/3/CNO67/JAR//KVZ/4/MN72252/5/SHI#4414/CROW
63 TSI/VEE#5//KAUZ
64 KAUZ/5/PAT10/ALD//PAT72300/3/PVN/4/BOW
65 KA/NAC
66 ALTAR 84/AE.SQ//2*OPATA
67 FRET2
68 BABAX/AMADINA//BABAX (WEEBILL1)
† Nomenclature according to Purdy et al. (1968): The initial cross is indicated by a single slash (e.g. A/B), the second cross by a double slash (e.g. A/B//
C), and subsequent crosses in numerical order by flanked single slashes (e.g. A/B//C/3/D). Backcrosses are designated with an asterisk (*) and a number
indicating the dosage of the recurrent parent.
‡ Refers to the five MEs described in Table 2.
384 CROP SCIENCE, VOL. 44, MARCH–APRIL 2004
Table 2. Characterization of important spring bread wheat megaenvironments (ME) defined by CIMMYT (Rajaram et al., 1994).
Moisture Temperature Breeding objectives in Contribution of
ME regime regime addition to yield† Year‡ Yield trials wide crosses§
ME1IR Low rainfall, Temperate Resistance to lodging, SR, YR and 1945 Obregon: 700 mm by irrigations China, India, Synthetic,
irrigated LR, end-use quality Durum wheats
ME1HT Low rainfall, Hot As for ME1IR plus tolerance to heat 1945 Obregon: 700 mm by irrigations, China, Argentina, India,
irrigated late planting Synthetic, Durum wheats
ME2 High rainfall Temperate Resistance to SR, YR, LR, Septoria 1972 Toluca: high rainfall (800 mm) China, Brazil,
spp., FHB, BYDV, waterlogging, Durum wheats
pre-harvest sprouting. end-use
quality
ME3 High rainfall Temperate As for ME2 plus Al and Mn tolerance, 1974 Toluca: high rainfall (800 mm), China, Brazil
P-use efficiency seedling test for low pH and
Al toxicity
ME4 Low rainfall Temperate Resistance to SR, YR, LR plus 1970 Obregon: one pre-seeding Argentina, Nepal,
or hot tolerance to drought, end-use irrigation, 300 mm available Synthetic wheats
quality
†SY⫽ Stem rust, YR ⫽ Yellow rust, LR ⫽ Leaf rust, FHB ⫽ Fusarium head blight, BYDV ⫽ Barley Yellow Dwarf Virus.
‡ Refers to the year in which breeding for the respective ME began at CIMMYT.
§ Refers to the 68 lines included in this study.
divides the total variance into variance components due to
ing G
ST
value for all loci was 0.09 and for EST-SSRs
intra- or inter-ME differences and tests their significance. Prin-
(0.10) just slightly higher than for genomic SSRs (0.09).
cipal coordinate analysis was performed on the basis of the
The mean COP value over the 68 wheat genotypes
modified Rogers’ distances to visualize the dispersion of the
was 0.14 and ranged from 0.01 to 0.87 for closely related
genotypes (Gower, 1966). The K-means clustering algorithm
pairs (Table 4). The mean COP values within MEs did
was used to identify groups of similar lines, on the basis of a
not substantially differ between the five MEs. Thirty-
least-squares partitioning method, which divides a collection
nine percent of the COP values were smaller than 0.10,
of objects into k clusters depending on minimum distances to
indicating that theoretically less than 10% of the genetic
the centers of the clusters (MacQueen, 1967). COP values
material segregating in ancestral populations was identi-
were calculated to the six progenitors most frequently used in
cal by descent in any two cultivars.
the crosses and averaged within each ME and K-means cluster.
GS for all pairs of lines ranged from 0.39 to 0.91 with
All analyses were performed with the Plabsim software
(Frisch et al., 2000), which is implemented as an extension of
an average of 0.59 for all genotypes (Table 4). Similar
the statistical software R (Ihaka and Gentleman, 1996). The
to the COP values, MEs were not significantly different
AMOVA was performed by the software package Arlequin
in their mean GS, but with an equal range of values. In
2.0 (Schneider et al., 2000).
the specific cases in which two progenies of the crosses
Alucan/Duluca, PF869107/CEP8825//Milan and Babax/
Amadina//Babax were selected for different MEs, GS
RESULTS
values were high (0.88, 0.86, and 0.72, respectively), as
For the 68 CIMMYT advanced lines analyzed with
expected.
99 SSRs, a total of 425 alleles was detected with an
The mean COP and GS values between MEs were
average of 4.3 alleles per locus. The average number of
of similar size as the means within MEs (Table 5).
alleles was considerably lower in EST-SSRs (3.3) than
ME1HT was most distant to ME2 and ME3 on the basis
in genomic SSRs (5.4), with seven out of the 52 EST-
of COP, and ME1IR most distant to ME3 on the basis
SSRs being monomorphic (Table 3). However, an im-
of GS values. The AMOVA confirmed these results in
portant feature of EST-SSRs was the high-quality frag-
that 92% of the total variation was found within MEs
ment patterns obtained, which were devoid of stutter
and just 8% between MEs (data not shown). The corre-
bands, resulting in a higher reproducibility (98.8%) and
lation between GS and COP values was r ⫽ 0.43.
lower residual heterozygosity (1.4%) than with genomic
The principal coordinate analysis based on modified
SSRs (89.5 and 3.7%, respectively).
Rogers’ distances did not separate the genotypes ac-
The total gene diversity (H
T
) varied widely among
cording to their targeted MEs (Fig. 1). Fourteen of the
loci from 0.01 at DuPw138 to 0.83 at Xgwm437, with an
chosen genotypes cluster somewhat together because
average of 0.47 (Table 3). Considering the two different
of their resistance to acid soil. The K-means cluster
SSR sources, the average H
T
and H
S
values were lower
algorithm identified more than one solution, the most
frequent (90%, 1000 repetitions) comprising three defi-for EST-SSRs than for genomic SSRs. The correspond-
Table 3. Average number of alleles per locus, residual heterozygosity, reproducibility and gene diversity estimated over the two different
sources of markers used in this study.
Gene diversity†
No. of Avg. no. of
SSR source SSR alleles/locus Heterozygosity Reproducibility H
T
H
S
G
ST
Genomic SSRs 47 5.4 3.7 89.5 0.57 0.52 0.09
EST derived SSRs 52 3.3 1.4 98.8 0.37 0.33 0.10
Total 99 4.3 2.5 95.7 0.47 0.43 0.09
†H
T
⫽ total gene diversity, H
S
⫽ diversity within megaenvironments (MEs), G
ST
⫽ diversity between MEs.
DREISIGACKER ET AL.: GENETIC DIVERSITY AMONG CIMMYT WHEATS 385
Table 4. Total and unique number of alleles, number of monomorphic loci, mean genetic similarities (GS), and mean coefficient of
parentage (COP) within each megaenvironment (ME).
GS COP
No. of No. of No. of No. of mono-
ME lines alleles unique alleles† morphic loci Mean Min. Max. Mean Min. Max.
ME1IR 15 267 20 21 0.59 ⫾ 0.06 0.44 0.85 0.18 ⫾ 0.11‡ 0.02 0.59
ME1HT 15 289 33 17 0.59 ⫾ 0.07 0.45 0.91 0.14 ⫾ 0.11 0.01 0.48
ME2 15 253 12 22 0.64 ⫾ 0.06 0.52 0.80 0.11 ⫾ 0.08 0.02 0.37
ME3 8 232 18 24 0.60 ⫾ 0.06 0.47 0.73 0.13 ⫾ 0.06 0.04 0.24
ME4 15 272 23 17 0.63 ⫾ 0.06 0.48 0.85 0.18 ⫾ 0.08 0.04 0.50
Total 68 425 – 7 0.59 ⫾ 0.06 0.39 0.91 0.14 ⫾ 0.09 0.01 0.87
† Alleles occurring only in one ME.
‡ Standard deviation.
nite centers. K-means tended to form the clusters on wheat lines, where on average 5.5 alleles per locus for
the basis of common progenitors used in the crosses
genomic SSRs and 4.1 for EST-SSRs were found, we
made during 1989 and 1996. Seven of the 11 lines with
detected slightly lower average numbers of alleles for
Kauz in the pedigree were included in cluster K1 (Table
both genomic (5.2) and EST-SSRs (3.2). The somewhat
6). Four lines containing Kauz did not group into this
higher level of diversity reported by Eujayl et al. (2002)
cluster. These lines had Chinese wheats in their pedigree
may be attributable to the more variable material in
or were selected in later segregating generations under
their study, which comprised a sample of various lines
different environmental conditions by the International
and landraces with different genetic backgrounds. Here,
Center of Agricultural Research in Dry Areas in Syria.
we studied advanced breeding lines ready for interna-
On the basis of COP, Weaver holds the highest parental
tional dissemination to developing country breeding
contribution to cluster K2 and Milan to cluster K3. Pro-
programs.
genitor Tinamou contributed about equally to clusters
Higher-order repeat motifs were applied in our sam-
K2 and K3. Twenty of the 68 genotypes had durum
ple of EST-SSRs. In comparison to dinucleotide motifs,
(T. durum Desf.) wheat, Chinese wheat, or synthetic
they are generally less polymorphic and insensitive to
hexaploid wheat in their pedigree. These lines were
single-nucleotide polymorphisms in the flanking regions
scattered all over the principal coordinate analysis plot
of the SSRs, which facilitate the designation of allele
because of different sources of Chinese lines and Aegi-
sizes (Chakraborty et al., 1997; Song et al., 2002; Mogg
lops squarrosa L. in the synthetic hexaploid wheats used
et al., 2002). This explains the higher reproducibility
as crossing parents.
and the lower degree of heterozygosity or heterogeneity
for EST-SSRs compared with genomic SSRs in our
DISCUSSION
study.
EST-SSRs are assumed to reflect more accurately the
Use of Genomic and EST-SSRs
effects of selection under which the germplasm has been
in Breeding Programs
developed. However, only a slightly better differentia-
Genomic SSR markers have been intensively used to
tion of the allele frequencies between MEs was observed
detect the variability between bread wheat genotypes,
for EST-SSRs than for genomic SSRs. Wheat is highly
but the large genome size of wheat is a challenge in
autogamous and, therefore, differential selection of fit-
identifying sufficiently robust and informative SSRs for
ness-related target loci will also affect genomic SSRs
fingerprinting. EST-SSRs present a novel source of
linked to them. Further evidence shows a functional
SSRs and have some intrinsic advantages over genomic
importance of genomic SSR structures, which may cause
SSRs. They can be developed from available EST data-
some form of balancing selection (Innan et al., 1997; Li
bases and their frequency is abundantly high in tran-
et al., 2000; Li et al., 2002).
scribed regions (Morgante et al., 2002). A concern is
Future opportunities to combine markers and pheno-
that the coding character of EST-SSRs limits their level
typic data in association studies may improve the appli-
of polymorphism.
cation of EST-SSRs in the evaluation of germplasm, as
Our results agree with other studies in rice (Cho et
exemplified in maize by Thornsberry et al. (2001). A
al., 2000), grape (Vitis ssp., Scott et al., 2000), and wheat
specific marker could then be used to examine the func-
(Eujayl et al., 2002) in that the overall level of polymor-
tional diversity at a certain locus. We speculate that
phism for genomic SSRs was higher than for EST-SSRs.
However, compared with the latter study with 64 durum EST-SSRs from genes contributing to specific ME adap-
Table 5. Mean coefficient of parentage (above diagonal) and genetic similarity estimates (below diagonal) between five CIMMYT
megaenvironments (ME).
ME1IR ME1HT ME2 ME3 ME4
ME1IRR 0.16 ⫾ 0.12† 0.11 ⫾ 0.06 0.11 ⫾ 0.05 0.18 ⫾ 0.10
ME1HT 0.57 ⫾ 0.05 0.10 ⫾ 0.05 0.10 ⫾ 0.07 0.16 ⫾ 0.10
ME2 0.62 ⫾ 0.06 0.59 ⫾ 0.05 0.14 ⫾ 0.11 0.12 ⫾ 0.04
ME3 0.56 ⫾ 0.05 0.58 ⫾ 0.08 0.59 ⫾ 0.06 0.12 ⫾ 0.06
ME4 0.58 ⫾ 0.06 0.59 ⫾ 0.07 0.62 ⫾ 0.07 0.60 ⫾ 0.07
† Standard deviation.
