Comparative genetic mapping of allotetraploid
cotton and its diploid progenitors
C.L. Brubaker, A.H. Paterson, and J.F. Wendel
Abstract: Allotetraploid cotton species (Gossypium) belong to a 1–2 million year old lineage that reunited diploid
genomes that diverged from each other 5–10 million years ago. To characterize genome evolution in the diploids and
allotetraploids, comparative RFLP mapping was used to construct genetic maps for the allotetraploids (AD genome;
n = 26) and diploids (A and D genomes; n = 13). Comparisons among the 13 suites of homoeologous linkage groups
permitted comparisons of synteny and gene order. Two reciprocal translocations were confirmed involving four
allotetraploid A
t
genome chromosomes, as was a translocation between the two extant A genome diploids. Nineteen
locus order differences were detected among the two diploid and two allotetraploid genomes. Conservation of colinear
linkage groups among the four genomes indicates that allopolyploidy in Gossypium was not accompanied by extensive
chromosomal rearrangement. Many inversions include duplicated loci, suggesting that the processes that gave rise to
inversions are not fully conservative. Allotetraploid A
t
and D
t
genomes and the A and D diploid genomes are
recombinationally equivalent despite a nearly two-fold difference in physical size. Polyploidization in Gossypium is
associated with enhanced recombination, as genetic lengths for allotetraploid genomes are over 50% greater than those
of their diploid counterparts.
Key words: restriction fragment length polymorphism (RFLP), Gossypium, evolution, polyploidy.
Résumé : Les espèces allotétraploïdes de cotonnier (Gossypium) appartiennent à un lignage vieux de 1 à 2 millions
d’années résultant de l’union de génomes diploïdes ayant divergé l’un de l’autre pendant 5 à 10 millions d’années.
Afin de caractériser l’évolution génomique chez les diploïdes et les allotétraploïdes, une analyse comparée de
cartographie RFLP a été réalisée pour établir des cartes génétiques chez les espèces allotétraploïdes (génome AD;
n = 26) et diploïdes (génomes A et D;
n = 13). Une comparaison des treize groupes de liaison homéologues a
permis d’étudier la synténie et l’ordre des gènes. L’existence de deux translocations réciproques impliquant quatre
chromosomes du génome allotétraploïde A
t
a été confirmée, tout comme une translocation entre les deux espèces
diploïdes à génome A existantes. Dix-neuf différences quant à l’ordre des gènes ont été décelées parmi les deux
génomes diploïdes et allotétraploïdes. La conservation des groupes de liaison parmi les quatre génomes indique que le
processus d’allopolyploïdisation n’a pas été accompagné de réarrangements chromosomiques importants chez le genre
Gossypium. Plusieurs inversions incluaient des loci dupliqués ce qui suggère que le processus menant à des inversion
n’est pas parfaitement conservateur. Les génomes allotétraploïdes A
t
et D
t
de même que les génomes diploïdes A et D
sont équivalents sur le plan de la recombinaison en dépit du fait que l’un soit pratiquement deux fois plus grand que
l’autre sur le plan de la taille physique. La polyploïdisation chez le genre Gossypium s’accompagne d’un accroissement
de la recombinaison puisque la somme des distances génétiques chez les génomes allotétraploïdes est plus de 50%
supérieure à celle de leurs contreparties diploïdes.
Mots clés : polymorphisme de longueur des fragments de restriction (RFLP),
Gossypium, évolution, polyploïdie.
[Traduit par la Rédaction]
Brubaker et al.
203
Polyploidy is common in plants (Grant 1981; Lewis 1980;
Leitch and Bennett 1997; Masterson 1994; Soltis and Soltis
1993; Stebbins 1950, 1971) and probably has been involved
in the evolution of all eukaryotes (Leipoldt and Schmidtke
1982; Sidow 1996; Spring 1997; Wolfe and Shields 1997).
In synthetic allopolyploids, illegitimate chromosome pairing
often disturbs meioses, and thus it is thought that evolution-
ary mechanisms which promote exclusively bivalent pairing
will be selectively favored in the critical first few genera-
tions following natural allopolyploid formation. However,
prior to the restoration of diploid-like chromosome behavior,
interactions among homoeologous chromosomes may result
in the rapid accumulation of structural rearrangements that
alter gene order and synteny (Ahn et al. 1993; Feldman et al.
1997; Leipoldt and Schmidtke 1982; Liu et al. 1998; Song et al.
1995).
The cotton genus,
Gossypium L., is an ideal system for
examining genome evolution in polyploids. Gossypium com-
Genome 42: 184–203 (1999)
© 1999 NRC Canada
184
Corresponding Editor: G.J. Scoles.
Received March 29, 1998. Accepted August 14, 1998.
C.L. Brubaker.
1
CSIRO Plant Industry, GPO Box 1600,
Canberra, A.C.T. 2601, Australia.
A.H. Paterson. Department of Soil and Crop Science, Texas
A&M University, College Station, Texas 77843–2474, U.S.A.
J.F. Wendel. Department of Botany, Iowa State University,
Ames, Iowa 50011, U.S.A.
1
Author to whom all correspondence should be addressed
(e-mail: curtb@pi.csiro.au).
prises approximately 45 diploid and five allopolyploid spe-
cies distributed throughout the arid and semi-arid regions of
Africa, Australia, Central and South America, the Indian
subcontinent, Arabia, the Galápagos, and Hawaii (Fryxell
1979, 1992). The diploid Gossypium species fall into eight
cytological groups, or genomes, designated A through G,
and K (Beasley 1940; Edwards and Mirza 1979; Endrizzi et
al. 1985; Phillips and Strickland 1966; Stewart, 1994). The
five allotetraploid species, indigenous to the New World, de-
rive from a single allopolyploidization event that united the
Old World A genome with the New World D genome, in an
A genome cytoplasm (Wendel 1989; Wendel and Albert 1992).
Judging from differences in meiotic pairing, Gossypium
allotetraploid genomes appear to be more fully “differenti-
ated” from one another than are the descendents of their dip-
loid progenitors (Endrizzi 1962; Mursal and Endrizzi 1976).
Specifically, allotetraploid-derived haploids form an average
of less than one bivalent per cell during meiosis, whereas
chromosomes of extant A and D diploid F
1
’s average 5.8 and
7.8 bivalents at metaphase I (Endrizzi and Phillips 1960;
Mursal and Endrizzi 1976; Skovsted 1937). These data sug-
gest that stabilization of the newly evolved allotetraploid in-
volved
genic
or
genomic
modifications
that
inhibit
homoeologous pairing while promoting exclusively biva-
lent formation between homologues (Kimber 1961).
To examine whether genome evolution in Gossypium
allotetraploids was also accompanied by chromosome struc-
tural rearrangements, we constructed RFLP maps for the A
and D diploid genomes, using a common set of nuclear
probes, most of which were previously or simultaneously
mapped in the allotetraploids (Reinisch et al. 1994). Com-
parisons
between
diploid
genomes,
between
diploid
genomes and their corresponding allotetraploid genomes,
and between allotetraploid genomes allowed us to address
whether allopolyploidization in Gossypium was accompa-
nied by rapid chromosomal structural change and compare
recombination rates among four homoeologous genomes.
The A and D genomic linkage maps were based on the multilocus
genotypes of 58 A genome (G. herbaceum L., A
2
[ = A
1
]–97, ×
G. arboreum L., A
2
–47) and 62 D genome (G. trilobum (Moc. &
Sessé ex. DC) Skovsted ×
G. raimondii Ulbr.) F
2
individuals. The
AD genomic linkage map was based on a
G. hirsutum race
‘palmeri’ × G. barbadense K101 F
2
population (see Reinisch et al.
1994 for details). DNA extractions followed Paterson et al.
(1993) or Brubaker and Wendel (1994). Restriction digestion,
electrophoresis, blotting procedures, RNA probe preparation, hy-
bridization protocols, and autoradiography followed Brubaker and
Wendel (1994).
Three hundred and twenty-one mapped DNA probes were se-
lected from six libraries: anonymous PstI nuclear fragments from
G. raimondii (“G”-series; n = 75), G. herbaceum subsp. africanum
(Mauer) Watt (“A”-series; n = 143), and G. hirsutum cultivar TM-1
[“M”-series (n = 1) and “P”-series (n = 16)]; low copy-number
nuclear sequences (Zhao et al. 1996) from G. barbadense L. cv.
Pima S6 (“PXP”-series; n = 8); and anonymous cDNA clones from
drought-stressed G. hirsutum accession T25 (“pAR”-series; n = 78)
(Reinisch et al. 1994). RFLPs between G. herbaceum and
G. arboreum and between G. raimondii and G. trilobum were re-
vealed using eight (EcoRI, HindIII, PstI, DraI, CfoI, BamHI, XbaI,
and EcoRV) and two restriction enzymes (EcoRI and HindIII), re-
spectively, the difference reflecting the relative levels of polymor-
phism between the A genome parents and the D genome parents.
We also assayed allelic segregation among the 58 A genome F
2
progeny at seven isozyme loci encoded by six enzyme systems:
aconitate
hydratase
(ACO1),
arginyl-specific
aminopeptidase
(ARG1),
leucyl-specific
aminopeptidase
(LEU1),
6-
phosphogluconate
dehydrogenase
(PGD1
and
PGD3),
triosephosphate isomerase (TPI1). Sample preparation, starch-gel
electrophoresis, and nomenclature follow Wendel et al. (1989).
Genetic interpretation of the RFLP phenotypes and nomencla-
ture followed the system of Reinisch et al. (1994). Loci are desig-
nated by probe, and arbitrarily assigned lower-case letters to
distinguish multiple segregating loci revealed by the same probe.
The letters A and D denote the A and D genome linkage groups,
respectively, followed by arbitrarily assigned numbers. MapMaker
Macintosh 2.0 (Lander et al. 1987) was used to infer the genomic
linkage maps (Figs. 1–11). Initial linkage groups were inferred
from two-point analyses using a minimum LOD score of 4.0 and a
maximum theta value of 0.40. Linkage groups were ordered by se-
lecting a subset of six or seven loci whose most likely order dif-
fered from the next most likely order by a LOD of two or more.
Remaining loci were added to the initial map using the “try” func-
tion. The “ripple” function confirmed the local order around new
loci and identified regions of uncertain order. Linkage-1 (Suiter et
al. 1983) was used to identify loci whose segregation ratios dif-
fered significantly (P < 0.05) from Mendelian expectations.
A genome map
The A genome map (Figs. 1–11) comprises 161 loci
encoded by 152 nuclear probes and 6 isozyme loci mapping
to 18 linkage groups (hereafter LGs). The 18 LGs encom-
pass 856 cM. Individual LGs include 2 to 24 loci (average =
8.94).
There was no discernible bias in allelic or genotypic fre-
quencies. Segregation ratios at only 14 loci (8.7%) from
seven LGs deviate significantly from Mendelian expecta-
tions (P < 0.05). This number is only slightly higher than
that expected from chance alone and none of the loci in
question occurred in discrete blocks.
Gossypium arboreum and
G. herbaceum differ by a recip-
rocal translocation (Gerstel 1953) and therefore the F
1
is a
translocation heterozygote. As a result, at least one of the A-
genome LGs should contain sets of loci for which three
point comparisons create unresolvable contradictions (e.g.,
locus A maps 5 cM from loci B and C, but locus B and C
segregate independently). These contradictory data will be
magnified in multipoint comparisons, making map orders
ambiguous. Linkage group A5 (Fig. 8) fits this description.
Comparisons to homoeologous LGs from the D, A
t
(“A” ge-
nome of the allotetraploids), and D
t
(“D” genome of the
allotetraploids) genomes reinforces this inference.
D genome map
The D genome map (Figs. 1–11) comprises 306 loci
encoded by 269 nuclear probes mapping to 17 LGs. The 17
LGs encompass 1486 cM. Individual LGs include 2 to 44
loci (average = 18.0).
In contrast to the A genome, there is evidence for segrega-
tion distortion within three genomic regions. Segregation ra-
tios at 37 loci (12.1%) from nine LGs deviate significantly
(P < 0.05) from Mendelian expectations. Twenty-four of
these loci map to three blocks on LGs D1, D5, and D9. On
LG D1 (Fig 1), three loci (pAR168b, A1559, and pAR173b)
© 1999 NRC Canada
Brubaker et al.
185