pAR3 maps to a corresponding region of A4; and
G1045b,
pAR319a, and
pAR21 map to the corresponding region of
LG A05. In HA5, these loci variously map to median re-
gions of LG D02, D7, and A13, and the upper region of LG
A03.
HA2 versus HA3
In HA2, Probes G1155 and G1070 reveal loci mapping to
D8 and G1155 reveals a locus on Chr. 22D; in HA3 these
probes reveal loci that map to D6 and A12.
HA2 versus HA5
In HA2, pAR101 and PXP2–75 map to the upper regions
of Chr. 22D and LG D05 and the former locus maps to D8.
In HA5, one or both of these loci map to the upper region of
D7 and the median region of LG U01, which itself repre-
sents the most distal region of homology with the other link-
age groups in HA5.
HA2 versus HA7A
pAR218, G1045, G1070, P6–25, and pAR179 map to
submedian regions of Chr. 22D, D8, and LG D05. In HA7A,
homologous loci map to D9 and Chr. 20D, albeit not as a
discrete block.
HA4 versus HA5
P10–56 and
A1214 map to the lower ends of LG D03 and
LG A02, respectively, in HA4. In HA5, duplicates of both
© 1999 NRC Canada
190
Genome Vol. 42, 1999
Fig. 3. HA3. The order of
pAR288, A1737, A1124, and
pAR127 on D6 is inverted relative to A12, as is
A1606 and
pAR8 on Chr. 23D
versus D6.
loci map to the lower end of LG D02 whereas only the latter
locus is duplicated on LG A03. Both of these chromosome
segments show evidence of homology with an additional
chromosomal region marked by HA10.
HA5 versus HA10
The loci PXP4–75, G1261, A1759, pAR144, P6–57 are
variously distributed on LG U01 + LG D02, D7, and LG
A03 in HA5. Duplicates of these loci map to HA10, where
one or more of the constituent loci map to LG D04, D11,
and Chr. 10A.
HA6 versus HA8B
In HA6, A1267 and A1345 map to A17; in HA8B dupli-
cates of these loci map to D3. We note that A1345 also maps
to LG A02 on HA4, raising the possibility that A17 is im-
© 1999 NRC Canada
Brubaker et al.
191
Fig. 4. HA4. The order of A1168, pAR309, A1590, and pAR118 on A16 is inverted relative to D4 and LG D03.
properly associated with HA6. A more highly saturated A
genome map should resolve this issue.
HA7A versus HA9
In HA7A, pAR26 and P12–20 map to orthologous regions
of D9 and Chr. 5A, respectively. Locus pAR218 also maps to
D9. Each of these three loci maps to one or more corresponding
regions of HA9 on linkage groups Chr. 25D, D10, and Chr.
10A.
HA7A versus HA10
P5–61, A1751, and
pAR48 map to corresponding regions
of Chr. 5A, A6, D9, and Chr. 20D in HA7A; in HA10 these
loci map to the subterminal regions of LG D04, D11, and
Chr. 10A.
Variation in recombination rates among genomes
To evaluate potential differences in recombination rates
among the mapped genomes, recombinational distances
(cM) between pairs of loci in colinear regions were summed
across the A, D, A
t
, and D
t
LGs (data available from CLB).
To remove any bias due to map density, each summation
was
calculated
directly
from
recombination
fractions
(Haldane 1919) for every pair of loci. When summed across
all
loci,
the
two
diploid
F
2
populations
differ
in
recombinational length by 5.8% (Table 2). Similarly, the two
genomes
of
the
allotetraploid
F
2
population
are
recombinationally equivalent (4.8% difference overall). The
two diploid F
2
populations, however, were genetically
shorter than their A
t
and D
t
counterparts, differing by 52%
and 59%, respectively.
Genome evolution in Gossypium
Comparisons among the four maps demonstrate that gene
order and synteny are sufficiently conserved among the A,
D, D
t
, and A
t
genomes that the expected 13 assemblages of
homoeologous linkage groups were easily identified. The re-
sulting maps confirm and clarify some of the inferred
homoeologies between linkage groups described previously
for allotetraploid cotton (Reinisch et al. 1994).
These comparisons among the A, D, D
t
, and A
t
linkage
groups provide genetic evidence for structural rearrange-
ments that have altered gene order and synteny. Map com-
parisons
revealed
two
reciprocal
translocations
that
differentiate the A diploid genome from the A
t
genome
(Figs. 7 and 8). Colinearity in these same genomic regions
among the A, D, and D
t
linkage groups suggests that these
translocations occurred in the A
t
genome subsequent to
allotetraploid formation. In addition, RFLP evidence identi-
fies the genomic location of a translocation that differenti-
ates G. arboreum and G. herbaceum (Fig. 8).
In addition to these translocations, comparisons of gene
order among linkage groups within HAs revealed 19 puta-
tive inversions. These inversions differentiated two or more
of the four genomes in 12 of the 13 HAs (Table 1). In most
cases the loci involved in the inversions were not polymor-
phic in all four genomes and hence could not be mapped in
one or more of the four genomes. Because of this, for 13 of
the rearrangements it is unclear whether the event occurred
at the diploid or allotetraploid level, or in which particular
genome. The remaining six inversions probably arose subse-
quent to polyploidization. Five of these differentiate D from
D
t
while one differentiates A from A
t
.
These inferences are remarkably congruent with those
based on chromosome pairing behavior in diploid and
allotetraploid hybrids. Classical cytogenetic approaches to
comparative genome analysis have shown that the two A
diploid genome species, G. arboreum and G. herbaceum,
differ by a single translocation, and that genomes of these
two species differ from the A
t
genome of allotetraploid cot-
ton by two and three translocations, respectively (Brown
1980; Brown and Menzel 1950; Gerstel 1953; Gerstel and
Sarvella 1956; Menzel and Brown 1954; Menzel et al.
1982). Beasley (1942) inferred the presence of one or more
inversions between the A and A
t
genomes, claimed that the
D and D
t
chromosomes differ by a minimum of four struc-
tural differences (by implication inversions), and suggested
that structural differences exist between all of the A and D
chromosomes.
If we accept the congruence between Beasley’s observa-
tions and ours as evidence that a minimum of 13 inversions
and one translocation occurred during the divergence of the
A and D diploid genomes, and set aside the caveats regard-
ing the timing and genomic location of these events, the esti-
mated rate at which inversions and translocations arose is
1.4 to 2.8 events per million years (13 inversions and one
translocation divided by 5 to 10 million years). Similarly,
the rate of fixation for translocations and inversions in the
allotetraploids is estimated as 2.6 to 5.2 events per million
years (6 inversions and 2 translocations divided by 1 to 2
million years, minus the rate for diploids). This suggests that
the rate of fixation of inversions and translocations in
Gossypium allotetraploids may be similar to or only moder-
ately higher than that which occurred in the A and D diploid
genomes. These calculations clearly are approximations and
are subject to several sources of error. Nonetheless, the
© 1999 NRC Canada
Brubaker et al.
193
Shorter genome
L
1
(cM)
a
Longer
genome
L
2
(cM)
a
Percentage difference
b
A
532.73
D
563.77
5.8%
D
t
532.40
A
t
557.89
4.8%
A
134.00
A
t
203.00
51.5%
D
769.20
D
t
1219.38
58.5%
a
L
1
and L
2
were calculated by summing the genetic distance between each adjacent pair of loci.
b
Calculated as L
1
–L
2
/L
1
, where L
1
= shorter genome length, and L
2
= longer genome length.
Table 2. Genetic length differences among the diploid (A, D) and allotetraploid (A
t
, D
t
) Gossypium
mapping populations.