have a lower than expected number of heterozygotes,
coupled with a consistent, albeit non-significant, under-
representation of G. raimondii alleles. On the same chromo-
some, loci between G1006 and pAR168a, particularly
pAR21a and
pAR24b, have fewer
G. raimondii homozygotes
than expected. The loci in the center of LG D5 (Fig. 8) have
fewer G. raimondii alleles than expected. This is coupled
with a slight over-representation of heterozygotes toward the
ends of the region. Finally, a 28.8 cM region on the lower
end of D9 (Fig. 7) contains five loci with an excess of het-
erozygotes.
Identification of orthologous and homoeologous LGs
Of the 659 potentially orthologous (related by common
ancestry and speciation, viz., A vs. D, A vs. A
t
, and D vs. D
t
)
and
homoeologous
loci
(related
by
duplication
via
allotetraploidy, viz., A
t
vs. D
t
), 424 (64%) fall into 12 suites
of loci that unite at least one A, D, A
t
, and D
t
LG. A 13th
suite of loci unites one D and one A LG with two different
LGs previously assigned to the D
t
genome, Chr. 22D and
LG D05 (Fig. 2) (Reinisch et al. 1994). Linkage group LG
D05 was tentatively assigned to the D
t
genome on the basis
of a single locus,
P11–23 (Reinisch et al. 1994). The rela-
tionships inferred here, however, suggest that LG D05 is of
A
t
genome origin, an inference supported by the addition of
new markers to the allotetraploid map (Paterson unpub-
lished). These 13 suites of loci are used to infer 13 groups of
putatively orthologous and homoeologous LGs, the number
expected in an allotetraploid derived from two diploids with
haploid complements of 13 chromosomes.
For discussion, the putatively orthologous and homoeologous
suites of LGs are termed homoeologous assemblages, or
HAs, and are denoted by numbers that correspond to their il-
lustrations in Figs. 1–10. The inclusion of multiple LGs for
a single genome within one HA implies that they represent
segments of the same chromosome (e.g., LG D10 and LG
D01 in HA 1; Fig. 1). Figs. 7 and 8 include two and three
HAs, respectively, united by A diploid genome transloca-
tions and are designated 7A, 7B, 8A, 8B, and 8C (discussed
below). Two A genome LGs (A21, A23) and two D genome
LGs (D15, D17) could not be included in any of the 13 HAs.
Four LGs from the allotetraploid (LG U07, Chr. 17D suppl.,
LG U02, and LG A04) had loci with putative orthologues in
at least one diploid LG but could not be assigned to one of
the HAs because of insufficient or contradictory data. These
unassigned LGs are shown in Fig. 11, with the map loca-
tions of putative orthologs indicated. No loci mapping to
allotetraploid LGs U09, A08, D11, or U05 had apparent
orthologous loci in either diploid map and are not consid-
ered or illustrated. Linkage groups U02 and A04 are unique
because all but one of their loci have homoeologues on other
LGs, but almost none of these homoeologous loci map to
LGs within the same HA. Either these LGs are artifactual or
they represent AD genomic regions that are extensively rear-
ranged. This “mosaicism” also occurs to a lesser extent on
the upper end of LG U01 (Fig. 5). The lower end of this LG
is clearly homoeologous to the A, D, and A
t
LGs in this HA,
but potential homoeologues to the loci on the upper end map
to a variety of LGs in other HAs, which again suggests ex-
tensive rearrangement.
Deviations from colinearity
Comparison of locus orders among LGs within HAs re-
vealed 19 locus order differences for which the assumption
of colinearity requires accepting an alternate locus order on
one or more LGs that is more than 100 times (LOD > 2) less
likely than the preferred order (Table 1). Nine of these in-
volve only two loci, and thus only weakly suggest the possi-
bility of a rearrangement. The remaining 10 involve three or
more loci and imply one or more inversions. Because these
are low-density maps, we clearly have underestimated the
actual number of rearrangements.
Confirmation of A
t
genome translocations
Homoeologous assemblages 7A & 7B (Fig. 7) comprise
two sets of D
t
, A, and D LGs (Chr. 20D/D9/A6 and
A14/D12/LG D07) with no evident relationship, except that
both contain loci with homoeologs mapping to A
t
LGs Chr.
5A and 4A. The homoeologous relationships shown (Fig. 7)
provide evidence that Chr. 5A and 4A have undergone a re-
ciprocal translocation relative to their D
t
, A, and D counter-
parts (i.e., after polyploidization). Both Chr. 4A and 5A have
two non-overlapping sets of loci, with each of the two sets
showing homology to one of the two sub-assemblages
(HA7A and HA7B) in Fig. 7. This inference is strengthened
by classical studies that suggest the A
t
genome will differ
from the D
t
, D, and A genomes by two reciprocal
translocations, one of which involves A
t
chromosomes 4 and
5 (Brown and Menzel 1950; Gerstel 1953; Gerstel and
Sarvella 1956; Menzel and Brown 1954; Brown 1980;
Menzel et al. 1982). Furthermore, the two A
t
LGs included
in Fig. 7 were assigned by Reinisch et al. (1994) to A
t
chro-
mosomes 4 and 5.
Inspection of this compound HA revealed several incon-
sistencies. The A6 and A14 homoeologs to pAR219b,
G1033a, and A1172 on Chr. 4A are interspersed with loci
whose homoeologs map to Chr. 5A. The intercalation of
G1033 and
A1172 with
A1159 and
pAR206 on LG A14 may
reflect a post-translocation inversion (the order of these loci
on A14 is inverted relative to D12). Loci pAR219 and A1543
on Chr. 4A are also intercalated with loci with homoeologs
mapping to Chr. 5A. In this case, it is unlikely that an inver-
© 1999 NRC Canada
186
Genome Vol. 42, 1999
Figs. 1–11. RFLP maps of diploid and allotetraploid
Gossypium. The A and D linkage groups are portrayed with their counterparts
from the allotetraploids (A
t
and D
t
; redrawn from Reinisch et al. 1994). Each figure represents homoeologous assemblages (HAs) of A,
D, A
t
, and D
t
linkage groups. Loci mapped to more than one linkage group (LG) within an HA are shown in bold. Bracketed numbers
beside loci indicate additional HAs (which correspond to Fig. numbers) with potentially homoeologous loci revealed by the same
probe. The order of linkage groups within HAs is (from left to right) D
t
– D – A – A
t
, except for in Figs. 7, 8, and 11, which are
discussed individually. When a genome is represented by multiple linkage groups, they are arranged in a single column. Lines
connecting orthologous and homoeologous loci (see Table 1 and text) indicate statistically supported deviations from colinearity. Loci
in boxes are duplicated within that linkage group.