between AA genome wild rice species
M. Akimoto1, Y. Shimamoto1 and H. Morishima2
Phylogenetic relationships among AA genome wild species related
to cultivated rice have been studied by several researchers (reviewed by
Morishima et a!. 1992), but their results were not always consistent. This
is probably due to different tools as well as due to insufficient sample
sizes used in respective studies. We examined variability of 22 quantitative
characters and 29 isozyme loci for a total of 168 strains of five AA genome
wild species; 57 strains of 0. rufipogon from Asia and Oceania, 49 strains
of 0. glwnaepatula from Latin America, 19 strains of 0. meridionalis from
Australia, 23 strains of 0. barthii from Africa and 20 strains of 0. longistaminata
from Africa. Polymorphisms at phenotype and isozyme loci were examined
by principal component analysis and scatter diagrams were generated using
the first and second principal component scores.
A scatter diagram of 148 strains plotted by the first and
second principal component scores based on the data of 22 characters was
generated. The strains of 0. rufipogon and 0. g!wnaepatu!a were distributed
in wider ranges than 0. meridionalis and 0. barthii, and seemed to be divided
into continuously varying two and three groups, respectively; 0. rufipogon-I,
H and 0. glumaepatula-I to ifi (Fig. la). The strains of 0. longistaminata
|
Research Notes
|
79
|
were not included in this character analysis, because they
did not flower during our study period, and consequently sufficient data
were not available. 0. rufipogon-I and 0. glumaepatu!a-I were plotted as
overlapping each other. While, 0. rufipogon-II overlapped with 0. meridionalis
and 0. barthii. Judging from character measurments, the former group tended
to show much resource investment to vegetative growth rather than seed
production, delayed reproduction and high pollen productivity, while the
latter group showed the opposite trend. We could consider the former group
as perennial-type wild rice and the latter as annual-type wild rice. Multiple
variant T2 test proved no significant differences at 5% level between principal
component scores of two perennial wild rice, 0. rufipogon-I and 0. g!umaepatula-I,
as well as between three annual wild rice, 0. rufipogon-II, 0. meridionalis
and 0. barthii (Table 1). 0. glumaepatula-H and -III differed significantly
from other taxa, indicating that they developed unique phenotypes, respectively.
Among 29 loci of 15 enzymes analyzed, 26 loci of 13 enzymes were polymorphic
for the entire germplasm evaluated. A scatter diagram of 168 strains plotted
by the first and second principal component scores revealed that the strains
belonging to five species formed respective clusters and were separated
from each other (Fig. Ib). Multiple variant T2 test proved significant
differences between all pairs of five species at 5% level (Table 1), indicating
that each species evolved with different isozyme genotypes. Intraspecies
differentiation of 0. rufipogon into 0. rufipogon-I and II as classified
in phenotype analysis was not clearly recognized (Fig. lb and Table 1).
On the other hand, 0. glumaepatula-1I was plotted separately from 0. glumaepatula-I
and III to some degree (Fig. ib), however, principal component scores did
not differ significantly at 5% level between 0. glumaepatula-I, II and
Ill (Table 1).
Phenotypes which are subjected to natural selection often
evolve convergently unlike isozyme genotypes which are considered to be
rather neutral. Plants, even with different
|
Table 1. Mean values and standard deviations (in parenthesis)
of the first and second principal component scores at 22 quantiative characters
and 29 isozyme loci
variant T2 test with 0.05 significant level.
|