吉田総合案内へ戻る．
本頁は　Japan. J. Breed.（育種学雑誌）35 : 204〜208 (1985)　に掲載したものです．

和文要旨

カンショの収量や収量関係形質の年次間相関
吉田智彦（九州農業試験場作物第二部，熊本県西合志町，〒861-11）

カンショ育種試験における収量試験の精度を調べるために，収量や収量関係形質の約20年間にわたる年次間
相関をデータベース化した育種試験成績から計算した．相関は実生二年目と三年目，三年目と四年目，四年目と
五年目，五年目と六年目の収量試験間の共通供試品種について計算した．実生二年目と三年目の間の相関の値
は20年間の平均で，いも収量が0.29，切干歩合が0.58，圃場判定（収量，外観などを総合的に観察した判
定）が0.15であり，いも収量と圃場判定の相関の値が小さかった．三年目と四年目の間の相関の平均はいも収
量が0.49，切干歩合が0.68，いもの大きさが0.48，一株いも数が0.48，圃場判定が0.35であり，五年目と
六年目の相関の平均（標準栽培）はいも収量が0.72，切干歩合が0.84，いもの大きさが0.65，一株いも数が
0.67，つる重が0.62，圃場判定が0.30，ネグサレセンチュウ抵抗性が0.91，ネコプセンチュウ抵抗性が0.85
等であった．これらの結果から一般的には初期の試験間よりも後期の試験間で相関の値が高く，初期の選抜で
は収量よりも切干歩合やセンチュウ抵抗性について選抜をするほうが良いことがわかった．


Correlation between Successive Yield Tests for Agronomic Characters in Sweet Potato

Tomohiko Yoshida
Kyushu Natl. Agric. Exp. Stn., Nishigoshi, Kumamoto, 861-11

   Correlation coefficients over cultivars between successive yield tests in sweet potato
  (Ipomoea batatas [L.] LAM.) for yield and other agronomic characters were computed
  from breeding records for about 20 years to evaluate the accuracy of the yield tests.
  The means of the correlation coefficients between years were from 0.29 to 0.79 for root
  yield, from 0.58 to 0.88 for dry matter %, from 0.48 to 0.65 for root size, from 0.48 to
  0.69 for roots/plant, and from 0.15 to 0.58 for general appearance, etc,. The correlation
  coefficients were generally higher in tests at advanced stages than at early stages. It
  was suggested that selection at early stages should emphasize dry matter % or nematode
  resistance, instead of root yield.
  KEY WORDS : Sweet potato breeding, Correlation between years, Ipomoea batatas [L.] LAM.

                        Introduction

 For efficient breeding, accuracy in measuring cultivar differences in yield and other
agronomic characters is absolutely necessary. In sweet potato (Ipomoea batatas [L.]
LAM.) breeding, clones from seedling plants are propagated vegetatively and evaluated
in a continuous succession of preliminary yield tests and advanced yield tests. During
these steps of evaluation, cultivar differences often fluctuate from year to year. Correla-
tions over cultivars between years, or between successive yield tests, were computed
to evaluate the accuracy of the yield tests and to establish the more efficient breeding
strategy for selection of high yield sweet potato lines.

 In sweet potato, IURA and CHUUMAN (1953) obtained the year to year correlations from
0.66 to 0.86 for root yield and from 0.88 to 0.96 for dry matter %, and SAKAI (1964)
computed the year to year correlations from 0.52 to 0.76 for root yield and from 0.35
to 0.92 for dry matter % in replicated yield tests. These values were obtained in
three to four years, which I do not consider enough to estimate the variation of the
year to year correlation values. This report describes the results of computations of
year to year correlations for sweet potato yield and other agronomic traits for about
20 years.
                    Materials and Methods

 A computerized data base from breeding records at the Kyushu Agricultural Experiment
Station was constructed (YOSHIDA 1984) and utilized in this study. The data base includes
records of crossings, selection of seedling plants in the first year, and yield tests in
succeeding years, designated as T2 for the second year of testing, T3 for the third
year, and so forth, for more than 20 years. Using the data base, it is relatively easy
to retrieve the target items and make numerical analysis, even when they contain a
large amount of data.

 Correlation coefficients were computed between the values of lines in T3 in 1981
and the values in the previous year (i.e., T2 in 1980), between T3 in 1980 and T2 in
1979, ......, and between T3 in 1957 and T2 in 1956. They are designated as T23,
hereafter. Correlation coefficients between T4 and T3 (T34), between T5 and T4
(T45), between T6 and T5 (T56) were also computed. Data of some characters and
some years were not obtained. Correlation coefficients were computed for 21 years in T 56
(standard culture), 20 years in T56 (short-term culture), 25 years in T45 and T34,
and 16 years in T23. Average degree of freedom was six in T56 (standard culture),
five in T56 (short-term culture), 28 in T45, 71 in T34, and 49 in T23. Average degree
of freedom in T23 was less than that in T34 because some data were missing in T2
and the correlation coefficients were computed without these lines. The data used for
computation were from records of breeding for starch yield. Data of lines for food
use were not included.

 Correlation coefficients were computed for tuberous root yield (root fresh weight),
root dry matter %, root starch %, root size (in gram), number of roots/plant, vine
weight (fresh weight at harvest), general appearance of the plant (scored subjectively
from two (poor) to six (good) by judging generally of the yield, root appearance etc.)
and root appearance (scored from two (poor) to six (good), subjectively), and resistance
to root lesion nematode (Pratylenchus spp.) and resistance to root knot nematode
(Meloidogyne spp.) scored from two (susceptible) to six (resistant). The experimental
methods of yield tests were, generally, as follows. T5 and T6 were conducted in
standard culture and short-term culture. Standard culture of T5 and T6 was by three
replications of four rows with 71 cm between rows x 4.2 m row-length with 35cm between
plants (52 plants/plot), and planted in the middle of May and harvested in the late
October after about 150 growing days. Short-term culture was by four rows x 4.2m
row-length with 30 cm between plants (60 plants/plot), and planted in the early May
and harvested in the late August after about 110 growing days. T4 was by two
replications planting 48 plants (four rows)/plot. T3 was by no replication planting 24
plants (two rows)/plot. T2 was by no replication planting 8 plants (one row)/plot.
T4-T2 were conducted in standard culture. Roots yield was determined by weighing
marketable roots (>50 g).

 Nematode resistance was evaluated in a nematode-infected field by planting ten clones
per line with two or three replications.
 Computation was by GRIMS/CGS, a package program for data base management,
provided by the computing center at the Ministry of Agriculture, Forestry and Fisheries
of Japan.
                  Results and Discussion

 Frequency distribution of correlation values for root yield and dry matter % are shown
in Table 1 and Table 2. Mean and standard deviation values of the correlation coefficients
for yield and other agronomic characters are shown in Table 3.

 For root yield, the correlation coefficients were from 0.20 to 0.94 (mean : 0.72) in
T56 (standard culture), from 0.32 to 0.97 (mean : 0.79) in T56 (short-term culture),
from 0.16 to 0.87 (mean : 0.62) in T45, from 0.17 to 0.76 (mean : 0.49) in T34, and

from -0.21 to 0.90(mean : 0.29) in T23. Correlation coefficients between years are af-
fected by the genetic variation in the population and the environmental factors. Extreme-
ly poor lines should have been eliminated at early stages of yield tests, therefore the
genetic variation should have been progressively smaller in the advanced yield tests.
But the mean values in advanced yield tests were fairly high, although in some years
the correlation values were low. On the other hand, values in yield tests at early stage
were low. Especially, values in T23 were low, suggesting that selection for root yield
in T2 was not effective, except discarding the extremely poor lines.

 For dry matter %, which has close association with root starch %, the correlation
coefficients ranged from 0.55 to 0.99 (mean : 0.84) in T56 (standard culture), from 0.58
to 0.99 (mean : 0.88) in T56 (short-term culture), from 0.31 to 0.95 (mean : 0.78) in T45.
from 0.21 to 0.94 (mean:0.68) in T34, and from 0.14 to 0.97 (mean :0.58) in T23.
The mean values were higher than those for root yield. Therefore, selection in yield
tests at early stages should emphasize dry matter %, instead of root yield. Correlation
coefficients for starch % were also high.

 The mean values of correlation coefficients were from 0.48 to 0.65 for root size and
from 0.48 to 0.69 for roots/plant. The mean values for vine weight in T56 were 0.62
and 0.48. Root size, roots/plant and vine weight were supplementary characters for
measuring root yield. The year to year correlation coefficients of these characters were
not higher than the values for root yield itself. Therefore, emphasis should not be
placed on these characters in selection for higher yield. Correlation coefficients for general
appearance were generally low and from 0.15 to 0.58. Standard deviation values for
general appearance were high. The mean values of correlation coefficients for root appear-
ance were lower than those for root yield. As the characters judged subjectively, i.e.
general appearance and root appearance, had low correlation coefficients, modification of
the judgement is necessary to improve the evaluations of these characters. Correlations
for nematode resistance seemed to be stable from year to year because the mean correla-
tion coefficients were high (0.91 and 0.85) and the the standard deviation values were
low (0.06 and 0.14), showing that the selection for these characters had been satisfactory.
Selections for nematode resistance had been done only at advanced stages. The results
indicate that they should be practiced at earlier stages because of the accuracy of the tests
and the relatively small amounts of the experimental materials required for the tests.

 In short-term culture, the correlation coefficients were a little lower for vine weight
and higher for general appearance and root appearance than in standard culture. The
standard deviation value in short-term culture was higher for vine weight and lower for
root size and root appearance than in standard culture. In general, however, there was
not serious discrepancy between correlation coefficients in standard culture and short-term
culture and the accuracy of these two tests seemed to be nearly same.

                   Literature Cited
 IURA,M. and K.CHUUMAN 1953. Year to year correlation coefficients for yield and dry
    matter % in sweet potato. Nogyo gijutsu. 8 : 225〜226.
 SAKAI,K. 1964. Studies on the enlargenment of variations and the improvement of
   selection methods in sweet potato breeding. Bull. Kyushu Agr. Expt. Sta. 9: 247〜
      397.
 YOSHIDA, T. 1984. Construction of data base from sweet potato breeding records. Japan.
    J. Breed. 34 : 115〜120.

以上