Application of narrow sense canonical characters to cucumber breeding

Cucurbit Genetics Cooperative Report 16:30-34 (article 10) 1993

Hongwen Cui and Yongtao Qi
Department of Horticulture, Northwestern Agricultural university, Yangling, Shaanxi, 712100, P.R. China; Beijing Vegetable Research Center, Beijing, 1000081, P.R. China

Previous studies have shown that the correlation between heterosis of early yield in F1 hybrids and its associated component (ACT) traits was near 1.0. Early maturity of F1 hybrids is closely related to the trait constitution of parental lines. In order to develop a hybrid which posesses early maturity heterosis, parental lines must have both early maturity and ACT. Thus, the establishment of a reliable and effective selection scheme is essential. Such a scheme should posess the following: 1) a means for overcoming complicate correlations among important agronomic characteristics and unfavprable traits; 2) a means for overcoming low heritabilities often associated with quantitative characters, and, 3) a means for overcoming interference of non-additive effects in the early generations. A selection scheme which incorporates independent multiple characters with maximum narrow sense heritabilities may be reliable and effective during the course of pedigree breeding. Such a scheme is described below.

The additive variance and covatiance of traits were separated from the genetic variance and covariance using analysis of combining ability covariance with eight parental inbred lines and 16 F1 hybrids. This produced a narrow sense correlation heritability matrix (H N). The HN was used to construct two narrow sense canonical multiple characters with maximum narrow sense heritabilities. Four inbred lines, Changmi-1, Zanqiu-m, Pinli-m, and Hei235 were used as male parents and four inbred lines, Yue82, Jing4-3-1, Xilong58-5, and 7742 were used as female parents to produce an incomplete diallel.

The resulting F1 hybrids were evaluated in a randomized complete block design with three replications. the narrow sense correlation heritability for two traits Xi and Xj (h2Nij) was estimated by means of variance and covariance of combining ability according to a random model.

The narrow sense correlation heritability matrix:

HN = h2N1 h2N12 h2N1m
h2N21 h2N2 h2N2m
h<2Nm1 h2Nm2 h2Nm

The concepts of narrow sense canonical character were expressed as:

P*Ni = bi’p (I = 1, 2,……….S, S Such that p*N must satisfy the following two terms: 1) p2(b’p, b’A) = max and 2) D(b’p) = 1,
The phenotypic random vector P = (p1 p2…..pm)’ and
The additive random vector A = (a1, a2…….am )’ consisting P*Ni is calculated by solving matriq equation (Hn -;Φ2Rp) β = 0.

The following traits were chosen as objective traits for the comprehensive selection scheme: 1) early hield per plant (S1); 2) number of early fruit (S2); 3) number of early female flowers on the main stem (S3); 4) total number of branches (S4); 5) number of leaves at the time of first flowering (S5); 6) the nodeposition at which the first female flower appeared (S6).

HN was estimated by means of combining ability using covariance analysis. Six eigenvalues and six eigenvectors of HN for Rp were obtained by solving the matrix equation (HN – 2Rp) β = 0 (Table 1).

Table 1. The broad sense eigenvalues and eigenvectors of HN for Rp.

Eigenvalues

0.998

φ1

0.773

 φ2

0.258

φ3

-0.895

φ4

-2.520

φ5

-5.306

φ6

Adjusted eigenvalues 0.998 0.773 0.258 0.000 0.000 0.000
S1 0.449 0.492 -1.187 0.356 0.081 -2.382
S2 -0.315 0.074 -0.094 0.328 0.099 4.179
Eigenvectors S3 -0.353 -0.251 -.385 0.482 -2.259 -1.538
S4 0.062 0.640 1.106 1.452 -0.164 1.025
S5 0.192 -0.904 -0.610 0.788 0.765 -1.482
S6 0.490 0.077 -0.963 -0.902 -2.457 0.818

Eigenvalues (φ) reflect the narrow sense heritabilities of corresponding canonical characters in biological sense (Table 2). For example, φ 23 = 0.2583, it is very low as a heritability, φ24, φ25, φ26 are negative numbers and can be regarded as zero.

Table 2. The heritabilities (φ21) and their corresponding eigenevectors for the broad sense and narrow sense canonical characters φ21, φ22, φ23 used to structure canonical characters.

Heritabilities

Bigenvectors

s1
s2
s3
s4
s5
s6
φ21 Broad sense 0.999 -1.640 2.184 0.194 0.836 -1.984 1.283
Narrow sense 0.998 0.449 -0.315 -0.353 0.062 0.192 0.490
φ22 Broad sense 0.999 0.517 -1.290 1.450 0.092 0.782 -0.823
Narrow sense 0.773 0.492 0.074 -0.251 0.640 -0.904 0.077
φ23 Broad sense 0.928 0.382 -0.743 -1.508 -0.906 0.113 -1.542
Narrow sense 0.258 -1.187 -0.094 0.385 1.106 -0.610 -0.963
φ24 Broad sense 0.897 1.118 -1.364 0.473 1.109 0.167 -0.990
Narrow sense 0.000 0.356 0.328 0.482 1.452 0.788 -0.902
φ25 Broad sense 0.785 1.600 -2.837 1.800 -1.416 0.143 1.515
Narrow sense 0.000 0.081 0.099 -2.259 -0.164 0.765 -2.457
φ26 Broad sense 0.724 -0.863 -0.896 0.405 0.244 -0.143 0.847
Narrow sense 0.000 -2.382 4.179 -1.538 1.025 -1.482 0.818

The broad sense canonical heritabilities is much larger than those of the narrow sense canonical characters. These results indicate that genes controlling each objective trait have many non-additive effects. There is a large difference between broad sense and narrow sense canonical characters in eigenvalues and weighting coefficients for objective traits.

The first canonical character (P*N1). In accordance with the quantity and direction of weighting coefficient of each target trait, we used a reverse P*N1 selection. That is, selecting the lineage with small P*N1 value (i.e., the lineage with low node position at which the first female flower appeared, more female flowers in main stem and early fruits harvested, fewer branches and leaves at first flowering). These characteristice are all important in early maturity.

The second canonical character (P*N2) . We employed a positive P*N2 selection. that is, the lineage with fewer leaves at first flowering and high early fruit yield were selected. Nevertheless, P*N2 selection was ineffective in reducing the number of branches.

Table 3. The narrow sense canonical character values of 8 parental and 16 hybrids (F1)

Narrow sense canonical character 1 (P*N1)

Narrow sense canonical character 2 (P*N2)

Yue 82 (M1) 1.0757 -3.7898
Jing 4-3 (M2) 1.6548 -0.8218
Xilong 58-5 (M3) 1.6158 -1.9314
7742 (M4) 3.8848 -3.5500
Changmi-1 (F1) 1.3303 -1.4092
Zhangqiu-m (F2) 7.2461 -6.9080
Pinli-m (F3) 1.9358 -1.9746
Hei235 (F4) 3.4756 -1.5404
H1 0.3051 -2.2553
H2 2.2231 -2.7390
H3 2.3907 -1.7190
H4 1.6023 01.4923
H5 1.2318 -0.3444
H6 2.5815 -1.1398
H7 2.0961 -0.5351
H8 2.8573 -0.0409
H9 2.2798 -0.9479
H10 3.1833 -1.6005
H11 2.4834 0.0571
H12 3.1118 -1.1334
H13 2.7619 -2.2661
H14 3.5662 -2.1424
H15 3.1481 -1.3824
H16 3.0969 -1.3155

F1, M2, and M3 (Table 3) were selected in P*N1 reverse selection and F1, F4, M2, and M3 were selected in P*N2 positive selection. F1 M2 and M3 were all chosen in the selection of P*N1 and P*N2. Data indicate that all inbred lines F1 (Changmi-1), M2 (Jing4-3-1) and M3(Xilung 58-5) posess better characteristics for early maturity. However, in the course of P*N2 selection, none of the inbred lines were outstanding and thus all lines tested have shortcomings in this regard.

Maximum heritabilities and independence of canonival characters embody the advantages of this breeding method (1) (2) . Yang De and DaiJunti (1983) reported on canonical character selection in the offspring of wheat populations (3) . Canonical analysis of early maturing cucumber lines overcame serious interferences with non-additive effects and thus is viewed as a reliable comprehensive selection scheme for early generations. As an overall appraisal, it can also define the best combinations from which unique early-mature diphyletic lines can be sunthesized during the course of pedigree selection. Nevertheless, only the use of ines with maximum broad sense heritabilities of canonical characters can avoid the serious interferences with broad sense canonical selection in early generations. In order to overcome such defeats, new breeding methods should be developed which allow for the establishment of narrow sense canonical charcters with maximum narrow sense heritabilities. Such methods will increase the accuracy and reliability of canonical selection and thus overall appraisal of parental lines and F1 hybrids.

Literature Cited

  1. Yongtao, Q. and C. Hongwen. 1991. High effective multiple selection of parental lines of cucumber hybrid with strong early mature heterosis. Scientia Agriculture Sinica 24:51-58.
  2. De, Y. and D. Junti. 1983. Approaches to canonical correlations of multiple quantitative traits. (I). canonical character and its heritability. J. Huanan Agr. college, No. (1): 27-36.
  3. De, Y. and D. Junti. 1983. Approaches to canonical of multiple quantitative traits. (I) canonical character selection for the offspring populations of crossing combination of wheat. J. Hunan Agr. college, No. (1):37-51.