Cucurbit Genetics Cooperative
Other Crop Genetics Cooperatives
Home About Membership Reports Gene Lists Conferences Links Search NCSU
Cucurbit Genetics Cooperative Report 17:14-17 (article 4) 1994

Cucumber Cultivar Cluster Analysis

Hongwen Cui and Meng Zhang and Junjun Deng

Department of Horticulture, Northwestern Agricultural University, Yangling, Shaanxi, 712100, P.R. China; Xi'an Vegetable Research Institute, Shaanxi 710068 P.R. China

Cultivar genetic clustering resulting from differences in trait expression can be the basis for parental selection in hybrid breeding programs. With a prerequisite of fitting cucumber types to local consumer customs, it is expected that heterosis can occur between cultivars which are the most different. Such differences among traits can be selected from corresponding categories according to the principle of complementary traits.

Method. An experiment was conducted at the Horticultural Station of Northwestern Agricultural University during the summer of 1991. fifteen cultivars and inbreds possessing contrasting trait differences were evaluated in a two-direction randomized block design with 3 replications. the cultivars and inbreds used were: 1) 58-663; 2) Jin 4-3-1;3) 'Jinan'; 4) Heiden-1; 5) 121713; 6) 3511;7) Changmi-1; 8) 'Mai'; 9) 'Pingli'; 10) Yue 82; 11) Shifeng-8; 12) Jin 1717; 13) 'Baitoushuang'; 14) Jiyu-2; and 15) 'Najingci'. Twenty-two quantitatively inherited traits were evaluated during the cucumber growing season. The traits evaluated were: average yield per plant (x1); number of harvested fruits per plant (x2); average fruit weight (x3); early yield per plant (x4); number of harvested fruit per plant in the early stage (x5); average fruit weight in the early stage (x6); number of leaf at first harvesting stage (x7); leaf area at the first harvesting stage (x8); the node position of the first pistillate flower (x9); number of leaves at the end of growth season (x10); number of nodes at the end of growth season (x11); number of nodes at the end of growth season (x12); total number of effective branches (x13); the days from sowing to the first pistillate flowering plant in the population (x14); the days from sowing to the pistillate flowering of 50% of the plants (x15); fruit length (x16); fruit volume(x17); fruit diameter (x18); average fruit developing average rate (x19); downy mildew resistance (x20); developing relative fruit rate at mid-growth period (x21); and developing relative fruit rate at early stage (x22).

After comparing several methods of estimating genetic distance, Mahalanobis distance (pivotal condensation) was selected for cluster analysis. Genetic distance estimates were clustered using each of the following methods: 1) nearest-neighbor method; 2) furthest-neighbor; 3) median; 4) centroid; 5) group-average; 6) flexible group-average; 7) flexible: and, 8) squares sum of dispersion. The results of clustering were contrasted and analyzed in order to compare clustering methods and to determine the most representative traits for each cultivar group.

Results. Comparisons among different methods of clustering of cultivars and inbreds indicated that the spatial arrangement obtained by the different methods were not similar. Moreover, differences that were observed were mainly in the pattern of the merging of large genetic distances between cultivar and inbreds. The results of different clustering methods can be divided into three types:

The clustering features of the nearest-neighbor method was similar to that of the centroid method (Fig. 1). Fifteen cultivars and inbreds were clustered based on smallest to largest genetic distance, and the clustering core was not distinct. This is especially true of cultivars and inbreds which were not related (i.e. large genetic distance) such as 'Nanjingei' (#15), Jin 4-3-1 (#2), 'Pingli' (#9), Shifeng-8 (#11), and Changmi-1 (#7).

The clustering feature of the furthest-neighbor method and squares sum of dispersion methods were similar and clustering cores were distinct (Fig. 2). All cultivars and inbreds were affected mainly by early maturity and high productivity. Jin 4-3-1 (#2) 3511 (#6), 58-663 (#1) and Jiyu-2 (#14) all possessed high productivity and downy mildew resistance and were classified into one group. Changmi-1 (#7) and Pingli (#9), which possessed early maturity, formed one clustering core.

The clustering results of group-average (representing a flexible method) and median methods were similar (Fig. 3). Clustering cores were obvious when the genetic distances were short as depicted by the furthest-neighbor method. When genetic distances among cultivars and inbreds were larger, the nearest-neighbor method was not to be found an adequate classifier and groups were merged one by one.

In summary, the clustering of 15 cultivars and inbreds by the furthest-neighbor method fit the synthetic al expression of traits, and therefore can be used as a guide for the selection parents in a hybrid breeding program,. Six groups of cultivars and in breeds were analyzed further, and representative traits and common features were found (Table 1).

Discussion. Parental selection is the key for cross-breeding. Cultivars or inbreds with distinctly different genetic distances are often selected for crossing. Thus, it is necessary to select the most appropriate methods for estimating genetic distance. Different methods directly affect the clustering of germplasm. It is regarded that the Majalanobis distance estimated by picotal condensation is a good predictor of early yield and productivity. This genetic distance estimate showed cultivar differences for quantitative traits related to yield in this study. Thus, the use of this method of genetic distance estimation coupled with an appropriate clustering method can be an important guide to the selection of parents in a plant improvement program.

Table 1. cultivar clustering and representative traits.

Cultivar or inbreds
Representative traits
Typical features
1 3511, Jiyu-2, 58-663, Jin 4-3-1 x1, x2, x10 x16, x17 Much more number of leaves in growth season, long and large fruit, more fruits per plant, high total production.
2 Heiden-1. 121713, Jin 1717 x3, x6 , x21 , x22 Light fruit weight, fruit developing quickly, not highly productive (between high productive type and early maturity type
3 Jinan, Mai Jue 82, Baitou shuang x7, x8, x11 Large number and index of leaf area in early stage, nutritive growth luxuriant, not good at later stage.
4 Pingli, Nanjingei x15 , x20 , x4, x5, x14 Early maturity, not resistant to downy mildew, more harvested fruit and high early yield, typical of early maturity cultivars.
5 Shifent-8 x4, x2 , x13, x14 Early pistillate flowering, fairly early maturity and high production.
6 Chang mi-1 x1, x2 , x4 , x5, x13 Many harvested fruits, and effective branches, the typical of early mature cultivars.

Literature Cited

  1. Camussic, A. 1985. Genetic distance based on quantitative traits. Genetics 111:945-956.
  2. Grower, J.C. 1967. A comparison of some methods of clustering analysis. Biometrics 23: 623-636.
  3. Zhonghu H. 1988. Stability o distance analysis. J. Bejing, Agri. Univ. 14:380-386.
Home About Membership Reports Gene Lists Conferences Links Search NCSU
Department of Horticultural Science Box 7609North Carolina State UniversityRaleigh, NC 27695-7609919-515-5363
Page citation: Wehner, T.C., Cucurbit Genetics Cooperative;
Created by T.C. Wehner and T. Ng, 1 June 2005; design by C.T. Glenn;
send questions to T.C. Wehner; last revised on 11 December, 2009