Studies On New Methods For Constructing And Evaluating Core Collection Of Plant Germplasm Resources

Core collection is an important field of germplasm research and utilization, constructing and evaluating core subsets are main contents of the core collection. Worked examples consisting of a cotton germplasm population and a rice germplasm population were used to investigate the methods of core collection construction and evaluation on the basis of previous research. Main results were achieved as follows.A new method of construting core subsets by sampling parameter optimization (SPO) was proposed. First, the initial core subsets were developed by randomly spampling strategy. Next, the evaluating parameters in core subsets with different sampling percentages were used for principal component analysis, and sampling parameter (SP) was designed, which could represent the evaluating parameters of core subsets. SP was as the sampling standard when core subsets were constructed. Each accession was removed gradually from initial collection, which remains SP of core subset as closed to SP of initial collection as possible. The sampling was performed until the core subset reached to the given sampling percentage. When core subsets were constructed by SPO method, the evaluating paramters presented by SP could be as optimal as possible. The optimal method of getting SP was achieved after compared the different getting ways for SP. The results showed that core subsets constructed by SPO method could representative to the initial collection, and were more reasonable than least distance stepwise sampling (LDSS) method. The core subset constructed by SPO method gave a fixed core accessions when given a sampling perentage, it will be beneficial to the utilization and management of germplasm resource.Improved least distance stepwise sampling (ILDSS) was proposed after the mothod on constructing core subsets by least distance stepwise sampling (LDSS) was improved. The process as follows: the genetic distances between accessions were calculated and accessions were grouped by cluster analysis based on genetic distance. A subgroup with the least distance in the dendrogram was found according to cluster result, the accession had a smaller the least distance with other accessions in the collection in this subgroup was removed, and another accession in the subgroup was sampled. Then, genetic distances among the remained accessions were calculated again, and the sampling was performed by the same way until the accessions reached to the given sampling percentage. Similar to LDSS method, cluster methods did not need to be considered when using ILDSS method to construct core subsets. However, the core subsets constructed by ILDSS method well preserved genetic variation and structure of initial population and had a better representativeness than that of LDSS method.A method of evaluating core subsets combining genotypic value data and molecular marker information was proposed. The data of genotypic values was graded and changed into qualitative data in the study. The evaluating parameters for qualitative characteristics were used to evaluate changed data of genotypic values and molecular marker information. Then, the evaluating results of two kinds of data were combined, and the mixed evaluating parameters (Mmix_D, Mmix_I and Mmix_PIC) were proposed. Analysis of variance was used to compare the significanly different of the same evaluating parameter among different sampling percentages of core subsets. The number of homogeneous populations of Tukey's test was considered the estimation standard of evaluating parameters validity. The results showed that initial rice population evaluated by this new mixed evaluating parameters could distinguish more homogeneous populations, and had better validity than reparate evaluating parameters for genotypic values or for molecular marker information. The method of combined evaluating parameters is not only for genotypic values and molecular marker information, but also able to combine any other continuous data and discrete data together to evaluate core collection.