Linear discriminant model construction based on microsatellite alleles for geographic distant populations of sea cucumber(Apostichopus japonicus)

A Discriminant Analysis of Principal Components(DAPC)method was adopted to construct a germplasm resource prediction model at individual level,for the sea cucumber Apostichopus japonicus.One hundred and thirty six alleles generated by twenty microsatellite markers were used as the raw data input.Three populations were sampled from Vladivostok,Russia(RS,131.55 E,43.08 N)and Dalian,China(CN1,CN2).Principal Component Analysis(PCA)and Discriminant Analysis(DA)were carried out sequentially.Linear Discriminant functions were constructed with the two China subpopulations as one overall CN population and two separated populations respectively.The results showed that with 75% individuals as training set and the remaining 25% as test set,the accuracy of constructed model reached up to about 80% for the three-pop model and over 90% for the two-pop model.Conventional genetic diversity and genetic structure analysis showed moderate genetic differentiations among the three geographic populations of sea cucumber,and relatively close genetic distances(D₁=0.18,D₂=0.159).Nice accordance of results of conventional analysis and our novel analysis indicated that the current method can be practically well applied on prediction and discrimination on germplasm resource at individual level.To our knowledge,this is the first application of DAPC with molecule marker information to genetic study.The novel method built in the current study can be well applied to the practical breeding project in future.