Abstract: Objective Based on the field trial on the triploid progeny produced from crossing between tetraploid and diploid birches, several superior parents and their combinations were selected with excellent growth and wood properties. This study aims to provide a scientific basis for parent selection in establishing birch seed orchards. Method This experiment was conducted in a field trial forest of 10-year-old progeny of 40 full-sib families derived from different parental combinations. Tree height (H), diameter at breast height (DBH), single tree volume, wood basic density, cellulose content, hemicellulose content, lignin content, fiber aspect ratio traits and retention rate of planted trees were measured respectively. SPSS 22.0 and WinNCII were used analyze variance, general combining ability (GCA) and specific combining ability (SCA) of each trait. Membership function and principal component analysis (PCA) were also used to identify the best overall performance of parental combinations. Result 1) Except the traits of wood basic density and tree retention rate, the variance of parental combination, GCA and SCA of remaining seven traits reached significant or extremely significant levels. The coefficient of variation of the seven traits ranged from 6.34% to 55.29%, and the broad-sense and narrow-sense heritability were between 38.83%–73.72% and 22.23%–68.48%, respectively. Some traits had significant variation and strong genetic ability among families, indicating that the selection opportunities were expected in this study. 2) Further analysis on the seven traits showed that the differences in the GCA and SCA effects of tree height, DBH, volume, lignin content and fiber length/width reached significant or extremely significant levels. The additive effect of parents was significant on traits of DBH, volume, cellulose hemicellulose content and fiber length/width ratio. By comparing the additive effect of maternal parent and paternal parent, it was found that the maternal parent had much greater effect on the observed traits than the paternal parent effect. The variance component of maternal parent was 43.11%–94.15%, while that of paternal parent was 0–4.86%. It was found that the lignin content and tree height traits were strong controlled by SCA effect only, which accounted for 52.03% and 44.55%, respectively. 3) Based on the combined analysis of membership function and PCA with GCA and SCA effect, F4, F7 and F10 were selected as superior male parents and Q19 with Q103 as superior female parents for growth or wood fiber traits, of which the volume genetic gain of Q103 reached 8.60%. Q33×F1, Q103×F10, Q13×F12 and Q19×F11 were selected as superior parental combinations, of which the volume genetic gain of Q33×F1 and Q103×F10 were 16.33% and 15.62% , respectively. Conclusion The main contributor to the performance of traits in this study is the additive effect of birch maternal parents. Based on growth and wood property measurements of birch triploid families and statistical analysis, three excellent male parents, two female and four superior parental combinations have finally been selected. This study provides a scientific basis for the establishing next generation birch seed orchard.

Key words: birch, triploid, tester mating design, combining ability analysis