Zhang et. al.·Silvae Genetica (2015) 64-3, 81-90 Genetic parameters and prediction of breeding values in Castanopsis hystrix A.DC. for growth and form traits By F. Q. ZHANG1), B. XU1), H. X. YANG1), W. H. ZHANG1), B. Z. ZHU1), Y. X. WANG1), H. Q. LIAO1) and W. PAN1),*) (Received 24th September 2014) Abstract (WHITE et al., 2007). Castanopsis hystrix A.DC. is one of the most important and multifunc- Castanopsis hystrix A.DC. is one of the most important and multipurpose tree species native tional tree species native to China because of its to China. 157 open-pollinated families collected good growth characteristics, desirable wood from 11 provenances in Guangdong, Guangxi, properties, and broad utilization range (HUANG and Fujian province were used to estimate et al., 2012). It is found naturally in southern genetic parameters for height (H), diameter at China in Guangxi, Guangdong, Hainan, and breast height (DBH), ground diameter (GD), Yunnan provinces between 21°N and 35°N and and crown width (CW) for each province and 109°E and 117°E. Castanopsis hystrix can be combined three provinces at ages from 3 to 9 planted as pure stand or mixed with other years. The variance component was small and species such as conifers and firs. Over the past non-significant among provenances but was three decades, C. hystrix resources in China highly significant among families within prove- have expanded as an important tree species of nances for H, DBH, GD, and CW. Heritability structural- and appearance-grade wood prod- estimates were significant except for a few ucts with the combined plantation area in traits from Fujian’s provenances. Heritability Guangdong and Guangxi has reached approxi- ranged from 0.20 to 0.57 for H, 0.19 to 0.38 for DBH, 0.21 to 0.55 for GD, and 0.09 to 0.39 for mately 5000 ha (ZHU et al., 2005). A comprehen- CW. Heritability estimates for H and DBH sive provenance-progeny tests was initiated in decreased with increased age for each province the 1990s with the aim of further selective and combined three provinces. Significantly breeding (ZHU et al., 2005; HUANG et al., 2012). high genetic correlations were observed for age- Breeding of for commercially impor- age and trait-trait correlations, indicating that C. hystrix tant traits is a relatively recent development for genetic performance at one trait was well corre- lated with another trait. In total, 22 families increasing the productivity of newly established and 60 individuals were selected for backward plantations. The program is following the prin- and forward selection based on breeding values. ciple of two successful long-term tree breeding programs: radiata and loblolly pine. Genetic Keywords: Castanopsis hystrix; additive variance; improvement of radiata pine [Pinus radiata D. individual heritability; genetic correlation; breeding Don] was initiated in the 1950s in Australia values. with the initial plus tree selection for open-pol- linated (OP) family testing. The first two gener- ations of radiata pine selective breeding focused Introduction on growth and form traits; the average rotation Forest tree species typically have long rota- length for radiata pine plantations decreased tions and long breeding cycles, indicating that from about 40–55 years to about 30–35 years tree improvement programs need to be well (LI and WU, 2005; WU et al., 2007, 2008). The organized and structured at the start to maxi- third generation of radiata pine breeding in mize genetic gain for short and long-term Australia focused on genetic variation in wood quality traits (MATHESON et al., 2008; GAPARE et al., 2009). Genetic improvement of loblolly pine was also initiated in the 1950s, and the North 1) Guangdong Provincial Key Laboratory of Bio-control Carolina State University Cooperative Tree for the Forest Disease and Pest, Guangdong Academy of Forestry, Guangzhou, 510520, P. R. China. Improvement Program (NCSU) with initial *) Corresponding author: WEN PAN. Tel/Fax: 86-20- provenance and progeny trials and achieved 87033558. E-mail: [email protected] genetic gains up to 30% using third-generation Silvae Genetica 64, 3 (2015) 81 DOI:10.1515/sg-2015-0007 edited by Thünen Institute of Forest Genetics Zhang et. al.·Silvae Genetica (2015) 64-3, 81-90 selectively bred loblolly pine when compared to the traits; and (4) estimate family and individ- unimproved genetic material (MCKEAND et al., ual-tree breeding values based on combined 2001, 2003). three provinces to select elite trees for grafting into seed orchards. Genetic variance among individuals is typi- cally partitioned into non additive and additive components. Non additive variance can be fur- ther divided into epistatic variance due to inter- Materials and Methods actions of genes at two or more loci and Genetic material dominance variance caused by intra-locus gene interactions (FALCONER and MACKAY, 1996). Genetic materials were collected from three However, genetic parameter estimates based on provinces as: OP seed orchards during forest tree breeding Guangdong provenance: Sixty-four OP fami- have largely focused on additive genetic models lies were collected from three provenances of in the form of family or individual tree models. Xinyi, Gaozhou, and Luhe in October and In C. hystrix, no reports are available on the use November 2001. Guangxi provenance: OP seeds of OP progenies for describing the structure of from 51 plus trees were collected from five the genetic variance. In addition, no estimates provenances of Pubei, Bobai, Rongxian, Pingxi- have been published of the breeding values of ang, and Donglan in October and November C. hystrix, although predictions of breeding val- 2000. ues for other forest tree species are a standard Fujian provenance: OP seeds from 42 plus prectices (XIE et al., 2007; ISIK et al., 2008; trees were collected from three provenances of IWATA et al., 2011; HARDNER et al., 2012; DENIS Zhangzhou, Jinshan, and Huaan in October and and BOUVET, 2013). At this time, studies on the November 2001. Details on seed source collec- genetics of have focused on height C. hystrix tion information for the 11 provenances at the (H), diameter at breast height (DBH), ground three provinces are provided in . diameter (GD), and crown width (CW); these Table 1 related traits are the most important for wood production at initial breeding effort (GREAVES et Provenance-progeny trial and measurement al., 1997). The 11 provenances from the three provinces The objectives of this study were to (1) com- were planted in 2002–2003 at Longyandong pare genetic variation at the additive levels in Forest Farm in Guanzhou City, Guangdong DBH, H, GD, and CW among the three Province, China (23°10’–23°18’N, 113°21’– provinces and combined three provinces; (2) 113°27’E). The site has an arenaceous shale lat- estimate heritability for growth and form traits eritic red soil (pH 4.75–5.20) and receives an at different ages using the family model for average annual rainfall of 1,760 mm (Table 1). each province and combined three provinces; (3) Location of natural Castanopsis hystrix prove- estimate additive genetic correlations between nances in China is described in Fig. 1. 3–5 trees Table 1. – Details of Castanopsis hystrix A.DC. provenance location. 82 DOI:10.1515/sg-2015-0007 edited by Thünen Institute of Forest Genetics Zhang et. al.·Silvae Genetica (2015) 64-3, 81-90 Figure 1. – Location of natural Castanopsis hystrix provenances in China: Xinyi, Gaozhou, Luhe, Pubei, Bobai, Rongxian, Pingxiang, Donglan, Jinshan, Gaoche, and Huaan. for family plot Seedlings were planted using y = Xb + Zu + e, (1) 2 3 m spacing, with five, six, and seven repli- ϫ where y is an n 1 vector of observations in the cations for populations from Guangdong, ϫ Guangxi and Fujian, respectively, using ran- trial. For combined province analysis, the b is a domized complete block designs. H and DBH vector of fixed effects (including the trial mean, were measured during 2005–2010 for all fami- replication, province and year effect, respec- lies of the three provinces (e.g. age 3 to 9 based tively), X is an nϫw design matrix that mapped the observations onto the fixed effects, on populations and traits see detail from Table is an n m matrix that mapped the observa- 3). DBH was measured at 1.3 m above the Z ϫ ground. GD of the three provinces was investi- tions onto the genotype effects (including prove- gated and CW was the average of east-west and nance, family within provenance), u is a pϫ1 north-south directions in 2005, which were vector of random effects (including family- important for wood production. within-provenance, province by replication, and family-within-provenance by replication inter- action), and e is an nϫ1 vector of unknown Statistical analyses residuals. For each individual province analy- Preliminary family mixed model analyses ses, the b is a vector of fixed effects (including were performed using the statistical software trial mean, replication), and u is vector of ran- ASReml (GILMOUR et al., 2009) to calculate dif- dom effects (including family across prove- ferent variance components. Provenance effects nance, and family by replication). were not significant and were dropped in the Individual-tree narrow-sense heritability was final analyses. The data were analyzed using estimated for each province and combined three univariate and multivariate family mixed mod- provinces, as follows: els. ASReml fits the general mixed model, and was used for combined and separate analyses for three provinces. (2) 83 DOI:10.1515/sg-2015-0007 edited by Thünen Institute of Forest Genetics Zhang et. al.·Silvae Genetica (2015) 64-3, 81-90 where 2 is the family variance component ing that = , and and are the square ␴ Fam ij im hj hm within provenance, 2 is the interaction root of heritability estimates for the juvenile ␴ Fam.Rep variance family by replication, and 2 is the age and the mature age, respectively; and is ␴ e rg residual variance component.