Genetic Variation And SSR Association Analyses In Growth And Wood Properties In Eucalyptus Cloeziana

Eucalyptus cloeziana is one of the most promising Eucalyptus species for high-value solid wood.Information on genetic variability and genomic loci associated with growth and wood properties is important for understanding the genetic basis of important traits and exploring the genomic resources usable for molecular breeding.Based on a progeny trial of 115 open-pollinated families of E.cloeziana,this dissertation established the near infrared（NIR）spectrum models for predicting wood mechanical properties,estimated the genetic parameters of growth and wood traits,selected out superior provenances and families,and identified simple sequence repeats（SSR）markers significantly associated with growth and wood properties.The main conclusions are as follows.（1）The NIR models for predicting wood mechanical properties in E.cloeziana had high reliability.A total of 75-109 trees were harvested for standard measurements,and NIR models for predicting basic density（BD）,green density（GD）,modulus of elasticity in static bending（MOE）,modulus of rupture（MOR）and compressive strength parallel to grain（σc）were established.The NIR cross-validation models were established using standard measurements of 85% of samples（64-93 trees）.The NIR models for BD,GD,MOE,MOR and cs showed 0.03,0.02,1.09,6.39 and 2.30 in root-mean-squares error of cross-validation（RMSECV）,0.78,0.77,0.81,0.78 and 0.80 in coefficient of determination（R²）,and 2.16,2.06,2.26,2.13 and 2.26 in residual predictive deviation（RPD）,respectively.The NIR models were further validated using the rest 15% samples（11-16 trees）,with BD,GD,MOE,MOR and cs showing 0.74,0.69,0.81,0.63 and 0.65 in coefficient of determination of prediction（2pR）and 0.060,0.041,1.100,12.500 and 4.790 in root mean squares error of prediction（RMSEP）,respectively.These parameters indicate the reliability of the NIR models established herein.（2）The genetic parameters of growth and wood mechanical properties including heritability,trait-trait correlation and correlation change with age in growth were estimated.The growth traits included diameter at breast height（D）,height（H）and individual volume（V）,and the wood properties included BD、GD、MOE、MOR and σc.Growth traits D and V as well as all wood traits showed significant difference between provenances and between families within provenance.Narrow-sense heritability ranged from 0.04 to 0.35 for growth across ages of 0.5-9.5 years and from 0.06 to 0.24 for wood mechanical properties at age 9.5,indicating low to moderate magnitude of additive genetic control of these traits.Phenotypic（rp）and additive genetic（rg）correlations at the final age of 9.5 years were consistently significantly positive between growth traits and mostly significantly positive between wood mechanical traits,while rp and rg were small and adversely significant,respectively,between the two types of traits,posing a challenge for concomitant improvement of growth and wood traits.In addition,rp and rg coefficients between age 9.5 and earlier years for each of the growth traits H,D and V were all positively significant(except for H_2.5)and had a generally increasing trend with age,suggesting the possibility of early selection.（3）A number of provenances and families superior in growth and wood properties were selected out using composite index selection against multiple traits,including H_9.5,D_9.5,V_9.5,BD,GD,σ_c,MOE and MOR.Both equations of I3 and I7 were preferred for multiple-trait index selection in terms of index selection heritability,comprehensive breeding value and index selection efficiency.Finally the coastal provenances 20725,20724 and 20722 and the families 65,56 and 51 ranked the top three in terms of composite index of provenance and family,respectively.（4）A total of 26 SSR markers were detected to be significantly associated with growth and wood mechanical properties.Based on 839 SSR markers spanning the genome of E.grandis,47 SSR markers were screened out against the high-value（at least 8 samples per traits）and low-value（at least 8 samples per trait）genotype pools for traits D_9.5,BD,MOE,MOR and sc（totaling at 32 samples）.The 47 SSRs were tested with 348 samples representing 115 open-pollinated families（2-4 trees per family）using both general and mixed linear models.The total numbers of significantly associated markers with D_9.5,BD,MOE,MOR and cswere 8,7,6,4 and 9（P﹤0.05）,respectively.The phenotypic variation explained per significant marker ranged from 5.7% to 44.1%,with 6 markers associated with two traits.The mean values of allelic plus effect and minus effect were 6.3% and 3.5%,respectively.