Table 2.
Summary of empirical studies of genomic selection in perennial tree species
| Species | Population type and size | Markers | Traits | Methods for building prediction models | Prediction accuracy | References |
|---|---|---|---|---|---|---|
| Apple (Malus × domestica) | 1120 individuals (seedlings generated from a factorial mating design of four females and two male parents) | 2500 SNPs | Six traits (fruit firmness, soluble solids, russet coverage, weighted cortex intensity, astringency, titratable acidity) | rrBLUP, Bayesian LASSO | 0.68 to 0.89 | Kumar et al. (2012b) |
| Japanese pear (Pyrus pyrifolia) | 76 Japanese pear cultivars | 162 markers (155 SSRs, 4 RAPD-STS, 2 ACC synthase genes, 1 S-RNase gene) | Nine traits (harvest time, resistance to black spot, firmness of flesh, fruit size, fruit shape in longitudinal section, acid content, total soluble solid content, number of spurs, vigor of tree) | BayesA, BayesB | −0.45 to 0.75 | Iwata et al. (2013a) |
| Oil palm (Elaeis guineensis) | 262 individuals (two parental populations involved in reciprocal recurrent selection with 131 individuals each) | 265 SSR markers | Eight traits (bunch number, average bunch weight, fruit-to-bunch, pulp-to-fruit, kernel-to-fruit, oil-to-pulp ratios, number of fruits per bunch, average fruit weight) | GBLUP, Bayesian LASSO, Bayesian RR, BayesCπ, BayesDπ | −0.41 to 0.94 | Cros et al. (2015) |
| Loblolly pine (Pinus taeda) | ca. 800 individuals (seedlings derived by crossing 32 parents in a circular mating design) | 4825 SNPs | Two traits at multiple ages (3, 4, and 6 years old for diameter at breast height and 1, 2, 3, 4, and 6 years old for height) | rrBLUP | 0.16 to 0.75 (0.63 to 0.75 at 6 years old) | M.F.R. Resende et al. (2012a) |
| Loblolly pine (P. taeda) | 951 individuals (seedlings derived by crossing 32 parents in a circular mating design) | 4853 SNPs | 17 traits (four traits related to growth, six traits related to development, two traits related to disease resistance, five traits related to wood quality) | rrBLUP, BayesA, BayesCπ, Bayesian LASSO, rrBLUP B | 0.17 to 0.51 | M.D. Resende et al. (2012) |
| Loblolly pine (P. taeda) | 149 clones (full-sib progeny derived from 13 crosses) | 3406 SNPs | Four traits (height, volume, lignin and cellulose contents) | rrBLUP including both additive and dominance effects | 0.30 to 0.83 when 10% of the clones were selected within each cross; 0.56 to 0.68 when 10% of the clones were selected randomly across all crosses | Zapata-Valenzuela et al. (2012) |
| Loblolly pine (P. taeda) | 165 clones (full-sib progeny from nine crosses) | 3461 SNPs | Two traits (height, volume) | GBLUP | 0.55 to 0.74 when 10% were selected within each cross | Zapata-Valenzuela et al. (2013) |
| Eucalyptus (E. grandis, E. urophylla, E. globulus and their F1 hybrids) | 820 individuals for one population and 920 for the other | 3564 SNPs for one population and 3129 for the other | Four traits (tree circumference, height growth, wood specific gravity, pulp yield) | rrBLUP | 0.55 to 0.88 | M.D. Resende et al. (2012) |
| White spruce (Picea glauca) | 1694 individuals | 6385 SNPs | 12 traits (cell population, fiber coarseness, crystallite width, wood density, microfibril angle, wood stiffness, ring width, specific fiber surface, cell radial diameter, cell tangential diameter, cell wall thickness, 22-year height) | rrBLUP and rrBLUP with pedigree effects | 0.33 to 0.44 when both training and testing data sets share individuals of the same families; 0.13 to 0.28 when training and testing data sets are made up of individuals of different families; −0.05 to 0.13 when families making up the validation data sets are from populations that are not represented in the training data sets | Beaulieu et al. (2014a) |
| White spruce (P. glauca) | 1748 individuals | 6932 SNPs | Four traits (average wood density, average microfibril angle, 17-year height, 17-year diameter at breast height) | Bayesian RR, Bayesian LASSO, and those with pedigree effects | 0.52 to 0.79 with Bayesian RR when the validation sets were built with individuals within full-sib families and both training and testing sets were from the same breeding group; 0.29 to 0.59 with Bayesian RR when the validation sets were built with full-sib families and both training and testing sets were from the same breeding group | Beaulieu et al. (2014b) |
SNP, single nucleotide polymorphism; RR, ridge regression; rrBLUP, random regression best linear unbiased prediction; LASSO, least absolute shrinkage and selection operator; SSR, simple sequence repeat; RAPD-STS, random amplified polymorphic DNA-sequence tagged sites; ACC, 1- aminocyclopropane-1-carboxylate; GBLUP, genomic best linear unbiased prediction.