In this study, we evaluated the potential of ranking and selec-tion for better solid-wood quality traits of outstanding pheno-types (plus trees) as an alternative to backward selection based on BV estimates on half-sib progenies. The evaluation was based on multiple genetic parameters: correlation between half-sib prog-eny BVs and plus-tree phenotype data, repeatability, and narrow-sense heritability (h2) based on parent–offspring regression as compared with half-sib correlation.
Fig. 2. Heritability estimation by parent–offspring regression based on different numbers of progenies for Pilodyn penetration depth, Hitman acoustic velocity, and MOE(ind). The number of ramets per mother clone varied among plus trees from one to three. [Colour online.]
0.3 0.4 0.5 0.6
2 4 6 8 10 12
Number of progenies
Heritability
Trait MOE(ind) Pilodyn Velocity
814 Can. J. For. Res. Vol. 49, 2019
Published by NRC Research Press
Fig. 4. Heritability estimates using parent–offspring regression of area-weighted values calculated from SilviScan data for each annual ring.
[Colour online.]
0.0 0.2 0.4 0.6
4 8 12 16
Ring
Heritability
Density MFA MOE
Fig. 3. Correlations of SilviScan data for each annual ring between breeding values (BVs) of plus trees estimated from the progeny and area-weighted phenotypic values from the plus trees. [Colour online.]
0.0 0.2 0.4 0.6
4 8 12 16
Ring
Correlation
Density MFA MOE
Zhou et al. 815
Published by NRC Research Press
The h2estimates for wood density, MFA, and MOE measured with SilviScan from increment cores were 0.43, 0.29, and 0.38 and 0.35, 0.15, and 0.28 based on progeny correlation and parent–
offspring regression, respectively. When using indirect measure-ments directly on standing trees, the h2estimates based on progeny correlation were 0.31, 0.20, and 0.28 for Pilodyn, velocity, and MOE(ind), respectively. Moreover, based on parent–offspring regression, the values ranged from 0.27 to 0.41, 0.13 to 0.29, and 0.13 to 0.30 for Pilodyn, velocity, and MOE(ind), respectively. Our h2values estimated by progeny correlation were in the range of those previously reported for wood properties in loblolly pine (Pinus taeda L.;Isik et al. 2011), maritime pine (Pinus pinaster Aiton;
Louzada 2003;Gaspar et al. 2008), lodgepole pine (Pinus contorta Douglas ex Loudon;Hayatgheibi et al. 2017), Norway spruce (Hylen 1997,1999;Hannrup et al. 2004;Hallingbäck et al. 2008), white spruce (Lenz et al. 2010), and British Columbia’s interior spruce (Ivkovich et al. 2002). Similarly, our h2estimates based on parent–
offspring regression also agree with previously reported values for wood properties in Norway spruce (Steffenrem et al. 2016), loblolly pine (Loo et al. 1984;Williams and Megraw 1994), and Sakhalin spruce (Picea glehnii (F. Schmidt) Mast.;Tanabe et al. 2015).
Our repeatability estimates for the indirect measurements based on the analysis of three ramets per plus tree were 0.52, 0.30, and 0.45 for Pilodyn, velocity, and MOE(ind), respectively. Previ-ously reported repeatability estimates for wood quality and growth in Norway spruce (Rosner et al. 2007;Gräns et al. 2009;
Steffenrem et al. 2016) and Sakhalin spruce (Tanabe et al. 2015) are also in concordance with our estimates, whereas other studies have reported either higher MFA and MOE values in radiata pine (Pinus radiata D. Don;Lindström et al. 1998) or lower MOE values in Sitka spruce (Picea sitchensis (Bong.) Carrière;Hansen and Roulund 1997).
Interpretation of the discrepancies between progeny and plus tree for ring width and wood properties
The observed discrepancies in developments across annual rings between the trees at the progeny trials and those at the clonal archive for ring width, wood density, and MOE could be attributed to a difference in spacing, including thinning of the clonal archive. During the first years, the trees of the clonal ar-chive were only 0.5 m apart and were under strong competition compared with the trees in the progeny trials. This is presumed to explain their thinner annual rings and higher wood density at these ages. The thinning performed at two occasions even out this difference in competition, and widths and densities become sim-ilar. Thinning results in more favourable growth conditions for the clonal archive trees regarding access to light and other re-sources, which is presumed to explain why trees at these ages instead have wider annual rings and lower wood densities. After topping of the trees, it is harder to relate this to the developments of growth and patterns.
Less spacing among trees is known to result in stronger compe-tition for resources. Under tight spacing, lower diameter is pri-marily the result of competition for light (Turner et al. 2009).
Trees tend to grow taller at the expense of diameter in their
at-tempt to outcompete the neighbouring trees in search of light.
Multiple studies in conifer species have reported effects of plan-tation density on growth (diameter and slenderness) and wood and fiber properties. Wider spacing at planting has been reported to be associated with higher tree diameter and lower MOE in Scots pine (Persson et al. 1995) and a number of coniferous species (Chuang and Wang 2001;Zhang et al. 2002;Clark et al. 2008;
Lasserre et al. 2008,2009;Schimleck et al. 2018). Ring width and wood density are negatively correlated, and MOE is negatively correlated with both wood density and MFA (Loo et al. 1984;Hodge and Purnell 1993;Zhang and Morgenstern 1995;Waghorn et al.
2007;Gaspar et al. 2008;Lasserre et al. 2009;Chen et al. 2014). The effect of spacing on growth and wood properties, together with their well-documented correlations, strengthens our previous in-terpretation regarding thinner rings and higher density for the clonal archive trees in the first rings and the reverse later on. It also supports our interpretation of the higher MOE, and lower MFA, at these latter ages for the progeny trees.
Although narrow spacing could account for the results that we obtained, it is also possible that additional factors have contrib-uted to the discrepancies between the two plantation types: abi-otic factors such as rainfall, temperature, or soil properties.
However, a previous study conducted on the same data from the progeny trials, both treated with similar silvicultural activities, revealed low genotype–environment interaction (Chen et al.
2014), which indicates that climatic conditions or soil properties are not factors behind the differences (at least in southern Swe-den, where all three plantations are located).
Potential for selection of Norway spruce plus trees on phenotype data at clonal archives
In operational breeding, selection of plus trees as gene donors to the next generation is usually conducted through evaluation of their OP progenies grown in common-garden experiments (prog-eny trials), a breeding design known as a backward selection. This is a method that involves multiple actions such as seedling pro-duction, seedling establishment (often in multiple sites), and as-sessment and evaluation of multiple tree properties when the trees in the trial have grown at least 10 rings at breast height. The high demands in time and costs of this approach motivate evalu-ation of alternatives such as plus-tree selection based on pheno-type data assessed at the clonal archive.
Phenotypic selection of plus trees is a common practice for establishing the foundations of a breeding program, while provid-ing early genetic gains (Zobel and Talbert 1984). Furthermore, two-stage selection strategies of plus trees, in which plus trees are first selected based on phenotype followed by a second stage based on clonal or progeny test, have previously been proposed in conifers (Danusevicius and Lindgren 2002,2004). Danusevicius and Lindgren concluded that when heritability is high, pheno-typic selection is a superior breeding strategy and a two-stage strategy based on progeny testing improves by the first stage of phenotypic selection.
Considering that repeatability and h2estimates are similar, we suggest that selection of MFA at the clonal archive would be an Table 1. Heritability and repeatability estimates based on measurements of wood density, MFA, and MOE from SilviScan, as well as Pilodyn penetration depth and Hitman acoustic velocity, which were used to estimate MOE as the indirect methods.
Methods
SilviScan Indirect methods
Density MFA MOE Pilodyn Velocity MOE(ind)
Parent–offspring regression (1 ramet) 0.35 (±0.16) 0.15 (±0.16) 0.28 (±0.16) 0.27 (±0.15) 0.13 (±0.15) 0.13 (±0.15) Parent–offspring regression (3 ramets) N/A N/A N/A 0.41 (±0.15) 0.29 (±0.15) 0.30 (±0.15) Half-sib correlation (offspring only) 0.43 (±0.09) 0.29 (±0.08) 0.38 (±0.08) 0.31 (±0.08) 0.20 (±0.08) 0.28 (±0.08)
Repeatability N/A N/A N/A 0.52 (±0.06) 0.30 (±0.04) 0.45 (±0.05)
Note: To allow comparison, all heritability estimates (± standard error) were based only on the 162 families for which three ramets were available in the clonal archive. Estimates that are statistically significantly different from zero are indicated in boldface type. MFA, microfibril angle; MOE, modulus of elasticity; MOE(ind), modulus of elasticity for indirect methods; N/A, not applicable.
816 Can. J. For. Res. Vol. 49, 2019
Published by NRC Research Press
effective alternative. However, given the low values of correlation among plantations, h2, and repeatability, a lower efficiency in tree improvement is expected for MFA than for other traits with higher h2(e.g., density;Chen et al. 2014). This conclusion can be extended to selection based on both progeny and plus-tree phe-notype. The heritability of MOE using three ramets based on parent–offspring regression (0.30) is higher than using half-correlation (0.28); however, considering that clonal repeatability for MOE (0.45) is higher than any h2estimate, we suggest that it would be more cost- and time-effective to select clonal archive trees based on MOE scored with indirect measurements. Previ-ously, MFA and MOE have been reported to have low and moder-ate heritabilities, respectively, in Norway spruce (Hannrup et al.
2004;Lenz et al. 2010;Chen et al. 2014), whereas higher heritabil-ities have been reported for MOE in Scots pine (Hong et al. 2015) and for MOE and MFA in loblolly pine (Isik et al. 2011). Similar to the other wood properties (repeatability for wood density is higher than correlation and h2), selection of trees at the clonal archive based on indirect measurements of this trait will be effi-cient. Considering that h2increases towards the bark, a higher response to selection is expected at older ages. Other studies also support our observation of higher heritability for wood density than for MFA and MOE (Lenz et al. 2010;Isik et al. 2011;Chen et al.
2014).