Tree Genetics & Genomes (2012) 8:303–311 DOI 10.1007/s11295-011-0441-z

ORIGINAL PAPER

Genetic effects on wood quality traits of plantation-grown white spruce (Picea glauca) and their relationships with growth Yill-Sung Park & Yuhui Weng & Shawn D. Mansfield

Received: 18 January 2011 / Revised: 14 September 2011 / Accepted: 10 October 2011 / Published online: 19 November 2011 # Her Majesty the Queen in Right of Canada 2011

Abstract Clonal repeatabilities on individual tree (Hi2 ) and clonal mean (HC2 ) bases for growth (14-year height and volume), wood quality traits (latewood proportion, wood density, fiber length, and microfibril angle), and genotypic correlations among the traits were estimated, using 30 white spruce (Picea glauca [Moench] Voss) clones from six fullsib families (five per family). These families were selected from a clonally replicated test to represent different early growth categories: fast, moderate, and slow. Wood increment cores of the 30 clones were collected from two contrasting sites at age 19 years. For growth traits, in contrast to most wood quality traits, more genetic variation was accounted for by clone within family than by family within growth category. Both growth and wood quality traits appear to be under moderate genetic control, with b 2 ¼ 0:20  0:36 and H b 2 ¼ 0:70  0:83. The only excepH i C b 2 ¼ 0:10 and H b 2 ¼ 0:34). tion was microfibril angle (H i

C

Generally, faster growth resulted in a significantly lower Communicated by R. Burdon Y.-S. Park (*) Natural Resources Canada, Canadian Forest Service—Canadian Wood Fibre Centre, PO Box 4000, Fredericton, NB E3B 5P7, Canada e-mail: [email protected] Y. Weng Tree Improvement Office, New Brunswick Department of Natural Resources, 3732 Route 102, Island View, NB E3E 1G3, Canada S. D. Mansfield Canada Research Chair of Wood and Fibre Quality, Department of Wood Science, University of British Columbia, 4030-2424 Main Mall, Vancouver, BC V6T 1Z4, Canada

latewood proportion and lower overall wood density. Selection for faster growth does not appear to impact on either fiber length or microfibril angle. Among the wood quality traits, significant genotypic association was observed only between latewood proportion and wood density. Despite the generally negative association between growth and wood density among families, several fastgrowing clones maintained above-average density. This implies that, by adopting multiclonal forestry, one can simultaneously improve growth and wood density. Keywords Breeding strategy . Deployment strategy . Genetic parameters . Tree improvement . Wood properties

Introduction White spruce (Picea glauca [Moench] Voss) is commercially important as a reforestation species in New Brunswick (NB), Canada, where about 15 million seedlings are planted annually. For this species, the manufacturing of wood fiber and lumber products is the most important use; therefore, wood properties can influence the end use of the timber. Longterm breeding programs for the species have been pursued in NB for the past 30 years, primarily targeting growth rate and adaptability (Weng et al. 2010). However, there is evidence that vigorous growth may adversely affect wood properties of spruce (Rozenberg and Cahalan 1997). Despite its economic importance, genetic studies on wood quality traits in white spruce have been limited. The few reported studies have generally focused on genetic variation among populations and/or half-sib families and have mainly examined wood density (Micko et al. 1982; Taylor et al. 1982; Corriveau et al. 1987, 1991; Yanchuk and Kiss 1993). The general findings based on previous

304

studies were: (1) wood density likely declines with increasing growth rate and (2) wood density was under strong additive genetic control. Although wood density represents a complex trait influenced by many interacting components, it has been proposed that the proportion of latewood is one of the most important factors (Louzada and Fonseca 2002). As earlywood and latewood have different densities, changes in the proportion of either will inevitability lead to changes in the overall wood density. For white spruce, information regarding how genetic selection affects latewood proportion and wood density, particularly at the full-sib and clone levels, is scarce. This shortcoming is of particular concern, as current reforestation strategies are gradually turning to the deployment of elite full-sib families and clones. Along with wood density, microfibril angle and wood fiber length are key traits that influence the wood products. Fiber length is generally acknowledged to affect pulp and paper quality, whereas microfibril angle has a substantial effect on both mechanical behavior and dimensional stability of wood, and as such, is an important quality trait for sawn timber. Information on the inheritance of these two traits in spruce and their relationships with growth rate is therefore required for developing breeding and deployment strategies. Currently, there is no available knowledge for either trait in white spruce. Genetic studies on both traits in other spruce (Hannrup et al. 2004) and pine (Pot et al. 2002; Eriksson and Fries 2004) species have demonstrated that improved growth yielded longer fibers and that both traits were under moderate genetic control. Using data collected from a 19-year-old, clonally replicated, genetic test of white spruce planted in NB, the objectives of this study were: (1) to calculate clonal repeatability of growth and wood quality traits and (2) to investigate the relationship between growth and wood quality traits. The wood traits investigated in this study included latewood proportion, wood density, microfibril angle, and fiber length.

Materials and methods Materials and measurement of wood quality traits As part of a white spruce breeding program, a clonally replicated full-sib progeny test was established at three sites [the Acadia Research Forest (ARF) near Fredericton, Sussex, and St. Quentin] in NB in 1990. At each site, 340 clones, from 75 full-sib families, were planted. The full-sib families were produced using a disconnected diallel mating design on 20 unrelated plus trees, which were phenotypically selected from natural stands based on height growth and tree form. The test at each site was established using a randomized complete block design with 16 blocks of single-tree plots. The 14-year height (HT14) and diameter at breast height were measured in 2005,

Tree Genetics & Genomes (2012) 8:303–311

and individual-tree volume (VOL14) was calculated using the volume equation of Honer et al. (1983). Details of producing the full-sib families and clones, development of clones by vegetative propagation, and genetic analysis for the 14-year growth traits were described by Weng et al. (2008). Wood samples were collected from two sites: ARF and Sussex in 2009 when the trees were 18 years old after planting. ARF (latitude 46.00° N, longitude 66.31° W, and elevation 80 m) represents a site of poor quality with a site index at age 50 (SI50) of 14 m; in contrast, Sussex (45.97° N, 65.43° W, and 165 m) has an SI50 of 21 m. Wood increment cores (10 mm in diameter) from 30 clones derived from six full-sib families (five clones per family) were collected manually. In order to sample genotypes across the growth range, the six unrelated families were selected based on previous height data to represent different growth categories: two each from fast- (families 223 and 242), moderate- (145 and 435), and slow-growing families (342 and 125). On average, four ramets per clone were randomly selected from the first 11 blocks of each site, resulting in 121 and 123 sampled trees in total from ARF and Sussex, respectively. The 10-mm increment cores were collected at breast height, from bark to pith on the north face for each tree. The increment cores were initially Soxhlet extracted overnight with acetone and allowed to dry. The samples were then precision cut to 1.68 mm thickness with a twin-blade pneumatic saw and allowed to equilibrate to 7% moisture before performing density analysis. All samples were then scanned from pith to bark by X-ray densitometry (Quintek Measurement Systems, TN) at a resolution of 0.0254 mm, and the data reported as relative wood density (WD) on an oven-dry weight basis. Intra-ring wood density information was also obtained for each core sample using X-ray densitometry. On average, there were 12 annual rings identified, and these were numbered by year. Ring width, latewood width, earlywood density, and latewood density were recorded for each of the rings. Demarcation between early- and latewood was done by calculating transition density between highs in latewood and lows in earlywood, which was around 325 kg/m3. Latewood percentage (LWP) was then calculated as the percentage ratio of the sum of latewood width to overall ring width over 12 annual rings. Fiber length (FL) was assessed, pith to bark, for all trees. Increment core samples were macerated in a solution of 1:1 30% hydrogen peroxide to glacial acetic acid for 48 h at 70°C as previously described by Ukrainetz et al. (2008), and length-weighted fiber lengths were determined on a Fiber Quality Analyzer (OpTest Equipment Inc., Hawkesbury, ON, Canada). Microfibril angle (MFA) was measured on the latewood portion of the growth ring corresponding to year 14 by X-ray diffraction (Megraw et al. 1998). The sample precision-cut sections used for densitometry were screened for 002

Tree Genetics & Genomes (2012) 8:303–311

305

diffraction arc T values using a Bruker D8 Discover X-ray diffraction unit equipped with an area array detector (GADDS) on the radial face of the latewood portion of individual growth rings, as previously described by Ukrainetz et al. (2008).

genotypic correlation (rG) between traits were calculated using the following formula (Weng et al. 2008):

Statistical analysis

d x,y is the estimated covariance between traits x where Cov b2x and s b2y are the estimated and y for all genetic effects, and s genetic variances for traits x and y, respectively. The probability of z score was used to examine the significance of estimates of variance component, clonal repeatability, and genetic correlation (Akesson et al. 2008). Note that here and elsewhere in the text, except where otherwise indicated, the term “significant” refers to P<0.05.

The data were analyzed using the Proc Mixed procedure in the Statistical Analysis System (SAS; Littell et al. 2006) for each trait, using the following model: y ¼ Xb þ Z1 s þ Z2 c þ Z3 t1 þ Z4 t2 þ e where y is a vector of a trait measurements, b is a vector of fixed effects (overall mean, site, and growth category), s is a vector of random family within category effects (including both general combining ability and specific combining ability effects), c is a vector of random clones within family and category effects, t1 and t2 correspond to vectors of random interactions between site and effects of s and c, respectively; and e is a vector of random residual errors. X is a known incidence matrix relating to observations in y to the fixed effects in b, and Z1–Z4 are known incidence matrices relating to observations in y to random effects s, c, t1, and t2, respectively. In order to handle heterogeneity between sites, the model was fitted by including the repeated statement with the group option. Variances associated with the random effects s, c, t1, t2, and e are referred to as s 2s , s 2c , s 2ls , s 2lc , and s 2e , respectively, these summing to the total phenotypic variance. Note that s 2e was estimated as the average of group residual variances and, actually, was very similar in size to that obtained without considering site heterogeneity for all traits except HT14. Means were calculated using the Estimate statement and the significance of the difference between the means were tested using the Contrast statement. We estimated phenotypic (CV P ) and genetic coefficients pffiffiffiffi c ¼ V b =X »100, where V is the of variation (CV G ) as CV c b2 and total phenotypic variance for CVP and the sum of s s

c G, respectively, and X is the trait mean value. b2c for CV s b 2) Estimates of clonal repeatability on individual-tree (H i b 2 ) bases were calculated as: and clonal mean (H C b2 ¼ H i

b2 ¼ H C

b2s þ s b2c s b2s þ s b2c þ s b2ls þ s b2lc þ s b2e s b2s s b2s þ s

b2c þs kb s 2 þb s2 b2c þ 1 lck2 e s

where k1 and k2 are the coefficients for s 2lc and s 2c , respectively, in the expected mean squares. Estimates of

d C ovx;y rG ¼ qffiffiffiffiffiffiffiffiffiffiffi b2x s b2y s

Results Effects of sites and growth categories As expected, trees growing at Sussex grew much faster than those at ARF (Table 1). Specifically, the trees were 40% taller (HT14) and had ∼246% greater volume (VOL14) at age 14 at Sussex. Site effects were significant for LWP and WD. Compared with ARF, Sussex had a 29.1% decrease in LWP and a 4.4% reduction in WD. Site also influenced the MFA and FL; trees at AFR had larger MFA (2.5%; averaging 16.6 vs. 16.2) and slightly longer fibers (0.5%; averaging 1.90 vs. 1.89) than those at Sussex (Table 1). These differences, however, were statistically nonsignificant. Among various traits, VOL14 showed the largest c P , followed by LWP, HT14, and MFA, whereas FL and CV WD showed notably lower phenotypic variation. Significant differences among three growth categories across sites were found for HT14, VOL14, LWP, and WD but not for MFA and FL. Moderate growth had significantly c P , in percent) of Table 1 Means and coefficients of variation (CV growth and wood-quality traits by site and across sites ARF

Sussex

Across sites

Traits

Mean

cP CV

Mean

cP CV

Mean

cP CV

HT14 (cm) VOL14 (dm3) LWP (%)

417.8 6.9 21.3

27.4 77.9 45.1

585.0 23.9 15.1

18.4 55.5 44.3

502.1 15.4 18.1

27.7 85.7 48.5

WD (kg/m3) FL (mm) MFA (°)

400.0 1.90 16.6

8.0 11.4 20.6

382.5 1.89 16.2

7.3 11.3 20.3

391.2 1.90 16.4

8.0 11.3 20.5

Note that the sample size was 121 for ARF and 123 for Sussex HT14 14-year height, VOL14 14-year volume, LWP latewood proportion, WD wood density, FL fiber length, MFA microfibril angle

306

Tree Genetics & Genomes (2012) 8:303–311

shorter MFA, while FL was not significant. Within each site, these differences were not always distinct; e.g., at ARF, there was no difference between fast- and moderategrowing families for LWP (Table 2). As expected, faster growth resulted in lower mean LWP and WD values. This was especially true at the good-quality Sussex site, e.g., compared with the “slow-growth” category, the “fast growth” category was −5.7% lower in WD at ARF, but the corresponding value was −7.1% at Sussex, suggesting that genetics and site have additive effects on WD and its components (i.e., LWP). The growth category effects on FL were nonsignificant at Sussex but significant at ARF, where the “moderate growth” had significantly longer fibers than the other two categories. The growth category effects on MFA were nonsignificant at both sites, although the “moderate growth” category had a slightly lower MFA relative to the other two categories. Genetic variation and genetic parameters Table 3 presents the estimates of variance components for all traits examined. Random error was consistently the most important contributor, accounting for 57–76% of the total phenotypic variation for all traits, with the largest being for MFA. Significant variances due to between families within category (s 2s ) were found for LWP, WD, and FL. Also, significant variances due to among clones within family (s 2c ) were found for HT14, VOL14, LWP, and FL. Thus, for b2c was the major contributor of the total growth traits, s b2s was important and always greater variation, whereas s b2c for wood quality traits. No significant genotype-bythan s site interaction effects were observed at both family and clonal levels for all traits except VOL14, which showed a marginally significant effect at family level. Among the c G for VOL14 and LWP were largest, traits studied, the CV

about 27%, followed by HT14, FL, MFA, and WD. The c G for WD was only 3.7%. CV b 2 estimates were moderate for VOL14 (0.20) and The H i b2 HT14 (0.35; Table 4). With the exception of MFA, the H i

of the wood quality traits were comparable to those of the growth traits, ranging from 0.24 to 0.36, with the latter b 2 for MFA was only 0.10, which calculated for LWP. The H i is assumed to be due to relatively strong environmental and b 2 estimates site-by-clone interaction effects. Similarly, the H C were high (0.70–0.83) and comparable for all traits except MFA (0.34). However, despite their magnitude in size, only those for HT14 and VOL14 were significantly different from 0. The rG between growth traits was high, 0.94 (Table 4). The rG among wood quality traits varied greatly; it was significant and positive between LWP and WD (0.97), nonsignificant and positive between FL and WD and between MFA and LWP, and nonsignificant and negative between MFA and WD or FL between FL and LWP. As expected, the rG between growth and wood quality traits varied. LWP was significantly and negatively associated with growth traits, particularly VOL14 (>−0.75). So too was WD, although to a much lesser extent. FL was positively correlated with growth traits, whereas MFA was negatively correlated, but the associations were weak and nonsignificant.

Discussion Site effects on growth and wood quality traits Site index is a measure of site quality relating to overall productivity. As expected, sites of better quality generally produce larger and taller trees, with lower LWP and WD

Table 2 Comparisons in growth and wood-quality traits between three growth categories by each site and across the sites Site

Category

HT14 (cm)

VOL14 (dm3)

LWP (%)

WD (kg/m3)

FL (mm)

MFA (°)

ARF

Fast Moderate Slow Fast Moderate Slow Fast Moderate Slow

462.4a 447.4a 345.5b 658.3a 559.0b 535.3b 562.7a 503.2b 440.4c

9.2a 7.5a 3.9b 33.8a 19.7b 17.8b 21.8a 13.6b 10.9c

17.6a 19.9a 26.5b 11.8a 14.1a 19.4b 14.6a 17.1b 22.9c

390.5a 395.1a 414.1b 368.3a 383.2b 396.3c 379.1a 389.1b 405.2c

1.84a 2.00b 1.86a 1.85a 1.90a 1.93a 1.84a 1.95a 1.89a

17.4a 14.8a 17.6a 16.8a 15.6a 16.1a 17.1a 15.2a 16.8a

Sussex

Across sites

Means within sub-columns with different suffix letters differ significantly at the P < 0.05 threshold (experiment-wise type 1 error rate)

Tree Genetics & Genomes (2012) 8:303–311

307

Table 3 Estimated variance components (VC), their percentages of the total phenotypic variance (in percent), and coefficients of genotypic c G ) for growth and wood quality traits variation (CV

HT14 (cm) 3

VOL14 (dm ) LWP (%) WD (kg/m3) FL (mm) MFA (°)

c G (%) CV

Family within b 2s category s

b2c Clone within family s

b2ls Site×family s

b2lc Site×clone s

b 2e Error s

VC

VC

Percent

VC

Percent

VC

Percent

VC

Percent

7.5 16.5 0.1 3.3 4.0 1.1

0.0ns 1.2ns 3.9ns 23.3ns 0.001ns 1.56*

0.0 1.4 6.0 2.6 1.9 14.2

6140.2 56.4 38.4 615.8 0.032 8.4

57.9 62.6 58.2 69.8 62.9 76.0

Percent

0.0ns 0.0ns

0.0 0.0

3672.1** 17.6**

34.6 19.5

794.8ns 14.9*

16.7** 158.3*

25.3 17.9

6.9* 56.1ns

10.4 6.4

0.1ns 29.1ns

0.011** 0.529ns

21.3 4.8

0.005* 0.431ns

0.002ns 0.12ns

9.9 3.9

The Wald test was used to calculate P value (if estimates are significantly different from zero): 0.05 and **significant at P < 0.01

(Zhang 1995; Lindström 1996; MacDonald and Hubert 2002, Table 1). As the latewood is the denser part of the annual ring, lower LWP leads to a lower overall WD. Similar effects on WD by site quality have been reported in other tree species (Abdel-Gadir and Krahmer 1993; Wimmer et al. 2008). Microfibril angle was marginally less in trees originating from the better site in this study, which is consistent with early studies on Norway spruce (Picea abies, Herman et al. 1999; Lundgren 2004) and other gymnosperm species (Raymond and Anderson 2005; Monteoliva et al. 2005; Wimmer et al. 2008). Our results also suggest that site effects on fiber length were negligible for white spruce. Differently, willow (Salix spp.) and Eucalyptus globulus growing on poorer sites exhibited significantly shorter fibers (Monteoliva et al. 2005; Wimmer et al. 2008). Limited information is available on the effects of site quality on wood properties of white spruce. As such,

ns

12.1 27.2 26.8 3.7 6.7 6.0

not significant at P > 0.05, *significant at P <

the information reported herein is of interest for seed source or clonal deployment. The Ecological Land Classification system in NB (Zelazny et al. 2007) will provide a useful framework within which the effects of site quality on wood properties can be investigated in the future. The established classification is based on climate and soil type, all of which can influence wood properties through their effects on tree growth and/or form. Genetic variation, heritabilities for wood traits, and their relationships with growth Wood density is the most extensively studied wood quality trait for most commercial tree species, especially in relation to growth rate. For white spruce, significant genetic variation in WD exists among populations and among half-sib families, with narrow-sense heritability being greater than 0.59 (Taylor et al. 1982; Corriveau et al.

b 2 ) and clonal mean (H b 2 ) basis for the growth and wood quality traits and genotypic Table 4 Estimated clonal repeatabilities on individual tree (H i C correlations (rG) between these traits b2 H i

Trait

HT14 (cm) 3

VOL14 (dm ) LWP (%) WD (kg/m3) FL (mm) MFA (°)

b2 H C

0.35 (0.00)

0.83

0.20 0.36 0.24 0.31 0.10

0.70 0.77 0.71 0.77 0.34

(0.01) (0.06) (0.10) (0.07) (0.19)

rG VOL14

LWP

WD

FL

MFA

0.94**

−0.49*

−0.34ns

−0.94**

−0.75** 0.97**

0.30ns 0.31ns −0.13ns 0.18ns

−0.31ns −0.09ns 0.12ns −0.21ns −0.43ns

P values for clonal mean repeatabilities were essentially the same as those on individual tree. The of Wald test was used to calculate P value (if estimates are significantly different from zero): ns Not significant at P > 0.05, * Significant at P < 0.05 and ** Significant at P < 0.01

308

b2 1987, 1991; Yanchuk and Kiss 1993). Our estimate of H i of 0.24, although on the broad-sense basis, is lower than the above-cited narrow-sense heritability estimate. The current value is also lower compared with the broad-sense heritability estimates reported for Picea sitchensis (0.32; Costa e Silva et al. 1998) and P. abies (0.35–0.55; Hannrup et al. 2004). This may be due to our sampling scheme, which employed the selection of families within each growth category, and thus could result in underestimation of variation among families. Generally, genetics influences WD in a similar way as site quality: faster growth leads to lower LWP and, thus, lower overall WD. Earlier studies have confirmed that growth and WD were moderately and negatively correlated for white spruce (Corriveau et al. 1987, 1991) and other spruce species (Rozenberg and Cahalan 1997; Livingston et al. 2004). The rG between WD and growth traits found in this study support the previous conclusions, suggesting that, in general, simultaneous improvement in both traits will be limited for white spruce. Our results further suggest that WD is genetically more strongly associated with volume (rG =−0.75) than with height (rG =−0.34) for white spruce. It is important to note that most of reported rG between WD and growth were obtained from genetic tests of small plots. In black spruce, Weng et al. (2011) found that height growth selection might not substantially and negatively affect wood density in plantation forestry and the predicted reduction in wood density in genetic tests of small plots might be inflated. In Sitka spruce, trees from fast-growing families were found to have significantly less latewood than unimproved control trees and the slow-growing families (Livingston et al. 2004; Cameron et al. 2005). Our results concur with these findings. LWP is substantially influenced by environment and appears to be under a low degree of genetic control b 2 in (Zhang and Jiang 1998; Hylen 1999). We estimated H i this study to be 0.36 for LWP, which implies that it is under moderate genetic control. Although it may relate to species differences, this inconsistency may in part be ascribed to the employed demarcation rule between earlywood and latewood densities. When using a prespecified density level in the demarcation between earlywood and latewood, Hylen (1997) found high heritability estimates. Microfibril angle is an important quality trait for sawn timber. In short, the lower the MFA, the better. Larger MFAs have been shown to cause increased longitudinal shrinkage and decreased transverse shrinkage in sawn lumber during drying operations (Cave and Walker 1994), resulting in a substantially greater proportion of drying degrade than in outerwood. MFA has greater practical influence on stiffness (modulus of elasticity) than on wood strength (modulus of rupture, Cave and Walker 1994). Our

Tree Genetics & Genomes (2012) 8:303–311

results showed that environment had a substantial impact b 2 ¼ 0:10, which is at the on MFA in white spruce, with H i low end of those in P. abies (0.12–0.26; Hannrup et al. 2004). Earlier studies with other spruce species have shown that faster growing trees always had a higher MFA (Pedini 1992; Lundgren 2004). In contrast, however, our results show selection on VOL14 or HT14 had a negligible or slightly negative impact on MFA. Increased FL is generally favorable for pulping, resulting in improved tear–tensile strength of paper. Heritability estimates for FL are scarce for white spruce. Our b 2 ð0:31Þ suggests that FL is under moderate genetic H i control. In support of our results, Khalil (1985) reported a repeatability of 0.33 in black spruce, and Hannrup et al. (2004) estimated a broad-sense heritability of 0.21 in P. abies. The finding that FL showed nonsignificant but positive rG with growth traits in this study accords closely with Zobel and Jett’s (1995) statement: “fast growth and fibre length are not strongly related genetically.” In our study, an average of 12 annual rings was identified. For white spruce, it has been proposed that between ages 12 and 15 is when the transition to “outerwood” occurs (Corriveau et al. 1991). Thus, the wood properties examined here, strictly speaking, apply only to wood with “corewood” characteristics. The relationship between wood traits and growth rate might vary with age. Fast growth has been shown to have a smaller impact on LWP, WD, and MFA as trees age (Zhang 1998; Zhang and Jiang 1998; Cameron et al. 2005). Thus, in this study, the reduction in WD due to faster growth might be overestimated for wood at its rotation age. Furthermore, the negative effect on WD based on selection for volume is likely to be greater than that for HT. It will be important to continue to assess these relationships in white spruce to monitor possible changes with age and growth traits. Growth improvement is expected to reduce the rotation age and thus increase the proportion of corewood. The information will be of interest for short rotation forestry as the amount of corewood will determine the final merchantable volume. Only two families per category were sampled in this study. Although these families represented the growth categories well (Table 2), it is considered a low number and, thus, variance due to families within category must have been estimated with low precision, leading to clonal repeatability with relatively high P values (Table 4). Also, genetic variation across categories was not included in the calculations as categories were considered a fixed effect in our model, and, thus, the clonal repeatabilities reported here may be underestimates. Relationships among wood traits The high rG between LWP and WD is not unexpected and parallels other studies on spruce species (Zhang 1995;

Tree Genetics & Genomes (2012) 8:303–311

Zhang and Morgenstern 1995; MacDonald and Hubert 2002; Bouffier et al. 2009). Thus, in addition to direct selection by WD, breeding for improving WD can indeed be realized by selecting for increased LWP (Zhang and Morgenstern 1995). Such a strategy may be attractive when b 2 and phenotypic considering the relatively higher H i variation for LWP compared with those for the associated WD values. Although variations in FL are not thought to have much of an impact on sawn timber, shorter FL has been associated with higher MFA (Zobel and Jett 1995; MacDonald and Hubert 2002). Our results confirmed such a negative relationship, although it was weak (Table 4). Mixed relationships between FL and MFA have been reported for spruce species with some being strongly negative (Hannrup et al. 2004), whereas others are negligible (Bergander et al. 2002) or even strongly positive (Hannrup et al. 2004; Cameron et al. 2005). Herein, we show that the rG between WD and FL was low but favorable for spruce (Table 4; Hannrup et al. 2004). In our study, the rG between MFA and WD was low and negative, which is comparable to results reported in P. abies (Hannrup et al. 2004). This negative relationship is expected, given that the lower LWP, the larger MFA and lower WD. Overall our results suggest that selection for WD might result in an increase in FL and a reduced MFA. Implications for white spruce breeding and deployment strategies The findings in the present study have several important implications for white spruce breeding and clonal deployb 2 for LWP, WD, and FL suggest ment. First, estimates of H i that moderate genetic responses can be obtained following the strategy of mass selection and subsequent vegetative b 2 for MFA propagation and deployment. However, the low H i suggests that such a strategy might not be so effective for white spruce for this trait. Second, current long-term breeding programs for white spruce have been focusing on improvements in height growth, tree form, and adaptability, whereas wood properties have been generally ignored. An alternative strategy needs to incorporate WD into breeding programs. However, results from this study suggest that it is difficult to simultaneously improve both growth and WD at the family level (Fig. 1a). One option is to improve growth and form while maintaining WD at an acceptable level, e.g., the same level as the unimproved seedlots or an average density. Such an approach is currently practiced in Sitka spruce improvement in Britain (MacDonald and Hubert 2002). However, as shown in Fig. 1b, there are three clones within the fast-growing families (marked as filled triangles) showing at least average density that are of interest to

309

a

b

Fig. 1 Scattergram of wood density against 14-year height by growth category, using a the least-square means of family and b the leastsquare means of clone. Note that the major girdles on the x and y axes represent the population mean of wood density and 14-year height, respectively

breeders. In general, breeders are interested in clones in the upper right quadrant. Our result indicates that fast growth without compromising WD is possible through multiclonal forestry, i.e., the deployment of tested clones in plantation b 2 calculated in this study forestry. The high estimates of H C suggest that such a deployment strategy will be very effective. Finally, selection for growth would have little or no unfavorable impact on FL and MFA. Furthermore, due to the favorable rG of WD with MFA and FL, selection for WD would lead to improvement in MFA and FL concurrently. Thus, it may be concluded that, from the viewpoint of genetics, there is currently no need to measure and breed for these traits in white spruce, particularly when WD is integrated into selection.

Conclusion Results from this study suggest that fast growth owing to better site quality and genetic selection for growth traits, particularly volume growth, might have a negative effect on WD; however, such an effect was negligible on FL and

310

MFA. Therefore, it appears that WD is a trait that needs to be incorporated into future breeding programs. Moderate genetic control for both growth and wood quality traits was found: The main contributor of variation in growth traits was the variation due to clones within family. For wood quality traits, variation due to families was greater than the clonal variation, although the latter was also substantial. Importantly, this study demonstrated that, despite negative correlation between growth and wood quality, clones that break such a correlation may be found through a multiclonal forestry approach aimed at improving growth without compromising WD. Acknowledgments We thank the New Brunswick Tree Improvement Council for providing help for the sample collection. We thank Dale Simpson and Greg Adams for their valuable comments on an earlier version of the manuscript and Caroline Simpson for editing. Valuable comments/suggestions from Dr. R. Burdon and two reviewers are also gratefully acknowledged.

References Abdel-Gadir AY, Krahmer RL (1993) Estimating the age of demarcation of juvenile and mature wood in Douglas-fir. Wood Fiber Sci 25:242–249 Akesson M, Bensch S, Hasselquist D, Tarka M, Hansson B (2008) Estimating heritabilities and genetic correlations: comparing the ‘animal model’ with parent–offspring regression using data from a natural population. PLoS One 3:1–10 Bergander AJ, Brändström J, Daniel G, Salmén L (2002) Fibril angle variability in earlywood of Norway spruce using soft rot cavities and polarisation confocal microscopy. J Wood Sci 48:255–263 Bouffier L, Raffin A, Rozenberg P, Meredieu C, Kremer A (2009) What are the consequences of growth selection on wood density in the French maritime pine breeding programme? Tree Gene Genom 5:11–25 Cameron AD, Lee SJ, Livingston AK, Petty JA (2005) Influence of selective breeding on the development of juvenile wood in Sitka spruce. Can J For Res 35:2951–2960 Cave ID, Walker JFC (1994) Stiffness of wood in fast-grown plantation softwoods: the influence of microfibril angle. For Prod J 44:43–48 Corriveau A, Beaulieu J, Daoust G (1991) Heritability and genetic correlations of wood characters of Upper Ottawa Valley white spruce populations grown in Quebec. For Chron 67:698–705 Corriveau A, Beaulieu J, Mothe F (1987) Wood density of natural white spruce populations in Quebec. Can J For Res 17:675–682 Costa e Silva J, Wellendorf H, Pereira H (1998) Clonal variation in wood quality and growth in young Sitka spruce (Picea sitchensis (Bong.) Carr.): estimation of quantitative genetic parameters and index selection for improved pulpwood. Silvae Genet 47:20–33 Eriksson T, Fries A (2004) Genetic analysis of fibre size in a full-sib Pinus sylvestris L. progeny test. Scand J For Res 19:7–13 Hannrup B, Cahalan C, Chantre G et al (2004) Genetic parameters of growth and woody quality traits in Picea abies. Scand J For Res 19:14–29 Herman M, Dutilleul P, Avella-Shaw T (1999) Growth rate effects on intra-ring and inter-ring trajectories of microfibril angle in Norway spruce (Picea abies). IAWA J 20:3–21

Tree Genetics & Genomes (2012) 8:303–311 Honer TG, Ker MF, Alemdag IS (1983) Metric timber tables for the commercial tree species of central and eastern Canada. Inf Rep M-X-140. Nat Res Can, Can For Serv—Atl For Ctr, Fredericton Hylen G (1997) Genetic variation of wood density and its relationship with growth traits in young Norway spruce. Silvae Genet 46:55– 60 Hylen G (1999) Age trends in genetic parameters of wood density in young Norway spruce. Can J For Res 29:135–143 Khalil MAK (1985) Genetics of wood characters of black spruce (Picea mariana (Mill.) B.S.P.) in Newfoundland, Canada. Silvae Genet 34:221–230 Lindström H (1996) Basic density in Norway spruce. Part III. Development from pith outwards. Wood Fiber Sci 28:391–405 Littell RC, Milliken GA, Stroup WW, Wolfinger RD, Schabenberger O (2006) SAS® for mixed models, 2nd edn. SAS Institute Inc., Cary Livingston AK, Cameron AD, Petty JA, Lee SJ (2004) Effect of growth rate on wood properties of genetically improved Sitka spruce. Forestry 77:325–334 Louzada J, Fonseca FMA (2002) The heritability of wood density components in Pinus pinaster Ait. and the implications for tree breeding. Ann For Sci 59:867–873 Lundgren C (2004) Microfibril angle and density patterns of fertilized and irrigated Norway spruce. Silva Fenn 38:107–117 MacDonald E, Hubert J (2002) A review of the effects of silviculture on timber quality of Sitka spruce. Forestry 75:107–138 Megraw RA, Leaf G, Bremer D (1998) Longitudinal shrinkage and microfibril angle in loblolly pine. In: Butterfield BG (ed) Microfibril angle in wood. University of Canterbury Press, Christchurch, pp 27–61 Micko MM, Wang EIC, Taylor FW, Yanchuk AD (1982) Determination of wood specific gravity in standing white spruce using a pilodyn tester. For Chron 58:178–180 Monteoliva S, Senisterra G, Marlats R (2005) Variation of wood density and fibre length in six willow clones (Salix spp.). IAWA J 26:197–202 Pedini M (1992) The variation in the microfibrillar angle within the juvenile wood of sitka spruce. IAWA J 13:261 Pot D, Chantre G, Rozenberg P, Rodrigues JC, Lloyd JG, Pereira H, Hannrup B, Cahalan C, Plomion C (2002) Genetic control of pulp and timber properties in maritime pine (Pinus pinaster Ait.). Ann For Sci 59:563–575 Raymond CA, Anderson DW (2005) Prior land-use influences wood properties of Pinus radiata in New South Wales. N Z J For Sci 35:72–90 Rozenberg P, Cahalan C (1997) Spruce and wood quality: genetic aspects (a review). Silvae Genet 46:270–279 Taylor FW, Wang EI, Yanchuk A, Micko MM (1982) Specific gravity and tracheid length variation of white spruce in Alberta. Can J For Res 12:561–566 Ukrainetz NK, Kang K-Y, Aitken SN, Stoehr M, Mansfield SD (2008) Heritability, phenotypic and genetic correlations of coastal Douglas-fir (Pseudotsuga menziesii) wood quality traits. Can J For Res 38:1536–1546 Weng YH, Park YS, Krasowski MJ, Tosh KJ, Adams G (2008) Partitioning of genetic variance and selection efficiency for alternative vegetative deployment strategies for white spruce in Eastern Canada. Tree Gene Genom 4:809–819 Weng YH, Tosh KJ, Fullarton MS (2010) Determining and projecting realized genetic gains: results from early-stage spruce improvement programs in New Brunswick, Canada. N Z J For Sci 40:5– 17 Weng YH, Tosh KJ, Fullarton MS (2011) Effects of height-growth selection on wood density in black spruce in New Brunswick, Canada. For Chron 87(1):116–121

Tree Genetics & Genomes (2012) 8:303–311 Wimmer R, Downes G, Evans R, French J (2008) Effects of site on fibre, kraft pulp and handsheet properties of Eucalyptus globulus. Ann For Sci 65:602–608 Yanchuk AD, Kiss GK (1993) Genetic variation in growth and wood specific gravity and its utility in the improvement of interior spruce in British Columbia. Silvae Genet 42:141–148 Zelazny VF, Martin GL, Toner M et al (2007) Our landscape heritage: the story of ecological land classification in New Brunswick, 2nd edn. New Brunswick Department of Natural Resources, Fredericton Zhang SY (1995) Effect of growth rate on wood specific gravity and selected mechanical properties from distinct wood categories. Wood Sci Technol 29:451–465

311 Zhang SY (1998) Effect of age on the variation, correlations and inheritance of selected wood characteristics in black spruce (Picea mariana). Wood Sci Technol 32:197–204 Zhang SY, Jiang ZH (1998) Variability of selected wood characteristics in 40 half-sib families of black spruce (Picea mariana). Wood Sci Technol 32:71–82 Zhang SY, Morgenstern EK (1995) Genetic variation and inheritance of wood density in black spruce (Picea mariana) and its relationship with growth: implications for tree breeding. Wood Sci Technol 30:63–75 Zobel BJ, Jett JB (1995) Genetics of wood production. Springer, Berlin, p 337