Effect of growth rate on wood properties of genetically improved Sitka spruce

Transcription

1 Effect of growth rate on wood properties of genetically improved Sitka spruce A.K. LIVINGSTON 1, A.D. CAMERON 1 *, J.A. PETTY 1 AND S.L. LEE 2 1 School of Biological Sciences, University of Aberdeen, MacRobert Building, 581 King Street, Aberdeen AB24 5UA, Scotland 2 Forest Research, Northern Research Station, Roslin, Midlothian EH25 9SY, Scotland *Corresponding author. a.d.cameron@abdn.ac.uk Summary This study examined the wood properties of 24-year-old Sitka spruce (Picea sitchensis (Bong.) Carr.) progenies with highly contrasting growth rates. The progenies were established as part of a breeding programme to improve growth rate and stem form. Trees from three progenies were selected with a high growth rate relative to an unimproved control of directly imported material from the Queen Charlotte Islands (QCI), British Columbia, Canada. Trees from a further three progenies were selected that displayed a similar growth rate to the QCI control. At the time of sampling, the fastgrowing progenies had a mean volume gain over the QCI and slow-growing progenies of ~70 per cent. Trees from the fast-growing progenies were found to have significantly larger branches, less latewood, and more compression wood in comparison with the QCI control and the slow-growing progenies. On the other hand, trees from the fast-growing progenies had a smaller grain angle than the QCI control. While fast-growing progenies had lower wood density than the control, this was not significantly different. These findings suggest that, in general, the criteria used to select Sitka spruce trees in the forest as potential candidates for the breeding population would indeed lead to significant improvements of the growth performance and grain angle from improved planting stock. Breeders need to be aware, however, of possible negative influences of such selection criteria on other stem and wood properties known to influence wood strength. Introduction The programme of Sitka spruce (Picea sitchensis (Bong.) Carr.) genetic improvement began in Britain in early 1960s with the objective of developing a breeding population that is well adapted to a range of site types, of improved stem form and growth potential, and producing trees with wood qualities that are suitable for the sawn timber market (Fletcher and Faulkner, 1972). This programme has reached a stage where improved plants are widely available and extensively planted. However, there is insufficient knowledge on how this genetic improvement, particularly in terms of growth rate, is influencing the wood properties of Sitka spruce, such as Institute of Chartered Foresters, 2004 Forestry, Vol. 77, No. 4, 2004

2 326 FORESTRY branch and knot size, wood density, grain angle and development of compression wood. These characteristics influence the mechanical properties and drying behaviour of timber. Several studies correlating growth rate with wood density (e.g. Wood, 1986; Lee, 1995, 1997) have made Sitka spruce breeders aware of the need to consider wood density when selecting for diameter increment if existing values of wood density are to be maintained. Lee (1997) found that basic wood density falls by 5 per cent as selections are made for vigour to improve stem diameter or height. The negative genetic correlation between wood density and growth rate is not unique to Sitka spruce; it is common in most conifers (e.g. King et al., 1988; Park et al., 1989; Dean, 1990; Allen, 1992), but it would appear to be more acute in Sitka spruce (e.g. Wood, 1986; Lee, 1995). Timber strength is determined by more than just wood density. Grain angle and compression wood are both of importance. Compression wood is seen as a serious defect, as sawn boards containing both normal and compression wood dry with differential longitudinal shrinkage and may warp and check (Timell, 1986). Compression wood forms in conifers on the underside of leaning stems or on the leeward side of trees exposed to strong winds (e.g. Nicholls, 1982; Desch and Dinwoodie, 1996). There is some evidence, however, that vigorous growth may result in the formation of compression wood around the entire stem (Walker, 1993). This is of particular concern since a high growth rate has been one of the main selection criteria of the Sitka spruce breeding programme along with stem straightness and, to a lesser degree, wood density (e.g. Lee, 1995). The aim of this investigation was to compare the wood properties of genetically improved Sitka spruce progenies displaying a high growth rate with those of an unimproved control consisting of material directly imported from the Queen Charlotte Islands (QCI), British Columbia, Canada and also genetically improved progenies displaying a growth rate similar to that of the QCI control. Comparisons between the slow-growing progenies and a QCI control of similar growth rate allow the effects of selection to be taken into account in the absence of differences in growth. Methods Sample trees were selected from a progeny trial located at Aultmore Forest, Moray District in north-east Scotland (57 31 N W; National grid ref. NJ431576). The soil is an iron pan with an underlying geology of Dalradian quartose mica schist. The elevation is 260 m and the site is moderately exposed with a mean annual rainfall of 964 mm. The progeny trial was planted in 1975 with trees established at 1.7 m 1.7 m spacing on tine ploughing, and has never been thinned. The objective of the progeny trial was to compare a range of characteristics of growth and stem form among 48 progenies derived from open pollinated trees and an unimproved control from Queen Charlotte Island. A randomized block design was used with eight-tree line plots replicated four times. To limit the scale of the study, only three of the four replicates were used and these were selected at random. Periodic measurements of stem diameter and height indicated a wide variation in growth rate between progenies. The aim of the selection procedure was to choose three progenies with a high rate of growth (assessed by diameter at breast height (d.b.h.)) relative to the QCI control, three with a rate of growth similar to the QCI control, and a QCI control. Each progeny and the QCI control had three replicates. Progenies were excluded where trees were damaged, missing or gaps had formed in adjacent plots. To avoid selecting trees that were suppressed, only the five largest trees from each selected plot of eight trees were used in the study. Great care was taken to ensure that trees adjacent to the sample plots selected had a growth rate similar to that of the sample trees to limit variations in the local growing environment. Details of the selected progenies and QCI control are shown in Tables 1 and 2. A pilodyn measure of wood density was made on the standing tree at 1.3 m from the tree base. The pilodyn fires a blunt pin into the tree with a constant force (6 joules) and the depth of pin penetration is negatively related to wood density. This allowed direct comparisons with density measurements made on wood samples taken from a similar location. Sample trees were felled and tree height and extent of the living crown

3 WOOD PROPERTIES OF GENETICALLY IMPROVED SITKA SPRUCE 327 Table 1: Summary of the percentage deviation in diameter at breast height (d.b.h.) between the different groups of Sitka spruce sample trees Progeny Progeny Percentage d.b.h. deviation Treatment descriptor identification relative to QCI control at 24-years old Control QCI QCI 0 Slow-growing progenies Slow I Slow II Slow III Fast-growing progenies Fast I Fast II Fast III Table 2: Mean growth characteristics of sample trees of Sitka spruce at the time of felling in 1999 (age 24 years) No. of Height d.b.h. Volume Treatment trees (m) (cm) (m 3 ) QCI control (0.25) (0.69) (0.015) Slow I (0.34) (0.93) (0.022) Slow II (0.34) (0.75) (0.016) Slow III (0.30) (0.62) (0.015) Mean slow-growing progenies (0.21) (0.44) (0.010) Fast I (0.27) (0.82) (0.025) Fast II (0.28) (1.18) (0.031) Fast III (0.21) (0.74) (0.022) Mean fast-growing progenies (0.16) (0.53) (0.015) Standard errors are in parenthesis. based on the location of the lowest living whorl were recorded. Stem taper was calculated as diameter at 1.3 m divided by tree height, both in metres. Internodal sample discs, 10 cm thick, were removed from the base of each tree and at 0.1, 0.3, 0.5, 0.7 and 0.9 tree height. A further 50-cm-long section was removed from the base of the stem, with the top of the sample at a height of 1.3 m, for the determination of all wood properties. In addition, a disc containing the 10th whorl from the top of the tree (approximately at the base of the living crown and hence with the largest live branches) was removed for the determination of branch size and angle. Sample discs were used to determine the historical patterns of height, diameter and volume growth through stem analysis. The upper surface of each disc was finely sanded then passed under a computer-linked digital positiometer on a north-to-south axis and the distances of the ring boundaries recorded on computer files. Calculations of historical height, diameter and volume

4 328 FORESTRY were carried out using the procedures described by Cameron and Watson (1999). The proportions of compression wood and knot area were also assessed on the stem section at 1.3 m. Compression wood was identified by its relative opacity to transmitted light in comparison with the translucence of normal wood (e.g. Pillow, 1941). A thin disc of 3 5 mm was cut from the top of each of the 50 cm main sample log. Areas of compression wood that showed orange/red and opaque were carefully drawn on tracing paper. Knot area on the surface of discs was highlighted individually to allow for exclusion from the compression wood area. Each tracing was scanned into Image-Pro Plus (Version 3.0, Media Cybernetics) and areas of compression wood and knots measured. Diameter at the base of each branch on each disc taken 10 whorls down from the tip was recorded using a digital calliper and mean values determined. Branch angle relative to the longitudinal axis of the stem of the tree was also measured on these discs. CT derived density (g cm Ð3 ) Wood density was measured on 10-cm-thick discs, removed from the top of the 50-cm stem section, by means of CT (computerized tomography) scans using procedures described by Lindgren et al. (1992). These were calibrated from a sub-sample of relative density measurements based on volume and weight values using samples at 12 per cent moisture content. A good relationship was found to exist between CTderived density measurements and relative density (Figure 1). All statistical analyses were based on the CT-derived measurements. CTderived density measurements were expressed as an average across radially diametric sections from bark to pith, and an area-weighted average where variation in ring width was taken into account. Latewood proportion was also determined from the CT scans where the boundary between the lower density early wood with the higher density latewood was clearly seen. Grain angle was measured on a 20-cm-thick disc removed from the upper section of the 50 cm Relative density (g cm Ð3 ) Figure 1. Regression of CT derived density and relative density (measured at 12 per cent moisture content) based on wood samples of Sitka spruce from all treatments (R 2 = 0.74, P 0.001).

5 WOOD PROPERTIES OF GENETICALLY IMPROVED SITKA SPRUCE 329 main sample logs. These discs were cleaved longitudinally across the diameter through the pith in a north south direction using an Instron 4483 Universal Testing machine. Grain angle (angle between the outer stem surface in the vertical plane and the orientation of the grain) was measured using a modified protractor. Two-way analysis of variance was performed on the data with three treatments and three replicates. Further one-way analysis of variance was carried out on the individual progenies within fast- and slow-growing treatments. Distributions of data involving ratios and percentages were sufficiently normal not to require transformation. Associations were carried out using least squares regression. Results Stem analysis revealed the high volume growth rate of the fast-growing progenies in comparison with the slow-growing progenies and the unimproved QCI control (Figure 2) The mean volume Volume (m 3 ) of the fast-growing progenies exceeded that of the QCI control by 70 per cent. The objective to select slow-growing progenies with similar growth rates to the unimproved QCI control appeared to have been achieved (see Table 2). At 24 years old, when the trees were felled, an analysis of variance showed that the difference in d.b.h. and volume between fast-growing progenies and both the slow-growing progenies and QCI control was highly significant (P 0.001). There were no significant differences in mean d.b.h. and volume between the slow-growing progenies and QCI control. The fast-growing progenies were taller than both the slow-growing progenies and QCI control by 10 per cent and 12 per cent, respectively, and these differences were significant (P 0.001; Table 2). Stem taper at the time of felling was greater in fast growing trees than the other two treatments (P 0.001; Table 3). While there was no difference in crown depth between control and slow-growing trees, fast-growing trees had significantly deeper crowns than both these treatments (P 0.001). However, when crown depth Year QCI Slow I Slow II Slow III Fast I Fast II Fast III Figure 2. Mean tree volume of three fast- and three slow-growing progenies and an unimproved QCI provenance of Sitka spruce. Bars represent ± 1 SE.

6 330 FORESTRY Table 3: Mean stem and crown characteristics of sample trees of Sitka spruce at the time of felling in 1999 (age 24 years) Stem taper d.b.h./ Live crown Proportion of crown No. of living Treatment height (m/m) depth (m) depth to tree height (%) whorls QCI control ( ) (0.26) (1.15) (0.22) Slow I ( ) (0.29) (1.15) (0.27) Slow II ( ) (0.31) (1.60) (0.31) Slow III ( ) (0.31) (1.43) (0.42) Mean slow-growing progenies ( ) (0.17) (0.81) (0.19) Fast I ( ) (0.30) (1.11) (0.33) Fast II ( ) (0.28) (1.51) (0.30) Fast III ( ) (0.29) (1.12) (0.31) Mean fast-growing progenies ( ) (0.23) (0.76) (0.18) Standard errors are in parenthesis. Table 4: Mean number, diameter and angle of branches measured on the 10th whorl from the top of the tree of fast- and slow-growing progenies of Sitka spruce, and of QCI origin (control) Mean no. of Mean branch Mean branch Treatment branches diameter (mm) inclination angle ( ) QCI Control (0.38) (0.58) (1.52) Slow I (0.31) (1.05) (1.68) Slow II (0.47) (0.99) (1.48) Slow III (0.51) (0.83) (2.20) Mean slow-growing progenies (0.25) (0.58) (1.08) Fast I (0.34) (1.34) (1.87) Fast II (0.53) (1.22) (1.69) Fast III (0.55) (1.26) (2.70) Mean fast-growing progenies (0.29) (0.74) (1.37) Standard errors are in parenthesis.

7 WOOD PROPERTIES OF GENETICALLY IMPROVED SITKA SPRUCE 331 is expressed as either a proportion of tree height or number of living whorls, no significant differences between treatments were found. Table 4 shows mean branch number (>5 mm diameter) and diameter and branch angle measured on the 10th whorl from the top of the tree. There were no significant differences in number of branches and branch angle between treatments. There was a significant difference in branch diameter between treatments, with the fast-growing progenies having larger branches than the other treatments (P < 0.001). Compared with the QCI control, there was a preponderance of bigger branches in both slow- and fastgrowing progenies with the latter group of trees having very big branches (Table 5). There was no significant difference in pilodyn pin (PIN) measurements between treatments (Table 6). While mean radius density and area weighted density for QCI samples were higher in comparison with both the slow- and fast-growing progenies, these were not significantly different. There was no significant relationship between mean radius density and tree size described by d.b.h. (R 2 = 0.16) suggesting that growth rate alone did not explain these differences. The proportion of latewood between treatments was significantly different (P 0.001) with QCI samples having the highest level and the fastgrowing progenies the lowest percentage of latewood. There was no significant relationship between PIN measurements and mean radial density (R 2 = 0.16) suggesting that the pilodyn was not effective in differentiating density in the sample trees used in this study. There was a highly significant difference in cross-sectional area of compression wood between treatments (P 0.001; Table 7). Fastgrowing progenies have significantly more compression wood than both the slow-growing progenies and QCI control. There was no significant difference in proportion of knotwood between treatments. There was no correlation between compression wood area and knotwood area. A series of simple linear regressions on compression wood percentage and tree height, d.b.h., tree volume, stem taper, depth of living crown, and number of living whorls revealed no significant relationships (R 2 < 0.2) for fast- and slow-growing progenies, the QCI control and all trees combined. This indicates that none of these parameters individually explains the level of compression wood found. A further regression between proportion of compression wood and disc area indicated that larger discs, and therefore trees, do not have a proportionally greater area of compression wood in comparison with small trees (R 2 = 0.16). The QCI control had a higher mean grain angle than that of both the slow- and fastgrowing progenies (P 0.001). There was no statistical difference in mean grain angle between the slow- and fast-growing progenies (Table 7). Discussion A high growth rate as a result of selection of parent trees in the forest appears to have influenced the wood properties of Sitka spruce in several ways. While selection for growth did not influence the number or angle of branches produced on individual stem whorls, branch diameter was significantly greater on the fastgrowing progenies. The proportion of bigger branches on the fast-growing progenies was also greater than for the other two treatments. It is the Table 5: Proportion of branches of diameter 11 20, and >30 mm measured on the 10th whorl from the top of the tree of fast- and slow-growing progenies of Sitka spruce, and of QCI origin (control) Proportion of branches (%) Treatment (mm) (mm) >30 (mm) QCI control Slow-growing progenies Fast-growing progenies

8 332 FORESTRY Table 6: Wood density measurements derived from Pilodyn Pin Penetration (PIN), CT (computerized tomography) scans, and relative density (12 per cent moisture content), and percentage latewood of fast- and slow-growing progenies of Sitka spruce, and of QCI origin (control) Manually measured variables CT measured variables Relative Mean radius Area weighted Proportion PIN density density density* of latewood Treatment (mm) (g cm 3 ) (g cm 3 ) (g cm 3 ) (%) QCI control (0.67) (0.022) (0.010) (0.011) (0.359) Slow I (0.55) (0.022) (0.011) (0.011) (0.484) Slow II (0.64) (0.041) (0.007) (0.007) (0.390) Slow III (0.76) (0.024) (0.011) (0.010) (0.509) Mean slow-growing progenies (0.37) (0.018) (0.006) (0.005) (0.262) Fast I (0.70) (0.020) (0.007) (0.007) (0.362) Fast II (1.16) (0.003) (0.005) (0.006) (0.414) Fast III (0.63) (0.013) (0.009) (0.007) (0.512) Mean fast-growing progenies (0.49) (0.009) (0.004) (0.004) (0.254) Standard errors are in parenthesis. * Area weighted density is a derived figure based on the radial density and the corresponding cross-sectional area, the calculation therefore assumes the stem is circular. These calculations exclude values from the pith. knots formed by these large branches that have a greater influence on timber strength in comparison with small knots (e.g. Dinwoodie, 2000). For a 24-year-old tree, the 10th whorl from the top (where the branch measurements were taken) will be located in the main log section when the tree is felled typically between 45 and 55 years old. Mean density, measured across the radius (i.e. pith to bark) of discs removed from the sample trees at 1.3 m, was lower in the fast-growing progenies, but not significantly different from the other treatments. In addition, there was no significant correlation between mean radius density and d.b.h., suggesting that a high growth rate does not appear to have had a deleterious effect on wood density. While density has been shown to be positively correlated with strength and stiffness of small clear samples of wood (e.g. Desch and Dinwoodie, 1996), the presence of other strength-reducing factors, such as compression wood (which is denser but weaker than normal wood) and knots, means that it is not always a good predictor of mechanical properties. Pilodyn measurements, used as a surrogate for assessing wood density, were not correlated with wood density. Normally these measurements are made on much younger trees where acceptable correlations with wood density are found (Lee, 1995). The QCI (control) trees had the highest proportion of latewood and the fastgrowing progenies the lowest; however, no significant relationship (i.e. phenotypic correlation) was found between proportion of latewood and density (R 2 = 0.19). Fast-growing progenies also had a significantly greater proportion of compression wood than both the slow-growing progenies and QCI control. Many environmental factors have been

9 WOOD PROPERTIES OF GENETICALLY IMPROVED SITKA SPRUCE 333 Table 7: Mean compression wood and knotwood area (percentage of cross-section), and mean grain angle of fast- and slow-growing progenies, and of QCI origin (control) Proportion of Proportion of Mean grain Treatment compression wood (%) knotwood (%) angle ( ) QCI control (2.11) (1.057) (0.403) Slow I (4.66) (1.473) (0.203) Slow II (4.63) (1.492) (0.326) Slow III (4.65) (3.365) (0.244) Mean slow-growing progenies (2.65) (0.711) (0.152) Fast I (3.78) (0.652) (0.159) Fast II (4.20) (1.841) (0.270) Fast III (3.68) 0.976) (0.214) Mean fast-growing progenies (2.21) (0.740) (0.129) Standard errors are in parenthesis. found to stimulate compression wood formation, but one of the most frequently cited is wind (e.g. Nicholls, 1982). Trees from the faster-growing progenies were greater in height and diameter than those in the other two treatments and, as a result, may have been more vulnerable to the effects of wind at this experimental site where treatments were set out as an intimate mix. At the age of 24 years when the trees were felled, the difference between the tallest and shortest trees was <2 m (Table 2), although even this small difference in height may influence the level of wind loading. These height differentials are likely to be reduced in plantations consisting exclusively of improved planting stock from highly selected parents. Conversely, there was no significant difference in depth of the living crown between treatments, suggesting that the level of wind loading may have been broadly similar for the fast- and slower-growing trees. Furthermore, compression wood content was found not to be correlated with measures of growth. Cameron and Watson (1999) found that, in a mixed species stand on an exposed site, the largest Sitka spruce trees had the lowest compression wood content. They suggested that these trees, being bigger than their neighbours from a young age, would have developed stabilizing features with the effect of reducing stem sway and the development of compression wood. However, the trees in the experiment reported by Cameron and Watson (1999) were relatively slow growing in comparison with the spruce studied in this current study. Vigorous growth in conifers has been observed to result in the formation of compression wood all around the stem (Walker, 1993) and may be the reason for the high level of compression wood observed in the wood of the fast-growing progenies. The lack of correlation between the proportions of compression wood and knotwood suggests that knots did not influence the development of compression wood in any meaningful way. The lack of clear differences in density measurements between treatments may, in part, be a result of the compensating effect of a greater proportion of compression wood (of higher density than normal wood) in fast-growing progenies. Mean grain angle (measured across the radius) was significantly lower in both the slow- and

10 334 FORESTRY fast-growing progenies in comparison with the QCI control. While the overall magnitude of the grain angle found, even in the QCI control is very low, it nevertheless appears that selection has had the effect of limiting this characteristic. Decreasing grain angle is likely to have a significant effect in improving the bending strength and stiffness of Sitka spruce timber (e.g. Maun, 1998). The findings presented in this study indicate that, in general, selection for high growth rates may have negatively influenced certain aspects of the wood quality of Sitka spruce. The presence of relatively high levels of compression wood is worrying and may result in weaker timber. Larger branches in these progenies will likely exacerbate the problem, although to what degree the decrease in grain angle will alleviate the problem is not yet known. Tree breeders need to be continually aware of the importance of traits other than growth rate in their breeding programmes. This has been accounted for in the Sitka spruce breeding programme in Britain by progeny testing large numbers of trees selected in the forest and further screening out of trees that do not meet certain wood quality criteria (Lee, 1995). Acknowledgements The authors wish to express their thanks to Mr David Moore, School of Biological Sciences, University of Aberdeen, for preparing samples used for estimating compression wood content and analysing the data. Thanks also go to Forest Research and the Scottish Agricultural College for organizing the CT scanner use. We thank the Scottish Forestry Trust for funding this work. References Allen, P.J Selection indices for the genetic improvement of Caribbean pine to increase sawn timber production. Aust. For. 55, Cameron, A.D. and Watson, B.A Effect of nursing mixtures on the stem form, crown size, branching habit and wood properties of Sitka spruce (Picea sitchensis (Bong.) Carr.). For. Ecol. Manage. 122, Dean, C.A Genetics of growth and wood density in radiata pine. Ph.D. thesis, University of Queensland, Australia. Desch, H.E. and Dinwoodie, J.M Timber Structure, Properties, Conversion and Use. Macmillan Press, London. Dinwoodie, J.M Timber: its Nature and Behaviour, 2nd edn. E & FN Spon, New York. Fletcher, A.M. and Faulkner, R A plan for the improvement of Sitka spruce by breeding and selection. Forestry Commission Research and Development Paper No. 85. Forestry Commission, Edinburgh, Scotland. King, J.N., Yeh, F.C., Heaman, J.Ch. and Dancik, B.P Selection of wood density and diameter in controlled crosses of coastal Douglas-fir. Silvae Genetica 37, Lee, S.J Multi-trait selection of Sitka spruce clones from progeny tests planted over an 11-year period. In Proceedings of the Joint Meeting of the IUFRO Working Parties S , 06, 12 and 14. Limoges, France. Lee, S.J The genetics of growth and wood density in Sitka spruce estimated using mixed model analysis techniques. Ph.D. thesis, University of Edinburgh. Lindgren, O., Davis, J., Wells, P. and Shadbolt, P Non-destructive wood density distribution measurements using computer tomography. Holz als Roh-und Werkstoff 50, Maun, K.W The relationship between growth characteristics of sawn timber, logs and machine grade stiffness. Building Research Establishment Client Report No. 166/98 (unpublished). Nicholls, J.W P 1982 Wind action, leaning trees and compression wood in Pinus radiata D. Don. Aust. For. Res. 12, Ni Dhubhain, A., Evertsen, J.A. and Gardiner, J.J The influence of compression wood on the strength properties of Sitka spruce. For. Prod. J. 38, Park, Y.S., Simpson, J.D., Fowler, D.P. and Morgenstern, E.K A selection index with desired gains to rogue jack pine seedling seed orchards. Information Report M-X-176. Forestry Canada Maritimes Region, Fredericton, NB, Canada. Pillow, M.Y A new method of detecting compression wood. J. For. 39, Timmell, T.E Compression Wood in Gymnosperms. Springer-Verlag, Berlin. Walker, J.F.C Primary Wood Processing: Principles and Practice. Chapman and Hall, London. Wood, P.E Variation and inheritance of wood properties of Sitka spruce. M.Sc. thesis, University of Oxford. Received 17 October 2003