Collection Mémoires et thèses électroniques
AccueilÀ proposNous joindre

Chapitre 4. Characterization of juvenile wood to mature wood transition age in black spruce (Picea mariana (Mill.) B.S.P.) at different stand densities and sampling heights

Table des matières

Le profil radial de la densité maximale du cerne et de la surface de cerne de trente six arbres d’épinette noire a été utilisé pour déterminer l’âge de transition du bois juvénile au bois mature. Les données ont été obtenues par densitométrie aux rayons X et la régression par segment ainsi que la régression polynomiale ont été utilisées pour trouver l’âge de la limite du bois juvénile. Trois densités de peuplement (1790, 2700 et 3400 arbres/ha) et trois hauteurs d’échantillonnage (2,4; 5,1 et 7,8 m) ont été étudiées. Bien que la densité maximale de cerne et la surface de cerne présentent un profil radial identique, ils donnent deux résultats significativement différents d’âge de transition. La densité maximale de cerne surestime l’âge de transition (17,6 années à 2,4 m) par rapport à la surface de cerne (14 années à 2,4 m). Les résultats montrent que la transition du bois juvénile au bois mature se produit après 12 ans à 7,8 m (versus 13,1 ans à hauteur de 5,1 m, et 17,6 ans à 2,4 m). Bien que l’âge de transition se produise plus tard dans le peuplement le plus dense (21 années à 2,4 m), la différence n’est pas significative entre les trois classes de densité. Néanmoins, l’âge de transition reste difficile à déterminer car il n’existe pas de définition standard. La transition se produit sur plusieurs années, et plus probablement il existe un bois de transition entre le bois juvénile et le bois mature. L’estimation de la proportion de bois juvénile en volume montre que celle-ci reste constante le long de la tige et qu’elle augmente avec la densité de peuplement.

The radial pattern of both maximum ring density and ring area of thirty-six black spruce trees were used to determine the transition age from juvenile wood to mature wood. The data were obtained by X-ray densitometry and both segmented linear and polynomial regressions were used to point out the age of the juvenile wood boundary. Three stand densities (1790, 2700 and 3400 stems/ha) and three sampling heights (2.4, 5.1 and 7.8 m) were studied. Although maximum ring density and ring area presented similar radial patterns, they gave two significantly different results of transition ages. The maximum ring density over-estimated the transition age (17.6 years at 2.4 m) in contrast to ring area (14 years at 2.4 m). The results show that the transition from juvenile wood to mature wood occurred after 12 years at 7.8 m (versus 13.1 years at a height of 5.1 m, and 17.6 years at 2.4 m). Although transition age occurred later in the high stand density group (21 years at 2.4 m), the difference was not significant between the three stand density groups. Nevertheless, transition age remains difficult to determine since no standard definition exists. The transition occurs over years, and most probably a transition wood exists between juvenile wood and mature wood. Estimation of the juvenile wood proportion in volume shows that it remains constant along the stem and it increases with stand density.

Key words: Stand density, sampling height, juvenile wood, transition age, ring area, maximum ring density, black spruce.

The radial cross-section of a stem can be divided into three zones (Clark and Saucier 1989): juvenile wood (JW), transition wood and mature wood (MW), but most of the time only two are considered, JW and MW. It has been widely accepted that JW, also known as immature wood, pith wood, or inner wood, is produced under the influence of the uppermost part of the tree crown. Formerly, the core JW was thought to be cylindrical, however, Zobel et al. (1959) proposed a more or less conical shape. Indeed, the formation of JW is related to the year of formation of the cambial initials (Yang et al. 1986), which is what induces the conical shape of JW. It has been shown that JW properties can be an indicator of MW properties in black spruce (Picea mariana (Mill.) B.S.P.) (Koubaa et al. 2000). However, in contrast with MW, many JW features are undesirable for most specific wood uses. This is of interest considering that the percentage of JW has a tendency to be higher in a suppressed tree than in a dominant tree (Yang et al. 1986). Juvenile wood is characterized by a lower density, shorter tracheids or fibers, lower latewood percentage, thinner cell walls, smaller tangential cell dimensions, lower cellulose content, lower strength, higher longitudinal shrinkage, higher microfibril angle, larger cell lumen, more reaction wood, more spiral grain, and a higher degree of knottiness than MW (Panshin and de Zeeuw 1980). Mature wood in softwoods is defined by a relatively constant tracheid length, whereas JW is characterized by an increasing tracheid length from pith outward (Yang and Hazenberg 1994). With the move to intensive silviculture, the proportion of JW relative to MW increases, resulting in warping problems during the drying process. Because of the particular physical and mechanical properties of JW, its proportion can have a significant impact on wood quality. For example, a high JW proportion involves a reduced lumber strength and a reduction of yields in pulp production.

The JW-MW boundary is difficult to establish because of the gradual transition between these two types of wood. Nevertheless, it is of great interest because of its implication on JW proportion. It has been shown that this transition occurs over years (Yang et al. 1986; Bendtsen 1978). Based on tracheid length, Yang and Hazenberg (1994) determined that the transition between JW and MW occurred between ring 11 and ring 21 in black spruce. Various criteria were used to locate the boundary between these two types of wood, including fibre length, microfibril angle, and longitudinal shrinkage. Among the relationships between tree growth and wood quality, the traits most frequently studied are ring density and ring width (Zhang et al. 1996). Wood properties change from pith to bark, and exhibit variable patterns, with some increasing and others decreasing. Since some ring traits of black spruce are affected by stand density and sampling height (Alteyrac et al. 2005), the JW boundary is also expected to depend on these factors. Nevertheless, the initial spacing has been reported not to influence the JW production period in Norway spruce and southern pine (Kucera 1994; Clark and Saucier 1989) while affecting the diameter of the juvenile core. A study by Yang (1994) also revealed that spacing had no influence on the number of rings of JW in 35-year old black spruce trees.

Variation within the tree is characterized by the decreasing number of years needed to attain a uniform tracheid length from the bottom to the top of the tree, meaning that the juvenile period is shorter in the top log (Yang et al. 1986). This results in major wood property differences between top and butt logs.

Previous studies have shown that a distinct JW-MW boundary could be determined by latewood density, growth ring specific gravity or tracheid length profile analysis (Abdel-Gadir and Krahmer 1993a; Sauter et al. 1999; Clark and Saucier 1989; Yang et al. 1986). Other techniques based on regression analysis were also proposed (Abdel-Gadir and Krahmer 1993a). Because of difficulties in estimating the transition age (TA), a reliable and quick method is necessary to determine it. The objectives of this study were to identify an appropriate methodology for the determination of the JW-MW transition age in black spruce wood, and to determine its variation as a function of sampling height and stand density.

Thirty-six sample trees were collected from a stand located in the Chibougamau area, 400 km north of Quebec City, 49°21’N and 75°05’W, at an altitude of 400 m. The stand density was 1967 trees/ha and the basal area was 41.3 m²/ha, with an average tree height of 13.8 m at age 79 (Horvath 2002). This stand was naturally regenerated from fire and exempt from silvicultural treatment. Therefore, all the trees had comparable ages and heights, but were randomly spaced. The sample trees were selected from the same stand in order to study the variability due to local stand density only. The Schütz index was used to determine the competition level for each sample tree (Ung et al. 1997). The local stand density was calculated from the number of surrounding trees in a 4-m radius circular sample plot (Alteyrac et al. 2005). Small trees (less than 8 cm at DBH) and dead trees were not considered in the calculation because of their limited competition effect. The 36 trees were selected and categorized into three stand density groups of the same span, namely: 1790 (1390-2190), 2700 (2390-2990) and 3400 (3190-3590) stems/ha or low, medium and high stand densities, respectively. The sample logs were removed at three sampling heights of 2.4, 5.1 and 7.8 m, then strips of 1.57 mm thick (Alteyrac et al. 2005) were extracted and analyzed by X-ray densitometry at Forintek Canada Corp. Eastern division. The microfibril angle (MFA) was measured by the Silviscan technology at CSIRO Forestry and Forest Products, Australia, on a total of 12 samples taken at 2.4-m high.

X-ray densitometry provided the radial patterns of several properties including ring basic density (RD), ring width (RW), and ring maximum density (MD). Ring area (RA) was computed from RW assuming a circular shape for the growth rings. The transition age (TA) was first visually estimated from the radial pattern of each property by a quantitative determination based on the regression analysis techniques described below.

Visual interpretation

The average RA, MD, RW and RD radial patterns of all 36 trees were plotted with respect to cambial age. The JW-MW transition age was then determined by visual interpretation of the slope changes of the above properties. The average radial pattern of MFA was also plotted to estimate TA.

Polynomial regression

The first regression technique used to estimate TA was a polynomial third-order regression and was inspired by the technique used to analyse the transition from earlywood to latewood (Koubaa et al. 2002) and the transition from JW to MW (Koubaa et al. 2004) in black spruce. From each tree, the radial profile of RA and MD was plotted with respect to cambial age (CA) and fitted with a third-order polynomial regression of the following form:.

(4.1)

From equation (4.1), the maximum value of Y was considered to be TA.

Segmented linear regression

The SAS NLIN segmented linear regression procedure was used as a second regression technique to characterize the RA and MD profile curves. This kind of segmented method or piecewise regression was used previously to analyse curves that present a changing slope (Bustos et al. 2003, Sauter et al. 1999, Abdel-Gadir and Krahmer 1993a). The model consists of finding two linear regressions to characterize the variation of traits with respect to cambial age. Equation (4.2) for JW and (4.3) for MW were used to find the intersection admitted to be TA from equation (4.4).

(4.2)

(4.3)

Where Y is the variable considered, RA or MD, a, b, c and d are the parameters, and X is the cambial age. X0 is unknown and automatically determined by iteration in such a way as to provide the best coefficient of determination of the model including the two regressions by a numerical optimization method.

The intersection point of the two lines considered as TA is calculated as follows:

(4.4)

Derivative function

According to the radial profile of MFA, a derivative function of the profile was used to determine the point from which the variation of MFA with respect to cambial age was null. Thus the radial profile was divided into two parts. A first part where MFA is decreasing was considered as juvenile wood, and the second part where MFA is constant was considered as mature wood.

Studied features

After visual examination, RW and RD were discarded, and MFA was kept as a comparison trait. The two remaining traits, RA and MD, were studied because their radial patterns suggested that they would be useful to find TA, as they presented a maximum at about age 20.

Ring area characterizes growth rate as RW does but RA was preferred to RW because the slope change in the RW radial profile that occurs at age 4-5 does not correspond to the main transition from JW to MW (Yang and Hazenberg 1994).

Maximum ring density showed a strong correlation with latewood density (r=0.76). Maximum ring density was therefore selected because this parameter further presented a radial pattern similar to RA. This trait was also used by Sauter et al. (1999) to estimate TA in Scots pine. Thus, the selection of RA and MD provided the advantage of using two traits representing growth rate and wood density, which are two major wood quality attributes.

Sampling height and stand density

The average TA was calculated with respect to sampling height and stand density. Three sampling heights: 2.4, 5.1 and 7.8 m; and three stand density groups 1790, 2700 and 3400 stems/ha were considered. An ANOVA and a Duncan test were carried out to compare the three stem levels and the three stand density groups.

Estimation of the juvenile wood proportion

An estimation of the JW proportion was calculated from the average TA found in each sampling height and each stand density group. Equations (4.5) and (4.6) present the radius of JW (JWR) and the corresponding area of JW (JWA) respectively, calculated from the RW average value of each category of sampling height and stand density.

(4.5)

and (4.6)

where j is the sampling height, and TAij is the transition age at the jth sampling height in the ith stand.

Equations (4.7) and (4.8) were used to calculate respectively the radius of the trees (TR) and the corresponding area. The trees were taken as an example in year 1980.

(4.7)

where RWh is the ring width at cambial age h.

and

(4.8)

Equation (4.9) was used to determine the JW proportion by area using values calculated from equations (4.6) and (4.8). The index number j indicates that the equation was formulated at the jth sampling height.

(4.9)

Log volume and JW volume were estimated using the formula of a truncated cone given in Equation (4.10) where L is the log length (Figure 4-1).

(4.10)

The truncated cone formula was preferred to the Smalian, Huber or Newton formulas because it refers only to the small and large ends dimension (Patterson et al. 1993). The data available to calculate the logj volume are the cross-section areas at the large and small ends, respectively denoted as radiusj-1 and radiusj.

The total volume of logij, which is the log from the j-1th height (index j-1) to the jth sampling height (index j) in the ith stand was calculated using equation (4.11).

(4.11)

Then, the JW proportion in volume was calculated from equations (4.10) and (4.11) in equation (4.12).

(4.12)

The profiles of MD and RA at 2.4 m sampling height from pith to bark are presented in Fig. 4-2. The two traits exhibit a similar radial pattern. They increase until about ring 20 and then decrease until bark. Only few rings near the pith presented a high MD value most likely due to the presence of compression wood. These data were omitted in the following analysis. These profiles show the expected transition age at a glance.

The profiles of RW, RD and MFA at 2.4 m sampling height from pith to bark are presented in Figure 4-3. The JW boundary is not defined as clearly as in Figure 4-2. As mentioned above, although RW presents a maximum at ring 4 which suggests a transition between two types of wood, it does not point out the JW boundary. According to the MFA profile, a first transition at about ring 10 and a second at about ring 25 are consistent with the presence of three zones in wood as described previously (Clark and Saucier 1989). Finally, the RD profile shows a minimum at about ring 10, also revealing a transition between two zones, but this can not be considered as the JW to MW transition.

According to Figures 4-2 and 4-3, TA was estimated to occur at about ring 4 to 5 with RW, ring 10 with RD, ring 10-25 with MFA, ring 20 with RA and ring 25 with MD. As a first consequence, the results suggest that RD and RW are not appropriate traits to determine the transition age. As the TA is expected to occur in the 11-37 years range in Douglas fir (Abdel-Gadir and Krahmer 1993a) and 11-21 years in black spruce (Yang and Hazenberg 1994), RA and MD were selected as the best traits to determine TA and MFA was used as a comparison trait. As a second remark, Figure 4-2 shows that MD leads to a later TA in comparison to RA. Nevertheless, their radial patterns are similar (Figure 4-4).

The efficiency of RA to determine the JW boundary should be emphasized. Indeed, both RW and RA refer to tree growth (Clark and Saucier 1989). However, RA provides information about the radial and tangential growth of the annual rings, and about the increasing tree diameter. From this point of view, ring area can be considered as a better indicator of growth rate than RW.

Variation of transition age with traits and method

The transition age for each tree was calculated from RA and MD, using polynomial and segmented linear regression as described above. These two methods applied to the two traits, RA and MD, led to four different results for TA at the 2.4 m sampling height (Table 4-2). The difference between the methods was significant while the difference between stand density was not (Table 4-1). The results also show that TA determined from MD is higher than TA determined from RA as it was already noticed in the preliminary visual analysis. Moreover, the methods of linear regression and polynomial regression led to two significantly different results. The segmented linear regression method, as shown in Table 4-2, results in an earlier TA when compared to the polynomial regression. The segmented linear regression method was selected because it gave results closer to the results found with MFA (Table 4-2).

The analysis confirms that the determination of TA is dependent on the trait considered. Although RA and MD have similar radial patterns, the TA value found by the two traits is significantly different.

The transition age was also calculated with MD by the segmented linear regression. Although RA is a growth rate indicator and should be suitable to find the JW boundary it is not measured directly as MD is. On the anatomical point of view the MD value corresponds to the external boundary of the annual growth ring and does not depend on the determination of the earlywood-latewood transition such that if latewood width is not accurately estimated it does not change the MD value. Moreover, results obtained with MD are closer to results found with MFA by the same method of regression. These reasons justify to select MD instead of RA to determine TA.

Table 4-2 shows the TA values obtained from microfibril angle measurement. The results were first obtained by a derivative function, showing a transition age around age 18 (Table 4-2), and then by a segmented linear regression with a resulting TA of 15.9 years. The radial pattern of MFA is given with respect to cambial age in Figure 4-3. It shows more than two wood zones, causing difficulties to distinctly define JW and MW. This is in agreement with the conclusions of Clark and Saucier (1989). However, since a segmented regression was used two zones instead of three could be defined, one zone where MFA consistently decreased, and one zone where the value of MFA was constant.

Variation of TA with sampling height

In the following part TA was estimated by the segmented linear regression carried out on the MD variable.

While there is no significant difference of TA among stand densities, there is a significant difference in TA among sampling heights (Table 4-3). The results show that TA was 17.6 years at 2.4 m, 13.1 years at 5.1 m, and 12.0 years at 7.8 m (Table 4-4). Thus, it was observed that the transition age occurred earlier with increasing sampling height (Tables 4-4 and 4-5). The shorter juvenile period up the stem could be a consequence of many factors, including a reduction of crown volume with tree aging which is linked to the production of JW, a possible attenuation of growth rate due to cambium aging, or most likely, a weak growth rate and activity of lower branches while stand canopy closes. These conditions induce a JW with more adult characteristics at the top of the tree (Alteyrac et al. 2005; Yang et al. 1986) because rings are smaller and more dense.

Variation of TA with stand density

No significant difference in TA among stand density groups and no interaction between stand density and sampling height were observed (Table 4-3). This is consistent with the results of Yang (1994) on black spruce and involves that TA does not depend on stand density in the stand density range considered in this study. Moreover, no significant difference in RD between the three stand density groups was found in a previous study (Alteyrac et al. 2005). The lack of difference was explained by both the low stand density range and the high average stand density value (Zhang and Chauret 2001) of over 1400 trees/ha.

Estimation of the juvenile wood proportion

Even though the number of rings corresponding to JW is unchanged by stand density variation, this does not necessarily indicate that the JW area is not affected. It was shown that growth traits depend on stand density, particularly in JW (Alteyrac et al. 2005). In that study, it was emphasized that a wide spacing allows a higher growth rate, which means larger rings during the juvenile period. In consequence, this induces a higher juvenile area for wider spacing, although the period of juvenility is unchanged. At the same time the diameter of the trees depends on tree spacing (or sampling height) so that both basal area of the tree and basal area of JW are changing with respect to stand density (or sampling height). This observation led us to estimate the JW proportion for the three different stand densities and sampling heights.

The JW proportion was estimated on an area or volume basis with equations (4.9) and (4.12) according to whether it is calculated on a cross section or a log. As an example, the proportion of juvenile wood was calculated for 60 years old trees corresponding to the annual ring produced in 1980 (Table 4-5).

During the first years of growth and up the stem, the trees produce JW only under the effect of the live crown. This is characterized by the horizontal line at 100% shown in Figure 4-5. Then the JW proportion decreases because the cambium stops to produce JW and starts to produce MW while the live crown moves up. The tree growth process involves that trees are producing JW in the live crown, while they are ending producing JW in the middle of the tree, and are producing MW in the butt log. Thus, while the tree is aging, there is a tendency of the JW proportion to reach a plateau (Figure 4-5) whatever the sampling height considered. The plateau can be explained by the ring width and ring area decreasing that induce a very low increasing of volume.

The results reveal a lower proportion of JW area for the low stand density (Table 4-5). These results are based on estimation and can not be supported by a statistical analysis of variance. Nevertheless they are consistent with some studies where a lower basal area proportion of JW associated to fast growing southern pine trees was observed (Clark and Saucier 1989). Moreover, in terms of volume, higher stand density produced a higher proportion of JW. This is mainly due to the fact that the total radius of the tree is reduced although TA is slightly affected by increasing stand density, which in turn induces a higher JW proportion for high stand densities.

1. Proportion of juvenile wood between 2.4 m and 5.1 m2. Proportion of juvenile wood between 5.1 m and 7.8 m

Considering the variation among sampling heights, a variation of JW transition age was observed, and consequently a variation of the JW radius. As both JW radius and total radius were found to vary with respect to sampling height, this resulted in an unchanged JW proportion (Table 4-5). A 60-year-old tree growing in a low stand density site has an estimated JW area of 45.8% at a 2.4-m height (Table 4-5). The variations of TA and JW diameter from the bottom to the top of the tree confirm former results about the conical shape of JW (Zobel et al. 1959, Yang et al. 1986).

Yang (1994) found the percentage of JW basal area to be 50% for black spruce trees at 35 years old, at breast height and in a plantation. Under the same conditions, a 35-year-old tree in our study (1965) would have a JW basal area of about 65% (Figure 4-5). These results show that fast growing conditions result in a lower proportion of JW area. Nevertheless, these close results are underlying the difficulty involved in determining the exact amount of JW and the TA. But as evidence, our results show that the proportion of JW decreases with age (Figure 4-5). This demonstrates the benefits of longer rotations for black spruce.

The results of this study lead to the following conclusions:

Two variables, maximum ring density and ring area, were used to determine the transition age at different stand density groups and sampling heights. From the observations of this study, the determination of transition age remains difficult since a standard method for this determination does not yet exist and because of the presence of a likely transition wood. Nevertheless, the use of maximum ring density and segmented linear regression seem to be suitable to determine a theoretical boundary between juvenile wood and mature wood.

Transition age varies significantly with sampling height (12 years at 7.8m, 13.1 at 5.1m and 17.6 at 2.4m), and occurs earlier at the top of the tree. It results in a lower juvenile wood diameter at top of the tree than at the bottom, producing a conical shape of the juvenile wood zone. Nevertheless, the variation of transition age does not seem to have an influence on the proportion of juvenile wood. The simultaneous variation of juvenile wood diameter and tree diameter with height lead to a constant proportion of juvenile wood.

No significant differences in transition age were found between the three stand density groups.

The authors would like to thank Dr. Geoff Downes, of CSIRO Forestry and Forest Products, Australia, for his assistance in determining microfibril angle with the SilviScan and for his great effort to run our samples in a short time. We would like also to thank the Natural Sciences and Engineering Research Council of Canada (NSERC) for funding under strategic grant no. 234774.

© Jérôme Alteyrac, 2005