Effect of Growth Rate on Fibre Characteristics in Norway Spruce ...

H. Mäkinen et al.: Fibre Characteristics in Norway Spruce

449

Holzforschung 56 (2002) 449–460

Effect of Growth Rate on Fibre Characteristics in Norway Spruce (Picea abies (L.) Karst.) By Harri Mäkinen1, Pekka Saranpää1 and Sune Linder2 1 2

Finnish Forest Research Institute, Vantaa, Finland Swedish University of Agricultural Sciences, Department for Production Ecology, Uppsala, Sweden

Keywords

Summary

Fertilisation Fibre length Fibre width Cell wall thickness

To study the effect of growth rate on fibre characteristics and their variations in Norway spruce, trees were sampled in a nutrient optimisation experiment in northern Sweden. Data was collected from 24 trees (40 years old) from fertilised and control plots after 12 years of annual nutrient application, as well as from older trees outside the experimental area. Fibre length, fibre diameter, cell wall thickness, lumen diameter and cell wall percentage were measured from every third annual ring at breast height and at a height of 4 m. Fibre properties, as well as their standard deviation, were closely related to ring number and distance from the pith. Intra-ring variation of fibre characteristics was high compared to their variation between trees. Fertilisation reduced fibre length and cell wall thickness, but increased fibre and lumen diameter in rings of the same age. The difference in fibre width, cell wall thickness and lumen diameter between fertilised and control trees was less apparent, but a greater difference in fibre length was found between the treatments with regard to distance from the pith. There was a similar effect of fertilisation on fibre properties in early- and latewood. The effect of enhanced growth rate was less pronounced at a height of 4 m (near the pith) than at breast height (in older rings). It was demonstrated that it is possible to model intra-tree variability of fibre characteristics using ring width and cambial age as independent variables. Models presented are, however, limited by the relatively young age of the sample trees used.

Introduction Dimensions of fibres have been subject of investigations for more than a century, and the general pattern of variations in fibre dimensions of conifers is well known (e.g. Panshin and de Zeeuw 1980; Zobel and van Buijtenen 1989). Fibre length and diameter increase rapidly and non-linearly during the first years of radial growth, and thereafter more gradually in mature wood. Within a stem, the main factor causing variation in fibre properties is ageing (maturation) of the cambium producing new fibres. Olesen (1982) stated that the vascular cambium is subject to two types of maturation processes, namely (i) formation of the cambium from the apical meristem (cyclophysis) and (ii) processes which the cambium undergoes after its formation. According to Olesen’s investigations in Norway spruce, fibre width changes both in a radial direction and with height in the stem. Helander (1933) found that trees growing in sites of lower fertility have longer fibres than trees on fertile sites. Differences in cell length, diameter, cell wall thickness and cell arrangement reflect the changes occurring in the cambium and the effect of environmental factors. Xylem development is influenced by external factors such as water availability, temperature, light and nutrients. Recent investigations have shown that increasing Holzforschung / Vol. 56 / 2002 / No. 5 © Copyright 2002 Walter de Gruyter · Berlin · New York

availability of light, water and nutrients increased tree growth and fibre diameter, but decreased fibre length (Lindström 1997; Dutilleul et al. 1998; Herman et al. 1998). In addition to maturation processes, fibre properties seem, therefore, to be related to the growth rate of trees. Contradictory results have, however, been reported on the effects of growth rate on fibre dimensions (e.g. Zobel and van Buijtenen 1989; Bergqvist et al. 2000). From a management point of view, more detailed information is needed on the effects of silvicultural treatments on fibre properties. Forest management has changed during the last decades and delayed thinnings are a problem, especially in young stands. At the same time there is a general trend for more intensive silviculture. For example, one of the main goals of Finland’s ‘National Forest Programme 2010’ is to intensify silvicultural treatments. More intensive silvicultural practices will increase the rate of tree growth, but will also lead to changes in fibre and wood properties. Characteristics of fibres largely determine the suitability of wood for further processing in the pulp and paper industry, as well as in the sawmill (e.g. Tyrväinen 1995). Fibre morphology and cell wall structure directly influence fibre flexibility, plasticity and resistance to processing and, therefore, influence the strength and other physical and optical properties of the end-prod-

450


uct. Therefore, knowledge of the processes determining fibre characteristics under different growing conditions and silvicultural treatments is of economic interest. Until now, the pulp and paper industry has mainly examined fibre properties and chemical composition of wood after processing, and the biological aspects related to wood formation have been taken into account to a lesser degree. In addition to mean fibre characteristics, their variation influences the quality of wood as raw material. Many paper properties are related to mean fibre characteristics, but knowing also their variation may lead to a better control of the properties of the end-product. Studies on frequency distributions of fibre characteristics are, however, scarce. As an exception, Ollinmaa (1959) and Herman et al. (1998) reported on the variation of fibre characteristics in relation to cambium age, growth rate and height within the tree. The aim of the present study was to investigate the effects of different growth rates obtained by fertilisation in a long-term nutrient-optimisation experiment on the fibre properties in young Norway spruce trees. The experiment was established to demonstrate the potential yield of Norway spruce, under given climatic conditions and non-limiting soil water, by optimising the nutritional status of the stand (Linder and Flower-Ellis 1992; Linder 1995). The yield of stemwood surpassed the best yields obtained by conventional silvicultural means (cf. Bergh et al. 1999), and should therefore represent the extreme impact of fertilisation on fibre properties. The effect of treatment on the branch characteristics in the same stands was presented in Mäkinen et al. (2001). Materials and Methods Site description and treatments The study was performed in a long-term nutrient-optimisation experiment at Flakaliden (64 °07’N; 19 °27’E; alt. 310 m a.s.l.) in northern Sweden. The principal aim of the experiment was to demonstrate the potential yield of Norway spruce (Picea abies (L.) Karst.), under given climatic conditions and non-limiting soil water, by optimising the nutritional status of the

stands, at the same time as leaching of nutrients to the groundwater was avoided (cf. Linder and Flower-Ellis 1992; Linder 1995). The experiment was established in 1986 in a young Norway spruce stand planted in 1963 with four-year-old seedlings of a local provenance. Before the initiation of the experiment, the site was classified according to Hägglund and Lundmark (1977) as fairly infertile (H100=17–19 m). The monthly mean temperature at the site varies from –8.7 °C in February to 14.4 °C in July and mean annual precipitation is approximately 600 mm, of which more than one-third falls as snow (Bergh et al. 1999). The treatments which began in 1987 included non-treated control plots, irrigated plots and two nutrient optimisation treatments. Treatments were replicated four times in a randomised block design, and each replicate consisted of a double plot made up of two 50 × 50 m plots. Each plot contained a net plot (1000 m2) surrounded by a buffer zone. In the present study, only control (C) and irrigated-fertilised (IL) plots were included. In the IL treatment, all essential macro- and micronutrients were supplied every second day during the growing season (mid-June to mid-August) and water was supplied to maintain a soil water potential above –100 kPa. The amount and composition of the nutrient addition was determined each year on the basis of nutrient analysis of foliage, the monitoring of nutrients in the soil water and predicted growth response. For further details regarding treatments, see Linder (1995). When establishing the experiment in 1986, the stand density was ca. 2400 trees per hectare and no thinnings were done thereafter. At that time the mean height of the trees in C- and IL-plots was 2.8 m and 3.0 m, respectively, and diameter at breast height 33.3 mm and 35.8 mm (Bergh et al. 1999). Measurements The 24 trees used in the present study were harvested in autumn 1998, after 12 years of treatment. On each replicated plot, three trees were chosen as sample trees according to the diameter distribution of the trees in the stands, i.e. one tree representing the mean diameter and two trees larger than 1.5 standard deviation added to the mean diameter. Suppressed trees were not measured because they will be removed in thinnings. Stem diameter at breast height, tree height, crown length, maximum crown width and crown width perpendicular to the maximum width were recorded for each sample tree. Statistics of the sample trees are given in Table 1. The sample trees were felled and 15 cm-thick stem discs were taken at breast height and at a height of 4 m. Along the south-north radius of each disc, two 3 cm-thick wedges were

Table 1. Statistics of the sample trees from control (C) and irrigated-fertilised (IL) plots, respectively. Twelve trees per treatment were used in the analysis. Values within brackets represent plot means in autumn 1996 based on Bergh et al. (1999) and an inventory of four replicated plots in autumn 1998 (Linder, unpubl.) Mean

STD

Min

Max

Diameter at breast height (cm)

C IL

9.4 (8.0) 14.8 (12.7)

1.6 (0.5) 2.8 (0.5)

6.6 10.8

11.7 17.9

Height (m)

C IL

6.6 (6.3) 8.9 (8.7)

1.0 (0.3) 0.7 (0.8)

5.2 7.9

8.1 10.1

Annual radial increment1 (mm)

C IL

2.0 (2.0) 4.1 (3.8)

0.5 (0.2) 1.3 (0.1)

0.93 1.34

Annual height increment1 (cm)

C IL

28.7 (29.2) 42.6 (47.9)

10.4 (2.9) 16.0 (5.5)

7.0 5.0

1 Mean

value for the treatment period 1987–1998

Holzforschung / Vol. 56 / 2002 / No. 5

3.46 6.73 63.0 79.0

H. Mäkinen et al.: Fibre Characteristics in Norway Spruce sawn through the discs. From the first wedge, annual radial increments were measured in two directions (south-north) based on computer aided tree-ring measurement equipment including a stereomicroscope connected to a video camera. At breast height, the sample trees had on average 21 annual rings at the time of the measurements. It was supposed that the last 12 annual rings formed during the fertilisation period represented mainly mature wood. At a height of 4 m on the stem, the sample trees had on average 10 annual rings, i.e. all rings were formed after the initiation of the experiment and represented juvenile wood. Beginning from the pith, earlywood zones of every third annual ring of the south radius of the wedges, from breast height and a height of 4 m, were split into small pieces and then macerated in glacial acetic acid and hydrogen peroxide solution (1:1, v/v) at 60 °C overnight (Franklin 1945). At breast height, latewood zones of the same rings of two fertilised and two control trees were also macerated. Suspensions of washed fibres were deposed on a microscope slide. The images (256 levels on a grey scale) were captured with a video camera (Cohu 4912; Cohu Inc., USA) attached to a light microscope (Olympus BH2; Olympus Optical Co., Japan). The resolution of captured images was 10.64 µm per pixel. The lengths of 50 unbroken fibres were measured with the help of an image analysis system (Image-Pro Plus; Media Cybernetics, USA). Measurements of different fibre properties on the individual trees at different stem heights are described in Table 2. In the earlywood, fibre diameters of the same rings were measured from the suspension by the means of Kajaani FiberLab (Kajaani Electronics Ltd, Kajaani, Finland). The number of measured fibre diameters per growth ring ranged between 784 and 3447, with an average of 2388. Since the FiberLab is not able to distinguish cut, curled and broken fibres or ray parenchyma elements from unbroken fibres, fibre lengths measured with the FiberLab were not used. For samples at breast height of 7 fertilised and 7 control trees, the same annual rings of the wedges were cut into 16 µm thick cross-sections on a cryo-microtome (–16 °C). The sections were stained with a 1% solution of safranin, immersed in a graded ethanol series, embedded with Canada balsam and mounted with a medium. Three parallel images from earlywood and latewood of each sample were captured using the equipment described above. In order to avoid irregular fibres around the boundary, the images were placed about five fibres away in the radial direction from the annual ring boundary, both in early- and latewood. Lumen diameter of an individual fibre was calculated as the average length of the diameters measured at two-degree-intervals joining two outline points of the lumen and passing through the centroid. They were measured from 7 fertilised

451 and 7 control trees using an objective lens with magnification of 20. Resolution of the captured images was 0.54 µm per pixel. The number of lumen diameters measured per image was on average 79, i.e. 237 per annual ring. Cell wall proportion from the whole image area was measured with a magnification of 40 and resolution of 0.27 µm per pixel. Cell wall proportion was only measured on the earlywood images. In earlywood and latewood, the thickness of double cell walls in the radial direction was measured with a magnification of 60 and resolution of 0.18 µm per pixel. The number of cell wall widths measured per image was on average 14, i.e. 42 per annual ring. In order to compare fertilised and non-fertilised trees of similar size, three older non-fertilised trees were sampled outside the experimental area. Fibre properties of the older trees were measured as described in Table 2. Statistical analyses All fibres of each annual ring were treated as individual observations, rather than as mean dimensions only. This approach made it possible to use all the information contained in the data set. Observations in the data have a hierarchical structure (block, plot, tree, stem height, annual ring, fibre), i.e. the individual observations belonging to the same batch are not independent of each other. In the mixed models, the mutual correlation structure of the dependent variable can be taken into account by allowing the parameters to vary randomly around the fixed population mean from one individual to another (e.g. Searle et al. 1992). Restricted maximum likelihood (REML) estimation in the MIXED procedure of SAS (SAS Institute, Inc. 1996) was used in estimating fixed and random parameters. The treatment effects on fibre characteristics were tested by using the model: Ybpthri = µ + τIL + βxbpth + ub + ubpt + εbpthri

(1)

where Ybpthri is a fibre characteristic (or its standard deviation) of fibre i in annual ring r assigned from block b, plot p, tree t and stem height h. The µ is the overall mean of the control plots, τIL is the effect of treatment IL, β is the regression coefficient, ub, ubp and ubpt are random effects for blocks, plots and trees, respectively, and εbpthri is random error. The initial differences between trees were removed by applying the arithmetic mean of each characteristic of the last annual ring formed before the treatment as a covariate (xbpth). At a height of 4 m, all annual rings were formed after the initiation of the treatment and no covariate was used. The data for three-year periods after the treatment (1 – 3, 4 – 6, etc.) were pooled at each stem height and the statistical tests were undertaken separately for each period.

Table 2. Number of sample trees used in the different measurements Property

Fibre length, earlywood Fibre length, earlywood Fibre length, latewood Fibre diameter, earlywood Fibre diameter, earlywood Cell wall thickness, earlywood Cell wall thickness, latewood Lumen diameter, earlywood Lumen diameter, latewood Cell wall %, earlywood

Stem height (m)

Control

Fertilised

Old non-fertilised

1.3 4.0 1.3 1.3 4.0 1.3 1.3 1.3 1.3 1.3

12 12 2 12 12 7 7 7 7 7

12 12 2 12 12 7 7 7 7 7

3 – – – – 2 – 2 – 2


452


To describe the profile of fibre characteristics with increasing age or distance from the pith, as well as to investigate random variation between blocks, plots and trees, the mixed models considered were: Ybpthri = β0 + β1zbpth + β2rbpthr + β3zbpth × rbpthr + β4τIL + ub + ubp + ubpt + βbrbpthr + βbprbpthr + βbptrbpthr + εbpthri

(2)

where β0, β1, β2, β3 and β4 are fixed regression coefficients, while ub, ubp, ubpt, βb, βbp and βbpt are random regression coefficients for blocks, plots and trees, respectively. The rbpthr is ring number (RN) or distance from the pith (DP), and zbpth is a dummy variable describing the stem height of 4 m. The τIL is the effect of treatment IL and always has a value 0 for the nonfertilised trees, as well as for the fertilised trees before the initiation of the treatment, but following the treatment fertilised trees are assigned a value of 1. To evaluate the model performance, the following error staˆ bpthri)/n), tistics were calculated: (i) mean error (E=∑(Ybpthri–Y ˆ bpthri|/n) and (iii) mean (ii) mean absolute error (|E|=∑|Ybpthri–Y ˆ bpthri)2/n) where Ybpthri is a meassquared error (E2=∑(Ybpthri–Y ˆ bpthri is a predicted observation and n is the ured observation, Y number of observations.

Results Fibre length Fertilisation increased radial growth, at breast height, threefold in sample trees (Fig. 1). In the earlywood of annual rings near the pith, fibre length was on average 1.2 mm, but increased rapidly outwards (Fig. 2A). The rate of increase slowed down, however, at 25–35 mm from the pith (Fig. 2B). The increased growth rate reduced fibre length at breast height, i.e. in mature wood. The fibres formed 10–12 years after the initiation of the experiment were on average 17% shorter in fertilised trees than in control trees (Fig. 2). This difference was statistically significant over the whole treatment period (Table 3). Furthermore, the difference between the fertilised and control trees was even more apparent when examined in relation to the distance from the pith, instead of ring number from the pith (Fig. 2B). Fibre length of the older non-fertilised trees from outside of the experiment area was rather similar to that of the younger control trees within the experiment (Fig. 2B). The difference in fibre length observed between the fertilised and control trees was also apparent between the fertilised trees and the older trees, i.e. between trees of similar size. At a stem height of 4 m, i.e. in juvenile wood, the difference in fibre length of earlywood between the fertilised and control trees was small when rings of the same age were compared. The fibres of the last-formed annual rings of the control trees were, however, again longer compared to the fertilised trees (Fig. 2C,D), but the difference was not statistically significant (Table 3). Within each annual ring, fibre length in earlywood was normally distributed (data not shown). Standard deviation of fibre length increased from the pith outwards, but the increase slowed down at 25–35 mm from the pith, i.e. at the same point as the increase in fibre Holzforschung / Vol. 56 / 2002 / No. 5

Fig. 1. Annual radial increments at breast height of the sample trees from control plots (continuous line) and irrigated-fertilised plots (dashed line). The initiation of the treatment in 1987 is indicated by a vertical line.

length (Fig. 2E-H). At stem heights of 1.3 m and 4 m, the difference in standard deviation of fibre length between the fertilised and control trees was mainly not statistically significant (Table 3). In earlywood, the relationship between fibre length and RN or DP was described by Eq. 2. Several alternative transformations of RN and DP were tested, but the logarithmic transformation was the best in terms of error statistics (E, |E| and E2) and consequently it was used (Table 4). The difference in fibre length between breast height and a height of 4 m was statistically significant, but small and in opposite directions in models 1, 2 and 3 (Table 4). Furthermore, at both heights, fibre length increased at the same rate from the pith outwards, i.e. β3 was not statistically significant. The dummy variable describing the fertilisation treatment (τIL) was statistically significant in all the models presented in Table 4. Its effect was, however, in some models in an opposite direction to what was found above, and it reduced the variance components and residuals of the models slightly. Its effect was also illogical and its importance low in several of the models used for the other fibre properties described below.Therefore, it was not included in the equations presented in Tables 4, 6 and 8 to 10. In addition, variance components describing average differences in fibre length between blocks and plots (ub and ubp) were not statistically significant; nor were the differences between blocks or plots in the increasing rate of fibre length from the pith outwards (βb and βbp). Thus, significant random variation was only found between individual trees (ubpt and βbpt, Table 4). They were, however, relatively small compared to the fixed parameter β2 and random variation εbpthri. Since no systematic trend in residuals of the models were found with respect to RN or DP (results not shown), the variation of fibre length, not explained by the fixed part of Eq. 2, was probably mainly intra-ring variation between individual fibres.


453

Fig. 2. Mean fibre length of the sample trees from control plots (continuous line) and irrigated-fertilised plots (dashed line) in earlywood at breast height (A, B) and at a height of 4 m (C, D) plotted against ring number and distance from the pith, as well as their standard deviations (E-H). For comparison, results from three older trees (thin line) of similar diameter as the fertilised trees are included (B).

Table 3. Tests on fibre length in earlywood at stem heights of 1.3 m and 4 m in the fertilised trees as compared to control trees.The numbers presented are values of variable τIL in Eq. 1. A negative value means that the fibres were shorter in the fertilised trees; p-values in parenthesis

Height

Years since the initiation of fertilisation 1–3 4–6 7–9 10–12

1.3 m

–0.200 (0.013)

Fibre length –0.322 –0.273 (0.000) (0.011)

–0.198 (0.093)

4m

–0.117 (0.370)

0.073 (0.495)

–0.017 (0.904)

1.3 m 4m

0.039 (0.757)

Standard deviation of fibre length –0.003 –0.041 –0.079 –0.043 (0.916) (0.138) (0.024) (0.290) 0.045 (0.235)

–0.004 (0.868)

0.078 (0.034)

–0.025 (0.437)

Latewood fibres were on average 11% longer than earlywood fibres. The length ratio of latewood and earlywood fibres (fibre length in latewood / fibre length in earlywood) was constant from pith to bark. Furthermore, the difference in fibre length between the fertilised and control trees was small in latewood (Fig. 3). The number of sample trees was, however, small with only two fertilised and two control trees.

Fibre diameter Fbre diameter in earlywood also increased from the pith outwards but the increase slowed down towards the bark (Fig. 4). At breast height, i.e. in mature wood, the fertilisation significantly increased fibre diameter when examined with respect to ring number, and the lastformed fibres of the fertilised trees were on average 12% wider (Fig. 4A, Table 5). However, the difference between the fertilised and control trees diminished when it was examined with respect to distance from the pith (Fig. 4B). At a stem height of 4 m, i.e. in juvenile wood, the difference between the fertilised and control trees was smaller, and it was only statistically significant in the annual rings formed 7–12 years after the initiation of the treatment (Table 5). Furthermore, fertilisation increased standard deviation of fibre diameter at both heights (Fig. 4, Table 5). Logarithmic transformation of RN and DP was used when their relationship to fibre diameter in earlywood was described by Eq. 2. As in the case of fibre length, a small difference was found between breast height and the height of 4 m (Table 6). The random coefficients describing blocks and plots (ub, ubp, βb and βbp) were not statistically significant.The random components for tree level (ubpt and βbpt) were statistically significant, but rather small compared to the unexplained variation (εbpthri). Since no systematic trends were found in residuals of the models with respect to RN or DP (results not shown), random variation was mainly intra-ring variation between individual fibres. Holzforschung / Vol. 56 / 2002 / No. 5

454


Table 4. Fixed and random regression coefficients, their standard errors in parenthesis and error statistics of the models (Eq. 2) for fiber length in earlywood as a function of ring number (RN) or distance from the pith (DP)

Intercept zbpth log(RN) log(DP) Variance components ubpt βbpt × log(RN) βbpt × log(DP) εbpthri Error statistics E |E| E2

Model 1

Model 2

Model 3

0.514 (0.046) 0.154 (0.009) 0.700 (0.031)

0.273 (0.054) –0.168 (0.008)

0.047 (0.031) 0.023 (0.031)

0.520

(0.028)

0.438 (0.050) 0.083 (0.019) 0.518 (0.034) 0.144 (0.034)

0.065

(0.031)

0.051

(0.031)

(0.031)

0.018 (0.031) 0.178 (0.031)

0.013 (0.031) 0.173 (0.031)

0.000 (0.031) 0.324 (0.031) 0.173 (0.031)

0.000 (0.031) 0.328 (0.031) 0.178 (0.031)

0.000 (0.031) 0.324 (0.031) 0.173 (0.031)

0.173

Fig. 3. Mean fibre length of the sample trees from control plots (continuous line) and irrigated-fertilised plots (dashed line) in latewood at breast height plotted against ring number (A) and distance from the pith (B).

Fig. 4. Mean fibre diameter of the sample trees from control plots (continuous line) and irrigated-fertilised plots (dashed line) in earlywood at breast height (A, B) and at a height of 4 m (C, D) plotted against ring number and distance from the pith, as well as their standard deviations (E-H). Holzforschung / Vol. 56 / 2002 / No. 5

H. Mäkinen et al.: Fibre Characteristics in Norway Spruce Table 5. Tests on fibre diameter in earlywood at stem heights of 1.3 m and 4 m in the fertilised trees as compared to control trees.The numbers presented are values of variable τIL in Eq. 1. A positive value means that the fibres were wider in the fertilised trees; p-values in parenthesis Years since the initiation of fertilisation 1–3 4–6 7–9 10–12 1.3 m

1.710 (0.016)

Fibre diameter 3.370 4.324 (0.000) (0.000)

4.744 (0.000)

4m

0.579 (0.805)

1.439 (0.357)

2.584 (0.000)

1.3 m 4m

3.918 (0.009)

Standard deviation of fibre diameter 2.704 4.590 4.283 4.522 (0.024) (0.002) (0.001) (0.000) 0.500 (0.855)

–0.490 (0.786)

3.387 (0.021)

2.459 (0.002)

455

Cell wall thickness, lumen diameter and cell wall proportion At breast height,cell wall thickness and lumen diameter in earlywood, as well as in latewood, increased from the pith outwards (Figs. 5 and 6). In both earlywood and latewood, the cell walls of the fertilised trees were thinner after the treatment initiation, but lumen diameters larger compared to the control trees when examined with respect to ring number. In most cases the difference was statistically significant (Table 7). Thus, the cell wall proportion of the fertilised trees was lower (Table 7,Fig.7).The difference in cell wall thickness, lumen diameter and cell wall proportion between the fertilised and control trees was, however, smaller when examined with respect to distance from the pith. Cell wall thickness, lumen diameter and wall proportion of the older unfertilised trees were similar to those of the younger control trees (Figs. 5, 6 and 7).

Table 6. Fixed and random regression coefficients, their standard errors in parenthesis and error statistics of the models (Eq. 2) for fibre diameter in earlywood as a function of ring number (RN) or distance from the pith (DP) Model 1 Intercept zbpth log(RN) log(DP) log(RN)*zbpth

23.418 (0.770) 3.297 (0.182) 5.320 (0.354) –0.226

Model 2 21.551 (0.722) 0.815 (0.096) 3.107

(0.121)

11.757

(0.031)

Model 3 21.619 (0.721) 0.935 (0.102) 1.559 (0.369) 2.936 (0.131)

(0.094)

Variance components ubpt βbpt × log(RN) βbpt × log(DP) εbpthri

255.931

(0.031)

4.053 (0.031) 255.700 (0.031)

2.389 (0.031) 255.675 (0.031)

Error statistics E |E| E2

0.000 (0.031) 9.755 (0.031) 255.931 (0.031)

0.000 (0.031) 9.739 (0.031) 255.931 (0.031)

0.000 (0.031) 3.735 (0.031) 255.675 (0.031)

13.093 (0.031) 2.710 (0.031)

11.714

(0.031)

Fig. 5. Mean cell wall thickness of the sample trees from control plots (continuous line) and irrigated-fertilised plots (dashed line) at breast height in earlywood (A, B) and in latewood (C, D) plotted against ring number and distance from the pith. For comparison, results from two older trees (thin line) of similar diameter as the fertilised trees are included (B). Holzforschung / Vol. 56 / 2002 / No. 5

456


Fig. 6. Mean fibre lumen diameter of the sample trees from control plots (continuous line) and irrigated-fertilised plots (dashed line) at breast height in earlywood (A, B) and in latewood (C, D) plotted against ring number and distance from the pith. For comparison, results from two older trees (thin line) of similar diameter as the fertilised trees are included (B).

Fig. 7. Mean cell wall proportion in the cross-sections of the sample trees from control plots (continuous line) and irrigated-fertilised plots (dashed line) of earlywood plotted against ring number (A) and distance from the pith (B). For comparison, results from two older trees (thin line) of similar diameter as the fertilised trees are included (B). Table 7. Tests on cell wall thickness, lumen diameter and wall proportion at breast height in the fertilised trees compared to the control trees. The numbers presented are values of variable τIL in Eq. 1. A positive value means that the fibre characteristic was larger in the fertilised trees; p-values in parenthesis Years since the initiation of fertilisation 1–3 4–6 7–9 10–12

As with the other fibre properties described above, cell wall thickness,lumen diameter and wall proportion in earlywood and latewood were closely related to ring number and distance from the pith (Tables 8 to 10). Distance from the pith better explained the variation in cell wall thickness, lumen diameter and wall proportion than ring number,but random variation between the trees was high.

Earlywood 1.3 m

–0.084 (0.715)

1.3 m

1.821 (0.097)

1.3 m

–3.043 (0.081)

Cell wall thickness –0.594 –0.560 (0.000) (0.001) Lumen diameter 2.391 4.547 (0.164) (0.005) Wall proportion –3.979 –5.177 (0.003) (0.000)

–0.414 (0.191) 4.158 (0.009) –5.650 (0.000)

Latewood 1.3 m

–0.859 (0.270)

1.3 m

2.107 (0.086)

Cell wall thickness –2.078 –1.356 (0.000) (0.001) Lumen diameter 3.832 3.316 (0.000) (0.000)


–1.208 (0.046) 3.342 (0.004)

Relationship between fibre properties Fibre properties were averaged for each individual annual ring, and correlations between them were calculated irrespective of ring age and fertilisation treatment.As could be expected on the basis of the results above, mean fibre length was positively correlated with fibre diameter, cell wall thickness and lumen diameter, but negatively correlated with cell wall proportion (Table 11). Accordingly, fibre diameter was positively correlated with cell wall thickness and lumen diameter, but negatively with cell wall proportion. Correlations between fibre properties were similar both in early- and latewood, excluding the relationship between cell wall thickness and


457

Table 8. Fixed and random regression coefficients, their standard errors in parenthesis and error statistics of the models (Eq. 2) for cell wall thickness as a function of ring number (RN) or distance from the pith (DP) Model 1 Earlywood Intercept log(RN) log(DP) Variance components ubpt βbpt × log(RN) βbpt × log(DP) εbpthri Error statistics E |E| E2 Latewood Intercept log(RN) log(DP) Variance components ubpt βbpt × log(RN) βbpt × log(DP) εbpthri Error statistics E |E| E2

2.524 (0.198) 0.322 (0.099)

0.540 (0.031) 0.635 (0.031)

Model 2

2.537

(0.176)

0.256

(0.073)

0.426

(0.031)

Model 3

2.258 (0.177) 1.610 (0.108) –0.913 (0.093) 0.428

(0.031)

0.073 (0.031) 0.286 (0.031)

0.034 (0.031) 0.280 (0.031)

0.000 (0.031) 0.416 (0.031) 0.284 (0.031)

0.000 0.418 0.286

(0.031) (0.031) (0.031)

0.000 0.413 0.280

5.463 (0.336) 0.319 (0.172)

5.457

(0.300)

0.272

(0.130)

1.227

(0.031)

0.284

(0.031)

1.539 (0.031) 0.406 (0.031) 1.686

(0.031)

0.000 (0.031) 1.005 (0.031) 1.686 (0.031)

(0.031) (0.031) (0.031)

5.097 (0.309) 2.061 (0.230) –1.222 (0.191) 1.282

(0.031)

0.233 (0.031) 1.692 (0.031)

0.119 (0.031) 1.675 (0.031)

0.000 1.005 1.687

0.000 1.004 1.675

(0.031) (0.031) (0.031)

(0.031) (0.031) (0.031)

Table 9. Fixed and random regression coefficients, their standard errors in parenthesis and error statistics of the models (Eq. 2) for lumen diameter as a function of ring number (RN) or distance from the pith (DP) Model 1 Earlywood Intercept log(RN) log(DP) Variance components ubpt βbpt × log(RN) βbpt × log(DP) εbpthri Error statistics E |E| E2 Latewood Intercept log(RN) log(DP) Variance components ubpt βbpt × log(RN) βbpt × log(DP) εbpthri Error statistics E |E| E2

6.533 (1.820) 8.032 (0.817)

46.143 (0.031) 9.307 (0.031)

Model 2

8.617

(0.889)

5.559

(0.301)

10.893

(0.031)

Model 3

7.489 (1.215) 6.713 (0.632) 0.675 (0.689) 20.336

(0.031)

(0.031)

1.246 (0.031) 25.359 (0.031)

2.728 (0.031) 25.218 (0.031)

0.000 (0.031) 3.826 (0.031) 25.231 (0.031)

0.000 (0.031) 3.826 (0.031) 25.328 (0.031)

0.000 (0.031) 3.820 (0.031) 25.185 (0.031)

25.263

3.022 (0.779) 2.315 (0.395)

8.433 (0.031) 2.178 (0.031)

3.580

(0.470)

1.596

(0.194)

3.051

(0.031)

4.523 (0.770) –5.561 (0.319) 5.635 (0.314) 8.212

(0.031)

(0.031)

0.521 (0.031) 6.975 (0.031)

0.624 (0.031) 6.886 (0.031)

0.000 (0.031) 2.038 (0.031) 7.055 (0.031)

0.000 (0.031) 2.022 (0.031) 6.975 (0.031)

0.000 (0.031) 1.995 (0.031) 6.886 (0.031)

7.055


458


Table 10. Fixed and random regression coefficients, their standard errors in parenthesis and error statistics of the models (Eq. 2) for wall proportion in earlywood as a function of ring number (RN) or distance from the pith (DP) Model 1 Intercept log(RN) log(DP)

Model 2

43.490 (3.528) –6.443 (1.434)

Variance components ubpt βbpt × log(RN) βbpt × log(DP) εbpthri

168.779 (0.031) 27.840 (0.031)

41.267

(2.235)

–4.287

(0.700)

39.976 (1.853) 7.167 (2.232) –9.470 (1.705)

65.767

(0.031)

41.553

(0.031)

(0.031)

6.409 (0.031) 6.247 (0.031)

3.813 (0.031) 6.370 (0.031)

0.000 (0.031) 1.924 (0.031) 6.016 (0.031)

0.000 (0.031) 1.865 (0.031) 5.660 (0.031)

0.000 (0.031) 1.878 (0.031) 5.777 (0.031)

6.658

Error statistics E |E| E2

Model 3

Table 11. Correlation of the mean fibre characteristics in earlywood and latewood; correlations above diagonal and their p values below diagonal

Fibre lengthearly Fibre widthearly Wall thicknessearly Wall thicknesslate Lumen diameterearly Lumen diameterlate Wall proportionearly

Fibre lengthearly

Fibre widthearly

Wall thicknessearly

Wall thicknesslate

Lumen diameterearly

Lumen diameterlate

Wall proportionearly

– 0.000 0.000 0.137 0.000 0.000 0.002

0.609 – 0.033 0.711 0.000 0.000 0.000

0.649 0.229 – 0.000 0.000 0.691 0.812

0.156 0.041 0.562 – 0.502 0.001 0.113

0.647 0.872 0.368 0.071 – 0.000 0.000

0.495 0.717 0.041 –0.342 0.739 – 0.000

–0.301 –0.736 0.023 0.167 –0.764 –0.732 –

lumen diameter. In earlywood, wall thickness increased with increasing lumen diameter, but in latewood wall thickness and lumen diameter were negatively correlated. In general, correlations between fibre properties were surprisingly high even though fibre lengths and diameters were not measured from the same fibres as cell wall and lumen properties. Discussion Fibre length followed the well-known age trend of conifers caused by maturation of the cambium (Dinwoodie 1961; Olesen 1982; Frimpong-Mensah 1987; Kucera 1994), i.e. fibres were short near the pith and their length increased with decreasing rate from the pith outwards. A similar non-linear trend from pith to bark was found both in the control and fertilised trees, but a faster growth rate led to the formation of shorter fibres.This is in agreement with earlier studies in different spruce species, where a negative correlation between ring width and fibre length has been reported (Stairs et al. 1966;Yang and Hazenberg 1994; Herman et al. 1998).A decrease in fibre length with increasing growth rate has also been found in Thuja occidentalis (Bannan 1960). As in the case of fibre length, the fibre diameter, cell wall thickness and lumen diameter increased from the Holzforschung / Vol. 56 / 2002 / No. 5

pith outwards both in the control and fertilised trees. Fertilised trees with wider annual rings had wider fibres and lumen diameters, but thinner cell walls than the control trees. This confirms earlier observations by Ollinmaa (1959), Denne (1973) and Atmer and Thörnqvist (1982) who found that mean fibre diameter in Picea abies and P. sitchensis was correlated with ring width and rate of shoot elongation. If fibre length is excluded, differences in fibre characteristics between the fertilised and control trees were less apparent when examined with respect to distance from the pith instead of ring number from the pith. Similar results have been reported by Olesen (1977, 1982) who found that the fibre diameter of Norway spruce was more strongly correlated with distance from the pith than with ring number. Therefore, fibre dimensions are rather determined by the number of anticlinal cell divisions taking place in the cambium than by cambium age, i.e. the maturation of the cambium is related to its activity (cf. Larson 1969). Average fibre properties in latewood differed from those in earlywood, where earlywood fibres were shorter than those of latewood. Most of the variation in cell wall thickness, lumen diameter and radial diameter of fibres was related to variation between earlywood and latewood, which is in agreement with earlier studies (cf. Stairs et al. 1966; Olesen 1977; Saranpää 1985). The rate


of change from the pith to the bark and the effect of fertilisation was, however, similar in early- and latewood. Since the samples were separately prepared from the earlywood and latewood zones of annual rings, determination of fibre properties in the whole wood material based on such data is affected by the variation in latewood content, which in Norway spruce decreases with increasing growth rate. Small differences in average fibre dimensions were found between different stem heights, which confirms earlier reports in which the fibre dimensions of Norway spruce changed only slightly at different heights along the stem (cf. Dinwoodie 1961; Stern 1963; Saranpää 1994).Average fibre length and diameter was less affected by enhanced growth rate in juvenile wood (at a stem height of 4 m) than in mature wood (at breast height). This result is consistent with studies in which young, intensivly cultured poplars were compared with older material from natural stands (e.g. Snook et al. 1986; DeBell et al. 1998). At least some part of the smaller effect in juvenile wood may be explained by the different rate of cell division and maturation in juvenile and mature wood (cf. Panshin and de Zeeuw 1980). The effect of fertilisation on fibre extension during their maturation could, therefore, be masked by the already rapid rate of cell division in juvenile wood. Intra-ring variation of fibre characteristics is strongly influenced by the occurrence of juvenile wood (Larson 1969). In the present study, the standard deviation of fibre length and diameter increased non-linearly with increasing ring number and distance from the pith. Thus, the range of intra-ring variation of fibre properties was narrower in juvenile wood than in mature wood (cf. Herman et al. 1998). Intra-ring variation increased, however, faster from the pith outwards in juvenile wood than in outer rings. The rapid growth rate of the fertilised trees did not significantly change the standard deviation of fibre length. On the contrary, intra-ring variation of fibre diameter at breast height and at a height of 4 m was increased by fertilisation. Random tree components were statistically significant in all the models for fibre properties. In addition, relatively large differences between trees were found in the rate of change from the pith outwards of fibre diameter, cell wall thickness, lumen diameter and wall proportion. However, intra-ring variations of fibre length and fibre diameter were high compared to random inter-tree variation (cf. Dinwoodie 1961; Sirviö and Kärenlampi 2000). It was shown that it is possible to model intra-tree variability of fibre characteristics using ring width and cambial age as independent variables. The models presented are, however, limited by the relatively young age of the sample trees. Future studies based on a larger amount of material from different geographical locations are necessary in order to develop more general models. Sampling more than two stem heights is also necessary to describe the effect of location along the stem more accurately.

459

Conclusions Fertilisation affected fibre properties mainly through the faster growth rate. Inter-ring variation of the fibre dimensions was characterised by the simultaneous increase in their average values and standard deviations from pith to bark. Faster growth rate reduced mean fibre length and cell wall thickness but increased mean fibre and lumen diameter when the fibre properties were examined in respect to ring number from the pith. However, fibre and lumen diameter, as well as cell wall thickness, were closely related to distance from the pith. Only the difference in fibre length between the fertilised and control trees was more apparent with respect to distance over ring number from the pith. In addition, the effect of enhanced growth rate was less apparent in juvenile than in mature wood. The changes in fibre properties could be similar in response to other silvicultural practices which enhance tree growth. For example, other studies on Norway spruce have demonstrated that rapid growth rate, caused by low stand density, resulted in similar changes in fibre characteristics as those observed in the present study (e.g. Herman et al. 1998). Therefore, increasing growth rate of trees probably causes similar kinds of changes in fibre characteristics irrespective of the factor promoting enhanced growth. In most forestry practices, stands are normally fertilised at an older age and only one or two nutrient elements are added. Therefore, the results presented cannot be used to predict the absolute change in fibre characteristics caused by fertilisation in Norway spruce stands. The results rather demonstrate the effects of growth rate on fibre properties and the magnitude of potential change and direction that could be caused by optimising the nutritional status of Norway spruce stands. Acknowledgements The study was made possible by financial support from The Academy of Finland and The Swedish National Energy Administration (STEM). We are greatly indebted to Dr. Antero Varhimo (KCL) for the measurements with the help of FiberLab, and Santtu Hietala, Satu Järvinen, Tapio Järvinen, Irmeli Luovula, Tapio Nevalainen, Carl Räihä, Bengt-Olov Wigren and Ulla Nylander for skilful technical assistance.

References Atmer, B. and T. Thörnqvist. 1982. The properties of tracheids in spruce (Picea abies) and pine (Pinus silvestris). Swedish University of Agricultural Sciences, Dept. of Forest Products 134, 1–59. (In Swedish, English summary). Bannan, M.W. 1960. Cambial behavior with reference to cell length and ring width in Thuja occidentalis L. Can. J. Bot. 38, 177–183. Bergh, J., S. Linder, T. Lundmark and B. Elfving. 1999. The effect of water and nutrient availability on the productivity of Norway spruce in northern and southern Sweden. For. Ecol. Manage. 119, 51–62. Bergqvist, G., U. Bergsten and B. Ahlqvist. 2000. Fibre properties of Norway spruce of different growth rates grown unHolzforschung / Vol. 56 / 2002 / No. 5

460 der birch shelterwoods of two densities. Can. J. For. Res. 30, 487–494. DeBell, J.D., B.L. Gartner and D.S. DeBell. 1998. Fiber length in young hybrid Populus stems grown at extremely different rates. Can. J. For. Res. 28, 603–608. Denne, M.P. 1973. Tracheid dimensions in relation to shoot vigor in Picea. Forestry 46, 117–124. Dinwoodie, J.M. 1961. Tracheid and fibre length in timber: A review of literature. Forestry 34, 125–144. Dutilleul, P., M. Herman and T. Avella-Shaw. 1998. Growth rate effects on correlations among ring width, wood density, and mean tracheid length in Norway spruce (Picea abies). Can. J. For. Res. 28, 56–68. Franklin, G.L. 1945. Preparation of thin sections of synthetic resins and wood-resin composites, and a new macerating method for wood. Nature 155, 51. Frimpong-Mensah, K. 1987. Fiber length and basic density variation in the wood of Norway spruce (Picea abies (L.) Karst.) from northern Norway. Commun. Norweg. For. Res. Inst. 40, 1–25. Hägglund, B. and J.-E. Lundmark. 1977. Site index estimation by means of site properties of Scots pine and Norway spruce in Sweden. Stud. For. Suec. 138, 1–33. Helander, A.B. 1933. Variations in tracheid length of pine and spruce. Foundation for Forest Products Research of Finland. Publ. 14, 75 pp. (In Finnish, English summary.) Herman, M., P. Dutilleul and T. Avella-Shaw. 1998. Intra-ring and inter-ring variations of tracheid length in fast-grown versus slow-grown Norway spruces (Picea abies). IAWA J. 19, 3–23. Kucera, B. 1994. A hypothesis relating current annual height increment to juvenile wood formation in Norway spruce. Wood Fiber Sci. 26, 152–167. Larson, P.R. 1969. Wood formation and the concept of wood quality. Yale University, School of Forestry Bulletin 74, 1–54. Linder, S. 1995. Foliar analysis for detecting and correcting nutrient imbalances in Norway spruce. Ecol. Bull. (Copenhagen) 44, 178–190. Linder, S. and J.G.K. Flower-Ellis. 1992. Environmental and physiological constraints to forest yield. In: Responses of Forest Ecosystems to Environmental Changes. Eds. A. Teller, P. Mathy, J.N.R. Jeffers. Elsevier Applied Sciences, London. pp. 149–164. Lindström, H. 1997. Fiber length, tracheid diameter, and latewood percentage in Norway spruce: Development from pith outwards. Wood Fiber Sci. 29, 21–34. Mäkinen, H., P. Saranpää and S. Linder. 2001. Effect of nutrient optimization on branch characteristics in Picea abies (L.) Karst. Scand. J. For. Res. 16, 354–362. Olesen, P.O. 1977. The variation of the basic density level and tracheid width within the juvenile and mature wood of Norway spruce. For. Tree. Improv. 12, 1–21. Olesen, P.O. 1982. The effect of cyclophysis on tracheid width and basic density in Norway spruce. For. Tree Improv. 15, 1–80. Ollinmaa, P.J. 1959. Study on reaction wood.Acta For. Fenn. 72, 1–54. (In Finnish, English summary).


H. Mäkinen et al.: Fibre Characteristics in Norway Spruce Panshin, A.J. and C. de Zeeuw. 1980. Textbook of Wood Technology. 4th ed. McGraw-Hill, New York. 722 pp. Saranpää, P. 1985. Length, diameter and cell wall thickness of tracheids in mature lodgepole pine bolewood. Silva Fenn. 19, 21–32. (In Finnish, English summary) Saranpää, P. 1994. Basic density, longitudinal shrinkage and tracheid length of juvenile wood of Picea abies (L.) Karst. Scand. J. For. Res. 9, 68–74. SAS Institute Inc., 1996. SAS/STAT Software: Changes and Enhancements through Release 6.11. SAS Institute Inc., Cary, NC. Searle, S.R., G. Casella and C.E. McGulloch. 1992. Variance Components. John Wiley & Sons, New York. 501 pp. Sirviö, J. and P. Kärenlampi. 2000. Two scales of variation in Norway spruce tracheid properties. Wood Fiber Sci. 32, 311–331. Snook, S.K., P.L. Labosky Jr.,T.W. Bowersox and P.R. Blankenhorn. 1986. Pulp and papermaking properties of a hybrid poplar clone grown under four management strategies and two soil sites. Wood Fiber Sci. 18, 157–167. Stairs, G.R., R. Marton, A.F. Brown, M. Rizzio and A. Petrik. 1966. Anatomical and pulping properties of fast- and slowgrown Norway spruce. Tappi J. 49, 296–300. Stern, K. 1963. Einfluß der Höhe am Stamm auf die Verteilung der Raumdichte des Holzes in Fichtenbeständen. Holzforschung 17, 6–12. Tyrväinen, J. 1995. Wood and fiber properties of Norway spruce and its suitability for thermomechanical pulping. Acta For. Fenn. 249, 1–155. Yang, K.C. and G. Hazenberg. 1994. Impact of spacing on tracheid length, relative density, and growth rate of juvenile wood and mature wood in Picea mariana. Can. J. For. Res. 24, 996–1007. Zobel, B.J. and J.P. van Buijtenen. 1989. Wood Variation. Its Causes and Control. Springer, Berlin. 363 pp. Received April 23rd 2001

Dr. Harri Mäkinen Finnish Forest Research Institute P.O. Box 18 01301 Vantaa Finland E-mail: [email protected] Dr. Pekka Saranpää Finnish Forest Research Institute P.O. Box 18 01301 Vantaa Finland Prof. Dr. Sune Linder Swedish University of Agricultural Sciences Department for Production Ecology P.O. Box 7042 75007 Uppsala Sweden