Tree Physiology 14,49-62 0 1994 Heron Publishing-Victoria,

Canada

Shapeof tree stems- a re-examination of the uniform stress hypothesis

JOHN MORGAN and MELVIN G. R. CANNELL*
Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

1 Department of Civil Engineering and Building Science, University of Edinburgh, Edinburgh EH9 3JL, Scotland, UK 2 Institute of Terrestrial Ecology, Bush Estate, Penicuik, Midlothian EH26 OQB, Scotland, UK Received April 8,1993 Summary The transfer matrix method of structural analysis was used to examine the hypothesis that tree stems grow to a shape that tends to equalize the average bending plus axial stressesto which they are subjected along their length. The method and computational procedures were checked by comparing computed heightdiameter profiles with those calculated using elementary stress theory for trees with simple force distributions in the crown. Measured height-diameter profiles for trees were then taken from the literature and shown to be well-fitted by profiles calculated to give uniform stress along the stems, using the most realistic average forces and force distributions within the crowns. At high wind speeds, the height-diameter profile giving uniform stress was more tapered than the profile giving uniform stress at low wind speeds. The profile giving uniform stress was similar over the normal range of average wind speeds of 2.5 to 10.0 m SC (at the top of the canopy). But a tree that had grown to give uniform stress along its stem in an average wind of 5 m s- showed markedly decreased stress with height at wind speeds of about 15 m s-t or more, and increased stress with height (to the crown base) at wind speeds of about 1.25 m SC or less. The fact that tree stems develop shapes in response to average conditions, but show varying stress distributions in extreme conditions, may help to explain some of the apparent evidence for non-uniform stress distribution in the literature. In general, our analysis supports the above hypothesis for the stem region above the butt swell. Keywords: axial stress, bending stress, force, structural analysis, transfer matrix method, wind speed.

Introduction
When trees bend in the wind, they experience maximum bending stress at the outer surface of their stems, close to the cambium. One of the most enduring theories in tree biology is that the cambium produces new wood in such a way as to equalize the distribution of stress along the outer surface of the stem (Metzger 1893, Larson 1965, Assmann 1970, Wilson and Archer 1979, Mattheck 1990, 1991). In other words, cambial growth tends to be greatest in regions of highest stress, and least in areas of lowest stress. This theory is sometimes referred to as the mechanistic theory of constant stress: but it is more exact to call it the mechanical theory of uniform stress because the stress distribution is not constant over time (or constantly uniform) when trees are subject to variable wind forces. There is some difference of opinion about the validity or universality of the uniform stress hypothesis. Early foresters observed that tree stems assume a shape that gives approximately uniform stress because the stem taper below the crown of

50

MORGAN

AND

CANNELL

many trees approximates a cubic paraboloid (Assman 1970). Others have been willing to accept the uniform stress hypothesis (Petty and Swain 1985, Dean and Long 1986) and Mattheck (1990, 1991) has assumed that it applies over the entire woody structure of trees. However, West et al. (1989) found that the stress distribution along the stems of Eucalyptus regnans F. Muell. depended on the wind velocity gradient within the canopy. In another analysis, Milne and Blackburn (1989) found that Picea sitchensis (Bong.) Can: stems tended to have a region of maximum stress, which occurred closest to the ground in stems with the greatest taper, and which coincided with the region where stem breakage often occurs in strong winds. Experimental and observational studies lend considerable support for the view that cambial growth is stimulated by continual high mechanical stress (force/area) or possibly, the associated strain (fractional change in length) (Zimmermann and Brown 1971). Most notably, several tree bending and guying experiments have shown that cambial growth is greatest where most bending occurs (Jacobs 1954, Larson 1965, Valinger 1990). The ways in which mechanical forces are translated into increased cell growth are unknown, although piezoelectric forces in crystalline substances, such as cellulose, and a redistribution of growth hormones may be involved (Jaffe 1973, Mitchell et al. 1977, Hunt and Jaffe 1980, Worrall 1980). In this paper, we used the transfer matrix method of structural analysis, previously used to examine the growth of branches (Morgan and Cannel1 1987, Cannel1 and Morgan 1988, 1989, 1990), to examine the evidence for uniform stress along tree stems. We offer a slightly modified hypothesis (see below) and show that some of the apparently conflicting evidence in the literature can be resolved by realizing that: (1) the profile of stem diameter with height is a poor test of the uniform stress hypothesis because bending stress is inversely related to the cube of stem diameter-consequently small differences in measured diameters associated with trunk irregularities give large differences in calculated stress; and (2) the distribution of stress along a stem depends on the wind force; the stress may be uniformly distributed in a steady wind close to the average wind speed, but will not be uniformly distributed at high or low wind speeds.

Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

Hypothesis If tree stems thicken preferentially each year in regions where mechanical weakness occurs (i.e., in areas of highest bending plus axial stresses), the stem shapes (heightdiameter profiles) observed will be the time-averaged response to the stresses produced by variable forces during the life of the tree. In other words, tree stems grow to a shape that tends to equalize the average bending plus axial stresses to which they are subjected along their length. The average bending plus axial stress will depend on the history of the tree crown and the average wind speed over the life of the tree.

SHAPE

OF TREE

STEMS-UNIFORM

STRESS

HYPOTHESIS

51

Method Elementary theory

Tree stems may be considered to be cantilever beams fixed at a point above the region of butt swell. Elementary theory states that the longitudinal stress, (3, in beams of circular cross section with bending moment, M, is given by:

Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

where d is the beam diameter (Timoshenko 1953, provided that beam diameter is much less than beam length, and the deflection is small. The height-diameter profile of tree stems having uniformly distributed stress (0) below and within the crown can be calculated using this elementary theory, in which the horizontal (wind) force, F, is distributed within the crown in a simple way. Consider the two simple cases in Figure 1. Equation 1 implies that, for stress to be uniformly distributed along the stem, the cube of the diameter at any height must be proportional to the distance from that height (h) to the center of pressure of the horizontal wind force on the crown (II,). That is, to give uniform stress along the

0.0

0.2

0.4 Relative

0.6 Diameter

0.8

1.0

Figure 1. Theoretical (lines) and calculated (points) height-diameter profiles of stems with uniform stress along their length. (a) A uniform distribution of wind force, and (b) a triangular distribution of wind force. The theoretical relationships (lines) were based on text Equations 2 and 4 for case (a), and Equations 2 and 6 for case (b). The points were calculated independently using the transfer matrix method. H = tree height; H, = length of crown, h = height, h, = height within crown, H, = height to the center of pressure from the wind force, F, acting at point g (see text).

52

MORGAN

AND

CANNELL

stem below the crown:

and
Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

M=F(Hg-h) therefore d3 = (Hg - h) .

(2)

For case (a) in Figure 1, the bending moment within the crown, which has a uniformly distributed force, is given by:
M=&(Hc-hc)2, c

(3)

where F is the force on the crown, H, is the length of the crown, and h, is the height above the base of the crown. It follows that, to give uniform stress along the stem within the crown in case (a): d3 0~ (Hc - /z,)~ or $/2 oc (Hc - hc) . For case (b) in Figure 1, the bending moment within the crown, which in this instance has a triangular distribution of force, is given by:

M=F3H2 WC- Q3 .
C

It follows that, to give equal stress along the stem within the crown in case (b): d3 0~ (H, - Q3 or d= (H,-A,). Equations 2, 4 and 6 were used to calculate the theoretical tree height-diameter profiles assuming that the crown occupied the top third of the tree (Hc = H/3). These (6)

SHAPE

OF TREE

STEMS-UNIFORM

STRESS

HYPOTHESIS

53

relationships are shown by the continuous lines in Figure 1, in which heights are given relative to tree height (H) and diameters are given relative to the diameter at the base of the tree (ignoring butt swell). Transfer matrix method Previously, we showed how the distribution of forces within cantilever beams of complex shape, with small or large deflections, can be calculated by dividing the beam into small sections and using a matrix equation to relate the conditions at the ends of each section (shear force, V, bending moment, M, angle, and deflection) to the distributed load, p (or any other load), given the length, diameter and Youngs modulus of the material (Figure 2) (Morgan and Cannel1 1987, Morgan 1989, see also Milne and Blackburn 1989). In studies on branches, we used iterative procedures to calculate the diameters required to support self weights, with given deflections, angles, taper and Youngs moduli (Cannel1 et al. 1988, Cannel1 and Morgan 1990). It is equally possible to specify the diameters and to calculate the resulting distribution of forces, or to specify that stress is uniform along the length of the beam with given forces and to calculate the resulting diameter-length profile. Our computer program was checked by determining the diameter-height profiles under uniform stress conditions for the two simple cases shown in Figure 1. The calculated values are shown as points and give an exact fit to the theoretical expectation (Figure 1).
Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

Figure 2. A section of a deflected stem showing the forces acting upon it. V = shear force, N = axial force, M = bending moment, W = weight, p = distributed force, 8 = angle. Subscripts L and R refer to left and right. (See also Morgan and Cannel1 1987, Morgan 1989).

54

MORGAN

AND

CANNELL

In the analyses reported below, the vertical force distribution on the crowns was taken directly from published values, or was assumed to be triangular, approximating the leaf area density distribution of many tree species (e.g., Landsberg and James 1971, Kellomaki et al. 1980, Massman 1982, West et al. 1989, Wang et al. 1990). Also, the effects of self weight, W, and of left and right axial forces, NL and NR, respectively, were incorporated into the transfer matrix equations as described by Morgan (1989) (Figure 2). Self weight affects the stress distribution in two ways: first, as the stem deflects, it causes an additional bending moment along the stem; and second, it produces an axial force (compression along the axis of the stem) and, therefore, an axial stress. The axial stress is generally small compared with the bending stress, but it does mean that the total bending at any point along a stem is no longer exactly proportional to d3 (Equations 1 and 2), but is proportional to the diameter to a power greater than three. Also, when self-weight is included, the stem shape giving uniform stress depends on the extent to which the stem is deflected from vertical, and hence, on the average force due to the wind.

Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

Measured and calculated height-diameter

profiles

Other workers have measured the height-diameter profiles of tree stems, estimated the force distribution on the crowns and then calculated the distribution of stress in the stems (West et al. 1989, Milne and Blackbum 1989). Our approach was to use the force distribution on the crowns to calculate the height-diameter profile that gave a uniform stress distribution. We then compared the measured height-diameter profiles with those calculated for uniform stress (Figures 3 and 4).

Ten trees of Picea sitchensis (Gardiner 1989)

The points in Figure 3 are stem diameters at different heights of ten trees of P. sitchensis, reproduced from Gardiner (1989), normalized to a breast height diameter of 1.5 m (i.e., the diameter at breast height is 1.0 in Figure 3). The trees were growing in Scotland at 3600 stems ha-, and averaged 15 m in height, 0.165 m in breast height diameter, 162 kg stem fresh weight, 50 kg branch fresh weight, with Youngs modulus 6.6 GPa, and a drag force of 1587 N at 20 m s- wind speed. The transfer matrix method was used to calculate the height-diameter profile giving uniform stress based on three assumptions. First, the crown was assumed to be distributed over the top half of the tree, as suggested by other measurements (Coutts 1986). Secondly, the vertical distribution of the force was assumed to be triangular, as shown in Figure 3, approximating the near-normal distribution of leaf area within crowns of 19 si/clrensis of similar age, as mcasurcd !)y Norman and Jarvis (1974). Thirdly, it was assumed that the average wind speed experienced by the trees was 5 m s-l, based on wind speed averages for the UK uplands (Shellard 1976), and that drag was proportional to the square of wind speed. The continuous line in Figure 3 shows the calculated height-diameter profile for the condition of uniform stress (bending plus axial stress) along the stem. There was

SHAPE OF TREE STEMS-UNIFORM

0.6

1.0 ... K..

STRESS HYPOTHESIS

55

Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

E .F
r B ; I d

Within crown
H

Relative

Diameter

Figure 3. Measured (points) and calculated (line) profiles of relative stem diameter (diameter at breast height = l.O), and relative stem height for Pica sitchensis. The measured values are for 10 trees, from Gardiner (1989). The calculations were made using the transfer matrix method for the condition of uniform stress along the stem, using the horizontal (wind) distribution shown in the inset diagram, a wind speed of 5 m SC, and the weight distribution and other tree properties given by Gardiner (1989).

SD [

(a) Eucalyptus

regnans

(b) Pica

sitchensls

25 -

12

16

20

24

Stem diameter

(cm)

Figure 4. Measured (points) and calculated (lines) height-diameter profiles (above butt swell) for single trees of (a) Eucalyptusregnans and (b) Piceasitchensis. The measured values are from West et al. (1989, their Figure la) and Milne and Blackbum (1989, their Figure 4). The calculated lines were derived using the transfer matrix method for the condition of uniform stress up the stems, using the horizontal force distributions shown in the inset diagrams and other conditions as stated in the text. The open points are diameters in the region of butt swell.

56

MORGAN

AND

CANNELL

good agreement between the measured height-diameter calculated for uniform stress. Single tree of Eucalyptus regnans (West et al. 1989)

profile and the profile

The points in Figure 4a are stem diameters at different heights within a single dominant 20-year-old tree of E. regnans growing in a thinned stand in Tasmania (West et al. 1989, their Figure la). West et al. (1989) also reported the vertical distribution of crown and stem fresh weights (their Figure lb), and the vertical distribution of force on the crown (their Figure 3). The latter was used directly in the transfer matrix calculations, taking the force distribution illustrated in Figure 4a (with their kr = 0.048, kZ = 0.021, see West et al. 1989). The overall force on the crown was 20 kg (about 200 N). The continuous line in Figure 4a shows the calculated height-diameter profile for the condition of uniform stress (bending plus axial stress), which deviated from the measured values only in the region of butt swell. Different basal stem diameters were tried, ignoring the region of butt swell, and the line shown is that giving the best fit of calculated to measured values. Single tree of Picea sitchensis (Milne and Blackburn 1989) The points in Figure 4b are the stem diameters at different heights within a single 22-year-old tree of P. sitchensis growing in a stand at 3800 stems ha- in Scotland (Milne and Blackbum 1989, their Figure 4). The wind drag forces on l-m thick layers of foliage within the canopy were supplied by Mime (personal communication) and were based on measured leaf areas, drag coefficients given by Landsberg and James (197 l), a wind speed of 15 m s- at the top of the canopy and a measured wind profile within the canopy (Milne and Blackbum 1989). Figure 4b shows the distribution of force on the crown. The continuous line in Figure 4b shows the calculated height-diameter profile for the condition of uniform stress (bending plus axial stress), which, as for E. regnans, deviated from the measured values only in the region of butt well. As for E. regnans, different basal stem diameters were tried, ignoring the region of butt swell. The optimal basal diameter was 20.4 cm which corresponded to a uniform stress of 0.34 MPa (see below). Sensitivity qf the test of equal stress to measured diameters Because bending stress is inversely proportional to the cube of stem diameter (Equation 1) (and bending plus axial stress is inversely proportional to the diameter to a power greater than three), a small change in diameter brings about a large change in stress. Conversely, many different stress distributions lead (in the transfer matrix method) to very similar predictions for the height-diameter profile. Consequently, the goodness of fit of the lines to measured diameters in Figures 3 and 4 was not a sensitive test of the hypothesis of uniform stress distribution in stems. Figure 5 shows the calculated height-diameter profile for a tree of P. sitchensis, with the same force parameters as for Figure 3 (Gardiner 1989), with either uniform

Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

SHAPE OF TREE STEMS-UNIFORM

STRESS HYPOTHESIS

57

1.0

r
Height

Lb!-

-10%

Equal +10x rtrerr

stress

Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

0.2 -

0.2

0.4 Relative

0.6 diameter

0.6

1.0

Figure 5. Calculated relative height-diameter profiles of stems of P. sitchensis assuming uniform stress along the stem (open squares and line) or allowing stress to increase or decrease by 10% from top to bottom. The solid points refer to a 10% increase in stress (thinner stem), star points refer to a 10% decrease in stress (thicker stem). Calculations were made assuming the same forces, with stem and tree properties as in Figure 3 (Gardiner 1989).

stress, or with stress increasing or decreasing by 10% from the top to the bottom of the tree. The 10% change in stress caused a change of only 5% in relative stem diameter. Measurements of diameters along single stems often vary by 5%, owing to eccentricity, or irregularities in the bark, as shown by the scatter of points in Figures 4a and 4b. Thus, the stress values calculated by West et al. (1989) and Milne and Blackbum (1989) from measured diameters may be subject to an error of about 10%. Conversely, the apparently good fit of the lines for uniform stress to measured diameters in Figures 3 and 4 provided evidence that stress was only approximately uniform (i.e., to within 10%).

Wind speeds and stress distribution

Height-diameter

in tree stems

profiles at different wind speeds

The transfer matrix method can be used to calculate the height-diameter profiles that give uniform stress in trees subject to different average forces corresponding to different average wind speeds. Figure 6 shows the profiles for such trees of P. sitchensis, based on the same assumptions as for Figure 3 (Gardiner 1989), and assuming drag was proportional to the square of the wind speed. The most notable result was that the height-diameter profile was not sensitive to a change in average wind speed over the range 2.5 to 10.0 m s-l. Many of the worlds forests will experience average wind speeds over their lifetime within this 4-fold range, as will trees growing at different spacings. Thus, it would appear that differences in crown dimensions and properties that affect the vertical force distribu-

MORGAN

AND

CANNELL

Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

0.0

0.2 Relative

0.4 diameter

0.6

0.8

1.0

Figure 6. Calculated relative height-diameter profiles of stems of P. sitchensis assuming uniform stress along the stem with different constant horizontal wind speeds. Calculations were made assuming the same tree properties as in Figure 3 (Gardiner 1989). The profiles giving uniform stress at high wind speeds signify greater stem taper. The profiles giving uniform stress at wind speeds 2.5 to 10.0 m s-r are sufficiently similar to be represented by one thick line.

tion probably have a greater impact on stem taper than differences in average wind speeds per se. At high wind speeds (20 m s-l), the height-diameter profile giving uniform stress was more tapered (Figure 6). At such high wind speeds, the stem deflected more than 50% of its length, which resulted in a relative reduction in the bending moment along the stem, permitting a relative decrease in diameter with uniform stress. At low wind speeds (< 5 m SC) the opposite was true, and the diameter required to maintain uniform stress changed relatively little with increase in height. As the wind speed decreased, the bending moment, and therefore, the tensile stress, became small and the axial stress, due to self-weight, became significant. In the limiting case, where wind speed was zero, there was no bending moment and the stress that remained uniform with height was solely the compressive stress due to self weight. Stress distributions at different wind speeds

If we assume that most trees grow in areas with an average wind speed of 5 m s-l and produce stems with uniform stress, we can calculate the stress distribution that develops when those trees are temporarily subject to higher or lower wind speeds. Figure 7 shows the stress distributions in such a tree of P. sitchensis, based on the same assumptions as for Figure 3 (Gardiner 1989). When wind speed was increased from 2.5 to 10.0 m SC, the stress at the base of the tree was calculated to increase from 0.6 to 11.3 MPa. However, the stress

SHAPE

OF TREE

STEMS-UNIFORM

STRESS

HYPOTHESIS

59

Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

0.2

0.0

0.2

0.4

0.6 Relative

0.8 stress

1.0

1.2

1.4

Figure 7. Calculated stress distribution in a tree of P. sirchensis which has a height-diameter profile giving uniform stress in a steady wind of 5 m ss. (This is the same profile as shown by the line in Figure 3.) The stress distributions are shown for wind speeds of 20, 10,5,2.5 and 1.25 m s-. Each stress distribution is scaled to a value of 1.0 at the base of the tree, which corresponds to 39.0, 11.3, 2.7, 0.6 and 0.1 MPa in wind speeds of 20, 10, 5, 2.5 and 1.25 m ss, respectively. Calculations were made assuming the same tree properties as in Figure 3 (Gardiner 1989).

distribution remained approximately uniform up the stem (Figure 7). As stated above (Figure 6), the heightdiameter profile for uniform stress was similar for forces produced by wind speeds over this range. The reason is that there are opposing forces. As a stem deflects, the effective lever arm of the wind force in the crown decreases with height, which reduces the relative bending stress with height, but, at the same time, the weights of the crown and stem produce an increase in relative stress with height. These two forces will cancel out until the deflection becomes large (> 20% of stem length). At high wind speeds (20 m s-l), the absolute stresses in the stem were large (see legend to Figure 7) but the deflection of the trees caused a substantial decrease in relative stress from the bottom to the top of the stem. Conversely, at a wind speed of 1.25 m s- (light breeze), the absolute stresses were very small, but the relative stress increased from the bottom of the stem to below the crown (Figure 7). The change in stress distribution with change in wind speed may help to explain some of the observations of non-uniform stress in the literature. The points and dashed line in Figure 8 reproduce the stress values calculated by Milne and Blackbum (1989) for the tree of P. sitchensis that is illustrated in Figure 4b (from their Figure 5). Milne and Blackbum (1989) based their calculations on a wind speed of 15 m s- at the top of the canopy. As mentioned above, our calculation of diameters given in Figure 4b was based on a wind speed of 5 m s-l, and assumed uniform stress of 0.34 MPa. When our tree (with diameters giving 0.34 MPa uniform stress) was

60

MORGAN

AND

CANNELL

Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

2 stress

3 (MPa)

Figure 8. Stress distributions in the tree of P. sitchensis whose height-diameter profile is shown in Figure 4b. Broken line = stress values calculated by Mime and Blackbum (1989, their Figure 5) based on measured diameters and a wind speed of 15 m s- at the top of the canopy. The open point is the calculated stress in the region of butt swell. Curved line = stress distribution produced by a wind speed of 15 m s-, m a tree with the same height-diameter profile as shown in Figure 4b, that is, with a shape giving uniform stress (0.34 MPa) in a wind speed of 5 m s-l.

subjected to a wind speed of 15 m SC (at the top of the canopy) the estimated stress increased to 3 MPa at the base of the stem, but decreased with height, giving a close fit to Mime and Blackbums values (Figure 8). West et al. (1989) reported that the stress distribution in E. regnans trees deviated from uniform when the wind speed, and hence, the force on the crown, decreased from the top to the bottom of the crown. However, it would appear that their stated wind speeds were arbitrary, because the overall force on the crown corresponding to a stated wind speed (at the top of the canopy) of 12 m s-l was given as only 200 N, which we estimated to give a deflection of only 0.35 m at the top of a 28 m tall tree (with diameters giving equal stress, shown in Figure 4a). Evidence for non-uniform stress was then based on a stated wind speed of only 5 m s-, corresponding to a total force of perhaps 60 N, which we estimated to give a deflection of only 0.08 m. If a tree with diameters giving uniform stress in response to a total force of 200 N (as in Figure 4a) is then subjected to a force of only 60 N, we estimated that the stress increased non-linearly by over 50% from the base to the top of the stem. We suggest that the use of very low force values may also account for the fact that West et al. (1989) found that the bending moment was proportional to the diameter to the power of only 2.3 (l/O.43 in their Table l), whereas our analysis, based on a 200 N force, gave a power value of 3.1.

Discussion
In general, our analysis supported the hypothesis stated at the beginning of the paper,

SHAPE

OF TREE

STEMS-UNIFORM

STRESS

HYPOTHESIS

61

at least to an approximation (say 10%) for the stem region above butt swell. Three salient points have come from our analysis. (1) Tree stems seem to grow to equalize the average stress (bending plus axial) at their outer surface. This apparent goal is a time-averaged response to a variable force. At any time, the actual stress distribution may deviate from uniform, because of lags that occur between changes in crown dimensions (and hence, of forces on the tree) and changes in stem diameter. (2) The stress distribution in a tree stem is not constant over time, but changes as the force on the crown changes. The stem shape that develops in response to average forces wili show non-uniform stress distributions if measurements or calculations are made based on very low or high force values. (3) Average wind speeds over the 4-fold range of 2510.0 m s- give similar force distributions in stems, which consequently have similar stem shapes under the condition of uniform stress. We conclude that most of the differences observed in the taper of tree stems (e.g., at different spacings) may be related more to differences in their crown dimensions than to differences in the average wind speeds that they experience. The transfer matrix method of structural analysis is clearly a powerful method for examining relationships between force distributions and the dimensions of trees. Stress can be calculated from diameter measurements (Dean and Long 1986, Milne and Blackbum 1989, West et al. 1989), or measured diameters can be compared with those calculated using assumptions about the distribution of stress (this paper). Nevertheless, either way, the test of the uniform stress hypothesis relies on diameter measurements that are subject to error and give rise to a 10% uncertainty about the distribution of stress. It is unlikely that further progress can be made until methods are developed to measure stress in tree stems directly. Meanwhile, if we accept the uniform stress hypothesis as a good approximation (e.g., Mattheck 1990, 1991), the transfer matrix method can be used to constrain models of assimilate partitioning in trees (Cannel1 and Dewar 1993), and to predict stem shapes for any given tree size, weight and force distribution in the crown that may result from genetic or silvicultural manipulation.
Acknowledgments We are grateful to Dr. R. Milne for making data available for the construction for discussions that stimulated further analysis of his data. We are also grateful critical review of an early manuscript. of Figures 4b and 8, and to Dr. R.C. Dewar for his

Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012

References Assmann, E. 1970. The principles of forest yield study. Pergamon Press, Oxford. Cannell, M.G.R. and I. Morgan. 1988. Support costs of different branch design: effects of position, number, angle and deflection of laterals. Tree Physiol. 4:219-23 1. Cannell, M.G.R., J. Morgan and M.B. Murray 1988. Diameters and dry weights of tree shoots: effects of Youngs modulus, taper, deflection and angle. Tree Physiol. 4:219-231.

62

MORGAN

AND

CANNELL

Cannell, M.G.R. and J. Morgan. 1989. Branch breakage under snow and ice loads. Tree Physiol. 6:307-317. Cannell, M.G.R. and J. Morgan. 1990. Theoretical study of variables affecting the export of assimilates from branches of Picea. Tree Physiol. 6:257-266. Cannell, M.G.R. and Dewar, R.C. 1993. Carbon allocation in trees: a review of concepts for modelling. Adv. Ecol. Res. 25:59-104. Coutts, MI? 1986. Components of tree stability in Sitka spruce on peaty gley soil. Forestry 59:173-197. Dean, T.J. and J.N. Long. 1986. Validity of constant-stress and elastic-instability principles of stem formation in Pinus contorta and Trifolium pratense. Ann. Bot. 58:833-840. Gardiner, B.A. 1989. Mechanical characteristics of Sitka spruce. Forestry Commission Occasional Paper 24, 11 p. Hunt, E.R. and M.J. Jaffe. 1980. Thigmomotphogenesis: the interaction of wind and temperature in the field on the growth of Phaseolus vulgaris L. Ann. Bot. 45:665-672. Jacobs, M.R. 1954. The effect of wind sway on the form and development of Pinus radiata D. Don. Austr. J. Bot. 2:35-5 1. Jaffe, M.J. 1973. Thigmomorphogenesis: the response of plant growth and development to mechanical stimulation. Planta 114:143-157. Kellomaki, S., Hari, P., Kanninen, M. and P. Ieonen. 1980. Eco-physiological studies on young Scats pine stands. II. Distribution of needle biomass and its application in approximating light conditions inside the canopy. Silva Fenn. 11:519-538. Landsberg, J.J. and G.B. James. 1971. Wind profiles in plant canopies. J. App. Ecol. 8:729-741. Larson, P.R. 1965. Stem form of young Larix as influenced by wind and pruning. For. Sci. 11:412-424. Massman, W.T. 1982. Foliage distribution in old-growth coniferous tree canopies. Can. J. For. Res. 12:10-17. Mattheck, G.C. 1990. Engineering components grow like trees. Materialwiss. Wekstofftech. 21:143168. Mattheck, G.C. 1991. Trees, the mechanical design. Springer-Verlag, Berlin, 121 p. Metzger, K. 1893. Der Wind als massgebender Faktor fur das Wachtsum der Baume. Mundener Forstliche Hefte 3:35-86. Milne, R. and P. Blackbum. 1989. The elasticity and vertical distribution of stress within stems of Picea sitchensis. Tree Physiol. 5: 195-205. Mitchell, C.A., Dostal, H.C. and T.M. Seipel. 1977. Dry weight reduction in mechanically-dwarfed tomato plants. J. Amer. Sot. Hort. Sci. 102:605-608. Morgan, J. 1989. Analysis of beams subjected to large deflections. Aeronaut. J. 93:356-360. Morgan, J. and M.G.R. Cannell. 1987. Structural analysis of trees trunks and branches: tapered cantilever beams subjected to large deflections under complex loading. Tree Physiol. 3:365-374. Norman, N.D. and P.G. Jarvis. 1974. Photosynthesis in Sitka spruce (Picea sitchensis (Bong.) Cam III. Measurement of canopy structure and interception of radiation. J. Appl. Ecol. 11:375-398. Petty, J.A. and C. Swain 1985. Factors influencing stem breakage of conifers in high winds. Forestry 58:75-84. Shellard, H.C. 1976. Wind. In: The Climate of the British Isles. Eds. T.J. Chandler and S. Gregory. Longman, London and New York, pp 96-128. Timoshenko, S. 1955. Strength of materials, 3rd Edition, Van Nostrand. Vabnger, E. 1990. Influence of thinning, fertilization, wind and tree size on the development of Scats pine trees. Dissertation. Swedish University of Agricultural Sciences, Department of Silviculture, Umea. Wang, Y.P., Jarvis, P.G. and M.L. Benson. 1990. Two-dimensional needle-area density distribution within the crowns of Pinus radiata. For. Ecol. Manage. 32:217-237. West, P.W., Jackett, D.R. and S.J. Sykes. 1989. Stresses in, and shape of, tree stems in forest monoculture. J. Theor. Biol. 140:327-343. Wilson, B. and R. Archer. 1979. Tree design: some biological solutions to mechanical problems. Bioscience 29:293-298. Worrall, J.G. 1980. The impact of environment on cambial growth. In Control of Shoot Growth in Trees. Ed. C.H.A. Little. Fredericton, New Brunswick, Canada. Zimmermann, M.H. and C.C. Brown 1971. Trees: structure and function. Springer-Verlag, New York.

Downloaded from http://treephys.oxfordjournals.org/ at Universite du Quebec a Rimouski on January 12, 2012