Dr. Hiroaki Ishii

Age-related development of crown stucture in coastal Douglas-fir trees

The study was conducted in 20-, 40- and 450-year-old stands of naturally established coastal Douglas-fir in the Wind River Ranger District, Gifford Pinchot National Forest in southwestern Washington State, USA. (45°N, 121°W; altitude 350 m). In this region, Douglas-fir establishes as a cohort following large-scale disturbance, and dominates in early stages of forest succession (Franklin and Hemstrom 1981). The 20- and 40-year-old stands originated after clear-cut harvesting, and are dominated by Douglas-fir. Stand height is approximately 15 and 30 m, stand density is 2125 and 1420 trees/ha , and stand basal area is 20.5 and 64.8 m2/ha, respectively. The 450-year-old stand originated after stand-replacing fire, and is dominated by Douglas-fir and western hemlock (Franklin 1972, Franklin and DeBell 1988). Stand height is approximately 60 m, stand density is 435 trees/ha, and stand basal area is 82.1 m2/ha. The three stands are located within 10 km of each other, have similar site conditions, and represent typical young and old stages of development for naturally established Douglas-fir stands in this region.

Field measurements
Five to six trees in each stand were chosen for measurement of crown structure (Table 1). The sample trees were chosen on the basis of DBH and tree height to cover the range of tree sizes observed in each stand. In the 20- and 40-year-old stands, the five sample trees in each stand were accessed from scaffolding towers. Persons taking the measurements climbed the trees while anchored to the tower by a harness and rope. For the 450-year-old stand, the six sample trees were accessed by the single-rope technique (Moffett and Lowman 1995) using a rope that was anchored to the main stem of the sample tree. The top 1-2 m of the sample trees could not be reached due to safety concerns, and was not measured. For all live (foliated) primary branches on the sample trees, branch height above ground was measured using a tape measure that was stretched vertically along the main stem of the tree. Branch diameter was measured immediately outside the branch collar using calipers for small branches and diameter tape for large branches. For the 20- and 40-year-old stands, branch length was measured for six branches on each sample tree. The crown of each tree was divided into three sections of equal depth: upper-, middle- and lower-crown. Two branches were chosen at random from each section on the northeast or southwest side of the tree. Diameter and length measurements for these branches were then pooled for each stand to obtain the following significant linear relationship between branch diameter (d, cm) and length (l, m):
l = 0.741 d + 0.47 (F = 98.34, P < 0.01, R2 = 0.830), (1)
l = 0.849 d + 0.22 (F = 137.85, P < 0.01, R2 = 0.810) (2)
for the 20- and 40-year-old stands, respectively. These relationships were used to estimate branch length from measured branch diameter for all branches of the sample trees. Because such a predictable relationship between branch diameter and length does not hold for old trees (Ishii, et al. 2000), branch lengths were individually measured for all branches of the sample trees of the 450-year-old stand by extending a one-inch-wide engineer's tape from the main stem to the farthest foliated section of the branch. For the 450-year-old trees, branches were also distinguished as being original (originated from the terminal bud of the main stem) or epicormic (originated from epicormic buds). Epicormic branches were distinguished non-destructively using several morphological characteristics: young bark, tangent angle of insertion to the main stem, multiple branches originating from a small area of the main stem, and smaller diameter relative to nearby original branches. See Ishii and Wilson (2001) for a detailed description of the distinguishing morphological characteristics of epicormic branches.

Measures of crown structure
The height above ground of each branch was translated into relative height (Hrel) in the measured live crown: Hrel = (branch height - lowest live branch height) / measured live crown depth, (3) where measured live crown depth = highest measured live branch - lowest live branch. (4) The measured live crown of each tree was then divided into ten height-classes of equal depth based on the relative height of each branch. Branch diameter squared multiplied by branch length was used as an indicator of individual branch volume (v): v = d 2 l. (5) This volume measure is analogous to (tree diameter)2 x (tree height), which has been shown to be correlated with tree biomass, and is expressed in terms of cm2 m sensu Fujimori et al (1976). This unit was used so that branches of varying sizes from the three stands could be expressed in reasonable digits. Using this measure, mean branch volume (V) was calculated for each height-class. Branch density (rB: mean number of branches per vertical meter of main stem) for each height-class was calculated as: rB = number of branches in height-class / depth of height-class. (6)

Analysis of crown development
The processes of branch growth, death and recruitment was analyzed using stand development as a conceptual analogue. Long and Smith (1984) recognize five stages of development of even-aged stands.
A. Initially, the stand is composed of small trees growing without competitive interaction.
B. Growth rates decline with the onset of competition, coinciding more or less with canopy closure.
C. Stand biomass and foliage area reach their maximum (full site occupancy). A size hierarchy develops among trees as a result of intraspecific competition.
D. Competition-induced mortality begins with the death of smaller, suppressed trees.
E. Individual tree sizes culminate, and large trees are unable to fill gaps created by mortality. Tree recruitment occurs in these gaps.
Because branches increase in age from upper- to lower-crown, the vertical change in mean branch size, branch density and branch size distribution within the tree crown can be considered as a chronosequence of cohorts of branches in various stages of development. In this analogy, branch growth corresponds to tree growth, branch death to tree death, and production of epicormic branches to tree recruitment.
Increase in tree size with increasing age initially follows the exponential growth curve during periods of increasing growth rate (stage A of Long and Smith 1984, see also Hozumi 1973). Eventually tree size reaches maximum (stage B), and tree growth is well represented by the general logistic growth curve (Hozumi 1973, Niklas 1994). Applying these models of tree growth, increase in mean branch volume (V, cm2om) from upper- to lower-crown was fitted by the exponential growth curve:
V = a exp (b Hrel). (7)
or the general logistic growth curve:
V = p /{1 + q exp (-r Hrel)} (8)
Parameters a, b, p, q and r were estimated using non-linear least-squares regression. Parameter p in equation (8) is the regression estimate of the maximum attainable branch volume. For the 450-year-old trees, the model of branch growth was applied to only the original branches which sequentially increase in age from upper- to lower-crown. Epicormic branches which vary in age depending on when they were released, were excluded from the analysis.
The decrease rate in tree density caused by competition-induced mortality (stage D of Long and Smith 1984) is relatively constant and can be represented by the negative exponential model (Hozumi 1973, Silvertown 1987):
r = r0 exp (-c t) (9)
where r is tree density, r0 is the initial tree density, t is time, and c is a constant. Applying this model, vertical change in branch density from upper- to lower-crown was modeled as a negative exponential process:
rB = r0 exp (-k Hrel) (10)
where rB is branch density, and r0 and k are parameters estimated using non-linear least-squares regression. In young trees, branch death does not occur in the upper-crown where branches receive sufficient light. In such cases, equation (10) may only be applicable to lower parts of the crown. The range of height-classes with the best fit of equation (10) was assessed by the highest r-square value obtained.
During stand development, competition-induced mortality occurs as a result of asymmetric competition where large trees have disproportionately greater growth rates than small trees, and eventually out-compete the small trees. The resulting pattern of growth and death of individual trees leads to changes in tree size distribution during stand development (Ford 1975, Mohler, et al. 1978, Hara 1988). Initially, there is little variation in size among individual trees (stages A-B of Long and Smith 1984). However, as the stand develops, small differences in size are augmented, because of differences among trees in their growth rate. With further stand development, a size-hierarchy develops and the size distribution becomes increasingly positively skewed and then bimodal (stage C). In later stages of stand development, small, suppressed trees begin to die, and the stand is comprised mainly of the surviving large trees (stage D). To infer the pattern of branch growth and death from upper- to lower-crown and with increasing tree age, we investigated the vertical change in branch diameter distributions of the sample trees.
The development of even-aged stands is characterized by the relationship between mean tree mass (W) and stand density (r):
W = A r B (11)
where A is a coefficient, and B is the scaling exponent. The development of crown structure from upper- to lower-crown and with increasing tree age was inferred from the relationship between mean branch volume (V) and branch density (rB) of the sample trees.
Finally, the vertical distribution of branch volume was investigated for each sample tree by calculating total branch volume for each height-class. This was used to infer age-related changes in the vertical distribution of crown biomass with increasing tree age resulting from the combined effects of branch growth, branch death, and production of epicormic branches.

References cited

Franklin, J. F., 1972. Wind River Research Natural Area. In: (Eds.), Federal Research Natural Areas in Oregon and Washington - A Guidebook for Scientists and Educators. Pacific Northwest Forest and Range Experiment Station, Portland. pp. WR1-WR11.

Franklin, J. F. and DeBell, D. S., 1988. Thirty-six years of tree population change in an old-growth Pseudotsuga-Tsuga forest. Can. J. For. Res., 18, 633-639.

Franklin, J. F. and Hemstrom, M. A., 1981. Aspects of succession in the coniferous forests of the Pacific Northwest. In: D. C. West, H. H. Shugart and D. B. Botkin (Eds.), Forest Succession. Springer-Verlag, New York. pp. 212-229.

Long, J. N. and Smith, F. W., 1984. Relation between size and density in developing stands: a description and possible mechanisms. For. Ecol. Manage., 7, 191-206.