Does Canopy Position Affect Wood Specific Gravity in Temperate Forest Trees?

Abstract

The radial increases in wood specific gravity known in many tree species have been interpreted as providing mechanical support in response to the stresses associated with wind loading. This interpretation leads to the hypothesis that individuals reaching the canopy should (1) be more likely to have radial increases in specific gravity and (2) exhibit greater increases than individuals in the subcanopy. Wood specific gravity was determined for three species of forest trees (Acer rubrum, Fagus grandifolia and Tsuga canadensis) growing in central Massachusetts, USA. Acer rubrum shows radial increases in specific gravity, but these increases are not more pronounced in canopy trees; the other two species show a pattern of radial decreases. The degree of radial increase or decrease is influenced by tree height and diameter. Of the dominant tree species for which we have data, A. rubrum, Betula papyrifera and Pinus strobus show radial increases in specific gravity, whereas F. grandifolia, T. canadensis and Quercus rubra show decreases. The occurrence of radial increases in B. papyrifera and P. strobus, which are often canopy emergents, suggests that it is overall adaptive strategy that is important rather than position (canopy vs. subcanopy) of any individual tree. It is suggested that radial increases in specific gravity are associated with early‐successional status or characteristics and decreases with late‐successional status or persistence in mature forest.

INTRODUCTION

The density or specific gravity of wood is a highly variable entity: variation among species ranges from approx. 0·12 for light‐wooded tropical trees, such as Ochroma, to approx. 1·3–1·4 for the densest woods, e.g. Krugiodendon ferreum (Zobel and van Buijtenen, 1989). The variation present within individual trees includes radial changes within the trunkwood (Wiemann and Williamson, 1988, 1989a, b; Omolodun et al., 1991; Butterfield et al., 1993; Woodcock et al., 2000). Various representations of strength are highly correlated with specific gravity (Zobel and van Buijtenen, 1989; Niklas, 1997). Biomechanical considerations, discussed below, lead to the hypothesis that trees reaching the canopy should (1) be more likely to have radial increases in specific gravity and (2) exhibit greater increases than subcanopy trees.

Specific gravity is known to vary within individual trees as follows: (a) there is often variation in the vertical direction (Taylor and Wooten, 1973; Zobel and van Buijtenen, 1989; but see Rueda and Williamson, 1992); (b) among trees exhibiting radial increases in specific gravity, differences between the inside and outside wood are particularly pronounced among tropical pioneer species (Wiemann and Williamson, 1988, 1989a). Increases have been widely reported for species found in wet‐tropical forests (Wiemann and Williamson, 1989a; Omolodun et al., 1991; Butterfield et al., 1993; de Castro et al., 1993; Woodcock et al., 2000) and are also known in species from tropical‐montane, dry‐tropical and temperate forests (Taylor and Wooten, 1973; Wiemann and Williamson, 1989a, b; Hernandez and Restrepo, 1995; Hernandez et al., 1998); (c) the maximum increases seen (0·2–0·25) can correspond to a change from light (specific gravity of 0·12–0·5) to medium (0·5–0·7), or from medium to high (>0·7) density categories. Rates of radial increase can be up to 0·11 g cm^–3 over 10 cm (Wiemann and Williamson, 1989a, b; de Castro et al., 1993); (d) rates of radial increase appear, at least in temperate trees, to be greatest during early growth or in small‐diameter trees (Panshin and de Zeeuw, 1980); (e) some temperate‐latitude trees, including some conifers and ring‐porous trees such as the oaks, display decreases rather than increases along the radius (Taylor and Wooten, 1973; Zobel and van Buijtenen, 1989; Wiemann and Williamson, 1989a); (f) the reaction wood that forms when trees are inclined from the vertical, functioning to right the tree, is generally higher in specific gravity than normal wood (Zobel and van Buijtenen, 1989); (g) variations in cell wall thickness, the proportion of late vs. early wood, the prevalence of fibres and ray density are among the anatomical characters that correlate with specific gravity (Panshin and de Zeeuw, 1980; Zobel and van Buijtenen, 1989; McDonald et al., 1995); (h) radial trends have also been related to differences between juvenile and mature wood and to the presence, in juvenile wood, of directional trends in a variety of wood characteristics (Zobel and Sprague, 1998).

The functional significance of variations in specific gravity needs to be seen in the context of the general functioning of wood within the tree. Conduction of water, structural support, the metabolic cost of wood production and the need to grow towards the canopy are all relevant considerations, and researchers have often proposed models of tree growth involving optimization of these and other factors (King and Louks, 1978; Niklas, 1992). Studying variations in wood density among species, Hacke et al. (2001) have shown that wood density is inversely correlated with the maximum (negative) xylem pressure that can be borne by the xylem elements without cavitation occurring. Variation in wood specific gravity within individual trees—specifically, the pronounced radial increases seen in tropical pioneers—has, on the other hand, been linked to mechanical strengthening of the trunk that would come from having stiffer/stronger material at the periphery (Wiemann and Williamson, 1989a, b; Wainwright et al., 1976).

In considering the way a tree experiences and responds to mechanical stresses that can cause mechanical failure or uprooting (including, perhaps most importantly, wind loading), the size and shape of the tree, anchoring by the root system and mechanical properties of the wood are all important (Ennos, 1997; Niklas, 1998). Bending stresses can also be resisted or accommodated through deformation. Using scaling analysis to study length and diameter relationships in Acer saccharinum, Bertram (1989) interpreted differences in geometry of distal and non‐distal branches as consistent with the idea that narrower distal branches bend or deform in response to wind, thus decreasing the drag forces experienced by the tree, whereas thicker, less‐distal branches tend to resist deformation. Scaling studies of other species have suggested that the entire tree, including the trunk, may bend in response to wind (King and Louks, 1978). It seems clear that the exact pattern of response may be quite variable given the diversity of tree architectural models, especially in the tropics, the range in values of wood specific gravity and associated structural characteristics of the wood, and the diversity of climatic conditions in the forested areas of the world.

The interpretation of Wiemann and Williamson (1989a, b) that radial increases in specific gravity are related to mechanical support rests on (1) the assumption that the trunk is resisting rather than deforming significantly in response to bending forces, and (2) the high correlation between specific gravity and the elastic modulus of the wood (Niklas, 1997), which together with the second moment of area determine stiffness. (Dependence on the second moment of area, I, which is equal to (πr⁴)/4 for a cylinder, shows the dramatic increase in stiffness conferred with an increase in diameter.) If this interpretation is correct, then radial increases should be more prevalent, and be of greater magnitude, in canopy trees than in individuals of the same species in the subcanopy because of the greater wind‐related stresses to which canopy trees are subject.

To test this hypothesis, three species of trees of Mixed Northern Hardwoods Forests were sampled: Acer rubrum L., Fagus grandifolia Ehrh. and Tsuga canadensis L. Carr. These species are capable of occupying both canopy and subcanopy positions. To gain information about patterns of radial variation in other dominant forest trees in the area, material was also collected from Quercus rubra L., Betula papyrifera Marsh. and Pinus strobus L., the main difference being that these samples were not stratified by canopy category. The study was carried out in mature forest in which the majority of trees sampled had ring counts greater than 100.

A subsequent analysis of a larger data set, including data from this study and the published literature, resulted in a model of specific gravity variation published elsewhere (Woodcock and Shier, 2002).

MATERIALS AND METHODS

The study site is in central Massachusetts, USA, on a tract of land adjoining the Prospect Hill section of Harvard Forest on the south (42°32′0′′N, 72°10′20′′W). The region supports Mixed Northern Hardwoods Forest that has largely grown up since the period of agricultural abandonment in the late 1800s. The mosaic of forest types represented is influenced by heterogeneity of landscape characteristics, land use, date of abandonment and logging history (Foster, 1992). Trees in several areas of the tract were cored to obtain sample sizes large enough for canopy comparisons. Included were many trees older than those generally found in the area (ring counts between 100 and 200). These came from a portion of the tract that may represent an old woodlot similar to that described by Foster et al. (1992) for a part of Harvard Forest.

Canopy categories were derived as follows (Fig. 1). In the case of A. rubrum and F. grandifolia, morphology was taken into account in addition to diameter and height. For these two species, three categories were used: (1) canopy trees, which reached into the canopy and had spreading crowns; (2) canopy entrants, which had leaders extending into the canopy, but not necessarily in gaps, and lacked well‐developed crowns; and (3) subcanopy trees, which were below the canopy. For T. canadensis, only the categories of canopy and subcanopy were used; this distinction rested essentially on height and position relative to the forest canopy. Ten trees per canopy category were sampled for these three species (n = 80). For the three additional species (Q. rubra, B. papyrifera and P. strobus), five individuals were sampled to give a range of diameters. Information about diameters and ring counts is listed in the Appendix.

Fig. 1. Height by diameter relationships by canopy category for Acer rubrum (A), Fagus grandifolia (B) and Tsuga canadensis (C).

Two cores, one directly above the other, were taken per tree at breast height using a 5‐mm increment borer. As far as was possible, sampling was limited to straight, unforked trees. One core was used for mounting and determination of ring counts, whereas the other was used for specific gravity measurements. Cores were hermetically sealed and stored frozen until specific gravity measurements were taken. An inclinometer was used to record tree height. Diameter was measured at breast height (1·4 m above the ground).

Cores were divided into 2‐cm segments from pith to bark. Green volume was obtained using the water displacement method (measured to the nearest 0·001 g; McDonald et al., 1995), with care being taken to maintain the moisture content above 30 %, the fibre saturation point. Cores were then oven‐dried at 103 ± 2 °C for 24 h, cooled, and then weighed to the nearest 0·001 g. Specific gravity was determined as the ratio of oven‐dry weight to green volume. This quantity is dimensionless and represents density relative to water (density of 1 g cm^–3 or specific gravity of 1).

Radial variation in specific gravity was evaluated using three values per tree: the inside, middle (average of all middle segments) and outside segment. This averaging procedure was carried out to account for variation in tree diameter (Hernandez and Restrepo, 1995). The high degree of variability typical of this entity makes ANOVA on grouped data a more robust analytical approach than regression against radial distance for individual trees. A balanced split‐plot design with repeated measures was used to compare specific gravity across canopy categories. ANOVA was carried out using all‐pairs comparisons with the level of significance set at 0·05.

RESULTS

Effect of canopy category on specific gravity

Averaged values of wood specific gravity for the three canopy categories are shown in Fig. 2. ANOVA did not identify canopy category as a significant influencing factor (Table 1). Thus, the hypothesis advanced here is not supported: canopy trees of the three species studied are not characterized by a higher frequency of radial increase or more pronounced increases. Canopy trees are characterized by denser wood overall in A. rubrum and by denser interior wood in the other two species. Deposition of extractives in the heartwood is one possible explanation for these patterns (Zobel and van Buijtenen, 1989). Maximum specific gravity shows the clearest difference between canopy and canopy entrant or subcanopy trees (Fig. 3): the difference is significant for A. rubrum and F. grandifolia, and marginally significant for T. canadensis.

Fig. 2. Average specific gravity as a function of radial position and canopy category for Acer rubrum (A), Fagus grandifolia (B) and Tsuga canadensis (C). Determinations based on increment cores cut into 2‐cm pieces with the middle sections averaged. n = 30 for A. rubrum and B. grandifolia, and 20 for T. canadensis.

Table 1.

ANOVA for canopy category and specific gravity by radial position

	d.f.	SS	F	P value	% Variance
Acer rubrum
Source of variation
Canopy category	2	0·01450	2·9504	0·0671
Position	2	0·00840	5·0573	0·0103*	6
Canopy category × position	4	0·00144	0·4329	0·7841
Tree (canopy category)	27	0·07787	3·4743	<0·0001*	54
Error	46	0·03820			26
Total	81	0·14410
Fagus grandifolia
Source of variation
Canopy category	2	0·00012	0·0262	0·9742
Position	2	0·01365	9·6643	0·0003*	11
Canopy category × position	4	0·00580	2·0521	0·1027
Tree (canopy category)	27	0·07436	3·8988	<0·0001*	57
Error	46	0·03250			25
Total	81	0·12930
Tsuga canadensis
Source of variation
Canopy category	1	0·00082	0·2175	0·6466
Position	2	0·03175	5·6452	0·0074*	15
Canopy category × position	2	0·01027	1·8264	0·1756
Tree (canopy category)	18	0·06800	1·3429	0·2202	32
Error	36	0·10124			48
Total	59	0·21210

* Statistically significant at the 1 % probability level.

Fig. 3. Maximum specific gravity per tree as a function of canopy category for Acer rubrum (A), Fagus grandifolia (B) and Tsuga canadensis (C). Box plots show the median (central bar), middle 50 % of the distribution (box), the outermost data points within the main portion of the distribution (bars), and outliers.

The radial trends seen in these trees (an increase in A. rubrum and decreases in F. grandifolia and T. canadensis) are significant and are associated with explained variances of 5–15 %. The categories of ‘tree (canopy category)’ and ‘error’ in Table 1 correspond to between‐tree and within‐tree variance, respectively. Values of both are high in the three species studied. Between‐tree variance is over 50 % in A. rubrum and F. grandifolia, whereas within‐tree variance is almost 50 % in T. canadensis and is higher than between‐tree variance in this species.

Factors affecting radial increase/decrease

Other variables for which we have data include tree height and diameter, number of rings (apparent age) and average growth rate (average number of rings per centimetre). Significant inter‐relationships exist among these variables. One way of looking at the effect of individual factors is by means of semipartial correlations, which show the effect of any one factor on the dependent variable with the influence attributable to other factors held to zero. The correlations obtained show that in general, diameter and height have the most direct effects on the degree of radial increase or decrease (Table 2). Significant but relatively small effects are evident for height in A. rubrum and diameter in F. grandifolia and T. canadensis. In other words, in A. rubrum, taller trees tend to have greater radial changes (which are increases), and in F. grandifolia and T. canadensis, larger‐diameter trees tend to have greater radial changes (which are decreases). In addition, in T. canadensis, ring count also has an influence: trees with higher ring counts have less pronounced radial changes (smaller decreases) than trees in the same diameter class but with lower ring counts.

Table 2.

Semiparital correlations for all variables

	Difference in specific gravity (outside‐inside)		Average specific gravity
	r	P value	r	P value
Acer rubrum
Canopy category	0·214	0·256	–0·010	0·955
Diameter	–0·123	0·51	0·013	0·943
Height	0·390	0·044*	0·006	0·973
Rings	–0·083	0·656	0·152	0·397
Average growth rate	0·046	0·805	–0·080	0·652
Fagus grandifolia
Canopy category	0·013	0·938	0·118	0·56
Diameter	–0·361	0·043*	0·028	0·89
Height	–0·010	0·953	0·164	0·42
Rings	0·272	0·12	0·083	0·682
Average growth rate	–0·129	0·452	–0·007	0·973
Tsuga canadensis
Canopy category	0·082	0·625	–0·296	0·146
Diameter	–0·664	0·001*	0·133	0·5
Height	0·293	0·096	–0·171	0·399
Rings	0·449	0·016*	–0·173	0·383
Average growth rate	–0·327	0·066	0·407	0·058

*Statistically significant at the 5% probability level.

Factors influencing average specific gravity

When average specific gravity is considered, analysis shows that the variables are not significantly related (Table 2). Note, however, that growth rate, as considered here, is a very generalized measure (average growth rate per tree). The present analysis is not capable of addressing the way in which year‐to‐year variations in growth rate may affect specific gravity. Although it is tempting to assume that growth rate has an influence upon specific gravity, numerous studies of a wide range of species have proved inconclusive on this point (Zobel and van Buijtenen, 1989; Zhang and Zhong, 1991; de Castro et al., 1993).

Radial trends

For the six species for which we collected material, two categories can be recognized with respect to radial trends (Fig. 4A and B). P. strobus, B. papyrifera and A. rubrum show radial increases, whereas Q. rubra, B. grandifolia and T. canadensis show decreases. The interpretation advanced is that species in the first category are early successional (or, in the case of A. rubrum, have both early‐ and late‐successional characteristics; Abrams, 1998) and that those showing radial decreases are either late successional or, in the case of Q. rubra, are capable of persisting as long‐lived elements of mature forest. Additionally, both B. papyrifera and P. strobus are often emergent trees, extending between one and several metres above the forest canopy. The occurrence of radial increases in these emergent species suggests that such increases should be interpreted in terms of overall adaptive strategies rather than position below or above the canopy of any given individual.

Fig. 4. Radial trends in specific gravity for species showing increases (A) and decreases (B). C, Average specific gravity for the same species. n = 30 for Acer rubrum and Fagus grandifolia; n = 20 for Tsuga canadensis; and n = 5 for Pinus strobus, Betula papyrifera and Quercus rubra. Box plots as in Fig. 3.

The plots in Fig. 4 are very generalized. There is a high degree of variability from tree to tree (Fig. 4C) as well as within trees; in addition, some individuals have radial trends opposite to the more usual pattern for the species (see the Appendix). For all these reasons, it is difficult to represent precisely the radial changes. That said, some observations can be advanced about the degree of change seen. The increases in specific gravity in the six species studied are, in absolute terms (i.e. difference between inside and outside values), generally 0·1 or less. Tropical species, on the other hand, can show increases as great as 0·25 (Wiemann and Williamson, 1989a). When considering the rate of change (increase or decrease over distance), increases are 0·024, 0·020 and 0·069 g cm^–3 per 10 cm for A rubrum, P. strobus and B. papyrifera, as compared with a high of 0·106 g cm^–3 per 10 cm for tropical species (Wiemann and Williamson, 1989b). The greatest rate of increase found in this study was for a pioneer species that remains a relatively small‐diameter tree (B. papyrifera).

DISCUSSION

Significance for biomass estimates

Specific gravity is a direct reflection of the amount of carbon present since biomass is about half carbon by dry weight. Current methods of estimating forest biomass rely on allometric equations that express the relationship between biomass and diameter for individual species, used in conjunction with plot data to produce estimates of biomass by region (Brown and Lugo, 1992; Brown et al., 1997). In this way, variations in specific gravity are subsumed. The difficulties in understanding patterns of radial variation in wood specific gravity (only one aspect of specific gravity variation within trees) certainly highlight the advantage of the allometric approach. It would, for a variety of reasons, still be valuable to know more about the way in which individual trees and forests at particular locations are taking up and apportioning carbon. Such information would provide a point of comparison to the existing estimates of forest biomass and carbon sequestration. It is also pertinent to small‐scale studies of carbon cycling in forested ecosystems, not to mention basic questions of tree biology and adaptation to the environment.

For now, however, it is possible to make a few, relatively limited observations concerning the way our data may relate to estimates of forest biomass. (1) Values of specific gravity determined here are slightly higher (4–14 %) than published values (Forest Products Laboratory, 1999) and may give an indication of the geographical variation in these measures (see Appendix). (2) Compared with those in the tropics, the radial increases seen are somewhat less, and decreases have been found in some dominant species. The more consistent occurrence of radial increases in tropical forest trees may have a more clearly interpretable significance in evaluating changes in carbon sequestration by these forests. (3) In the case of the central Massachusetts forest studied here, there appear to be two trends that may affect carbon storage over time. One is a tendency for radial increases in specific gravity in trees in early successional forest (with the exception of Q. rubra, and presumably other oaks) and decreases in trees characteristic of later‐successional forest (Fig. 4). Secondly, there is a successional sequence from trees with low specific gravity (P. strobus, B. papyrifera), through hardwoods with high specific gravity, to late‐successional forest in which the softwood T. canadensis with low specific gravity is an important element. Both trends suggest that early‐ to middle‐successional forests may be higher in biomass, or may constitute increasing sinks of carbon, compared with late‐successional association (although the floristic makeup of the forest will undoubtedly play a role). Of course, evaluation of overall trends in carbon storage must take into account other factors, such as tree size and density, and amount of carbon going into soils, litter and woody debris, etc.

Functional interpretations

Although the hypothesis that individuals in the canopy display increased structural reinforcement as reflected in radial trends in specific gravity was not supported, the occurrence of radial increases in species that are early successional, as appears to be the case for three of the species (including two ‘temperate emergents’), is in agreement with other studies showing significant radial increases in pioneer taxa, particularly in the tropics (Wiemann and Williamson, 1988, 1989a). The association between radial decreases and later‐successional taxa on the other hand needs further confirmation.

There are many indications that pioneering species tend to show radial increases. Among softwood species, the pines, which generally occupy a pioneering position relative to other tree species, are generally characterized by increases, whereas other taxa (Picea, Abies) generally display decreases (Zobel and van Buijtenen, 1989). Ring‐porous species like the oaks generally show radial decreases (Wiemann and Williamson, 1988; Zobel and van Buijtenen, 1989), but radial decreases are not limited to ring‐porous taxa. The occurrence of radial decreases in later‐successional trees may seem paradoxical since these trees are long‐lived and can support large crowns necessitating greater support. However, as the trees grow in diameter, they experience increased strength due to an increase in the second moment of area (Wainwright et al., 1976) and perhaps also a decreased tendency for uprooting (Vogel, 1988). Thus, they may not require the reinforcement at the periphery associated with specific gravity increases. All the species showing radial decreases have the capacity to grow into large‐diameter trees, whereas this is not the case for all species showing increases (B. papyrifera, for instance). An interesting contrast is that of A. rubrum and A. saccharum: the latter species, which tends to have radial decreases (Sadjak, 1967), is later‐successional and both larger and longer‐lived than A. rubrum.

The biomechanical significance of the radial trends seen here is difficult to assess definitively. One problem is the difficulty of generalizing across species that vary widely in specific gravity, size and shape, and type of rooting system. Another is the issue of deformation as a response to wind, and the possibility that flexibility as opposed to stiffness may be favoured in tall trees, especially emergents. Width contributes to structural stiffness so for a given amount of material, narrow, denser branches or trunks will be more flexible than wider, less dense structures. If it is advantageous for pioneer trees and/or emergents to be (or grow to be) proportionately narrower/denser (and thus more flexible) as wind‐loading increases, then this could explain the radial increases seen in these taxa. However, the mechanical trade‐off cited above is for a given amount of material, whereas in a living tree width and density may not be so clearly related. Narrow growth increments are not necessarily associated with denser wood, even for diffuse porous species. In ring‐porous species, including the oaks (high canopy trees, although not emergents, in the study area), narrow rings are lower in specific gravity because of clear anatomical correlates (the higher proportion of wide‐vesselled earlywood in narrow rings). In addition, attainment of canopy position is normally associated with release in growth and increased wood production (wider rings) in the trunk and probably also the branches.

Variations in wood density also have a physiological dimension (Hacke et al., 2001; Hacke and Sperry, 2001). In a study of 48 species of gymnosperms and woody dicots from a range of locations, Hacke et al. (2001) found that cavitation resistance is positively related to specific gravity and to wood‐anatomical characteristics (conduit dimension and cell wall thickness) which are important determinants of this property. It is not so easy to extend this analysis to within‐tree variation in specific gravity except to note that trees might be expected to have denser wood in high branches where xylem pressure is more negative (J. Sperry, pers. comm.).

The number of variables that may influence specific gravity is quite large and includes factors particular to the tree (diameter, height, growth rate) as well as others, such as elevation, and soil and water conditions. Often there are inter‐relationships among these variables, and it is difficult to control for all factors, especially for trees in natural as opposed to plantation settings. Still, wood specific gravity is an important measure in various contexts, and better information about its variability will advance our understanding of the biology of trees and the way that they sequester carbon. Thus, there remains a need for well‐designed studies that control for potential influencing factors to the greatest extent possible.

ACKNOWLEDGEMENTS

This project benefited from field assistance from Sadhna Vora and Kelly Seary; the assistance of numerous people at Harvard Forest and Steve Wofsy and associates in the Department of Earth and Planetary Science at Harvard; and advice and suggestions from Roger Hernandez, Bruce Cutter, Elisabeth Wheeler and John Sperry. Support from the Bunting Institute of Radcliffe College, Radcliffe Research Partnerships, and the National Science Foundation (grant ATM‐07899) is gratefully acknowledged.

APPENDIX.

Summary data for all species

				Radial					Wood handbook
				change in	Number	Number	Average
		Diameter	Ring	specific	showing	showing	specific	Average	Specific	% under‐
	n	(m)	count	gravity	increase	decrease	gravity	s.d.	gravity	estimation
Acer rubrum							0·57	±0·025	0·49	14
Canopy	10	0·25–0·46	77–160	0·004–0·127	6	4
Entrant	10	0·15–0·29	62–110	0·002–0·095	7	3
Subcanopy	10	0·10–0·14	57–83	0·002–0·058	7	3
Fagus grandifolia							0·587	±0·032	0·56	4·4
Canopy	10	0·34–0·52	77–133	0·024–0·124	1	9
Entrant	10	0·17–0·28	67–89	0·001–0·065	1	9
Subcanopy	10	0·11–0·16	60–80	0·014–0·052	3	7
Tsuga canadensis							0·384	± 0·034	0·38	–
Canopy	10	0·41–0·66	68–198	0·021–0·318	4	6
Subcanopy	10	0·13–0·43	48–134	0·009–0·093	4	6
Quercus rubra	5	0·33–0·55	51–135	0·039–0·107	–	5	0·59	±0·031	0·56	4·9
Betula papyrifera	5	0·18–0·31	69–110	0·023–0·091	5	–	0·534	± 0·036	0·48	10
Pinus strobus	5	0·24–0·58	61–121	0·035–0·390	4	1	0·393	±0·065	0·34	13

In A. rubrum, F. grandifolia, T. canadensis and Q. rubra there is high variability in the inner wood. In addition, A. rubrum shows some decreases toward the periphery while T. canadensis shows increases.

Received: 8 July 2002; Returned for revision: 27 September 2002; Accepted: 10 December 2002    Published electronically: 20 February 2003