The southeastern region of Buenos Aires province (Argentina) has a forest resource of

In Australia,

The value of

The contents of sapwood and heartwood are quality attributes that are important for the different end uses of wood. Heartwood contains extractives and a smaller content of moisture than sapwood, whereas sapwood contains living parenchyma cells. The cell walls in heartwood can be infiltrated by polyphenols, which allow reducing both shrinkage and the swelling ability of wood, and increasing its durability and other properties (Hillis, 1987; Taylor

In the living tree, the sapwood, in contrast with heartwood, is physiologically active, conducting water and nutrients from roots to leaves and storing food materials (Bamber, 1976; Hillis, 1987). The transformation of sapwood into heartwood is characterized by the death of parenchyma cells, development of tyloses in the vessels of many species and the biosynthesis of nonstructural compounds, leading to an important accumulation of extractives and to the differences in physical and chemical properties between sapwood and heartwood (Bamber, 1976; Hillis, 1987). Heartwood and sapwood in a tree vary with many factors, including species, age, rate of growth, climate, foliage area, site quality and tree vitality. Functions and causes of heartwood formation in trees are complex and consequently a great number of hypotheses, sometimes contradictory, may be found in the bibliography from the past decades, as reviewed by Bamber & Fukazawa (1985) and Hillis (1987). The regulation of water-conducting area in the stem and the preservation of non-functional wood are widely accepted features linked to heartwood formation (Bamber, 1976; Berthier

Several researchers have reported on the content of heartwood in

The aims of this study were: 1- to determine the heartwood and sapwood content and their distribution within the stem, 2- to analyze their relationship with the growing site, age and growth rate of the trees, and 3- to predict heartwood content by means of easy-to-measure variables (diameter at breast height and tree height), in relation to its potential use in the construction and furniture industries.

Hypothesis

Given the structure of non-uniform age forest, the age of the trees and the site of growth affect differently the heartwood content. Despite of that, it is possible to adjust a linear model to predict the heartwood content.

The study material consisted of four non-implanted forests of

The experimental material was obtained from 20 trees (five from each of the sites), randomly selected with two restrictions: they belonged to the higher diametric classes (greater than 10 cm in diameter at breast height - DBH-) and to the co-dominant stratum (Table S1, [supplementary]). The trees age varied between 9-32 years. Total trees height (TH) range was 12.6-18.8 m and DBH range was 12.5-32.2 cm. As shown in Table S1 [supplementary], trees of similar diameters showed different ages and vice versa. The same age was found in trees of different DBH values (development), typical of forest of non-uniform age.

To estimate the volume of the trees selected, we used Smalian’s formula (Caillez, 1980; Prodan

The age was estimated from a basal disc (taken 30 cm above ground level) by microscopic observation of histological cross-sections covering the entire radius with an optical microscope (Olympus BX50, Japan).

We defined four sampling heights in the stem: the base, the breast height (BH) (corresponding to 1.3 m above soil level), 30% of the total height (30%TH) and 50% of the total height (50%TH). The commercial height represented on average 57.5% (± 6.2) of total height (Table S1 [supplementary]), for which sampling up to 50%TH can be considered corresponding to the commercial portion of the trunk. A 5-cm-thick sample disc was taken from each sampling height. This material was naturally dried for 6-10 months in a covered storage room at room temperature. Sandpaper with decreasing grain size (80-100-120) was used to polish the cross-sectional surface of each slice on which naked-eye observations were made to delimit the heartwood region and the measurements.

The following measurements were performed with millimeter ruler on polished discs: a) North-South and East-West diameters of each slice, with and without bark, centered in the pith; and (b) North-South and East-West diameters of the heartwood area, centered in the pith.

The heartwood content was expressed in terms of:

1. Its proportion in the cross-section: area of heartwood in relation to the total area of the cross-section with bark; area/area ratio, for the analysis of the axial variation in heartwood content.

For the calculation of the areas, it was assumed that the cross-sections of the discs and heartwood were circular, with mean diameter resulting from the average of those measured in North-South and East-West direction, already mentioned. The following expressions were applied:

where AT
_{ob}
: total over bark area of the cross-section [cm
^{2}
]; A
_{HW}
: heartwood area [cm
^{2}
]; D
_{ob}
: mean over bark diameter of the cross-section [cm]; and D
_{HW}
: mean diameter of the heartwood [cm].

2. Its proportion in volume in each conical cross-section defined by the sampling: volume of heartwood in relation to volume with bark of each log into which the trunk was divided; volume/volume ratio, for the analysis of the axial variation in heartwood content. For this analysis, the trunk was divided into three conical sections or log (Base-BH; BH-30%TH; 30%TH-50%TH) each of which was measured for the calculation of its volume with bark and heartwood volume. For this calculation, we used Smalian’s formula.

The estimation of the total volume with bark and heartwood volume of each log was based on the stem heights (determined at the time of felling) corresponding to each axial position of the sampling (BH, 30%TH and 50%TH) and the respective transverse areas determined in accordance with

3. Its absolute value in heartwood volume of the whole tree (m
^{3}
), resulting from the sum of the heartwood volumes in each conical section (log) into which the trunk was divided.

4. Its proportion (%) estimated for the whole tree (used in the regression analyses):

I. as the simple average value per tree of the percentage of heartwood of each cross-section (area/area ratio),

II. as the simple average value per tree of the percentage of heartwood of each log into which the trunk was divided (volume/volume ratio),

III. as the value (%) per tree emerged from the ratio between heartwood volume (m
^{3}
) (according to section C.3) and total volume of the tree with bark (m
^{3}
) (according to section A); volume/volume ratio.

5. Sapwood thickness: The thickness of the sapwood was calculated in radial direction based on the mean diameter of each slice without bark and the mean diameter of the heartwood, according to the following expression:

where SW: sapwood thickness [cm]; D
_{ub}
: under bark mean diameter of the cross-section r [cm]; and D
_{HW}
: mean diameter of the heartwood [cm].

The analysis of variance (ANOVA) was developed under a statistical model for fixed effects, where the effects considered as fixed were: the site, the height at the stem, and the tree nested at the site [4]. The alpha probability value established to consider the differences between the sources of variation as significant was 0.05. Tukey’s test (honestly significant difference -HDS) was used as a method of multiple comparisons of means. The assumptions of normality and homogeneity of variance were verified by the Shapiro-Wilk and Levene statistical tests respectively.

where y ijkl: estimation of the variable corresponding to the k-th tree at the j-th height of the stem at the i-th site; a i: effect of the i-th site; ßj: effect of the j-th height of the stem; dk(i): effect of the k-th tree at the i-th site; aßij: effect of the interaction between the i-th site and the j-th height of the stem; e jk(i): effect of the k-th tree at the j-th height at the i-th site

Pearson correlation analysis was used for the analysis of the linear relationships. General linear models were used to analyze the relationships between heartwood content, DBH, total height (predictor variables), site and age tree (categorical independent variables) (Statistica 7). The models whose parameters were estimated by least squares were adjusted and ANOVA was developed to test the significance. The confidence intervals of the parameters of the models were calculated. The suitability (or quality of fit) of the models was assessed by the determination coefficient.

All the trees presented heartwood along the stem. The mean heartwood content (area/area and volume/volume) for the experimental material was 47.6% (±16.3) and 49.2% (±14.7) respectively. The average thickness of the sapwood was 2.18 cm (±0.75), whereas that of the bark was 0.81 cm (±0.28).

The ANOVA indicated that the site, the sampling height and the tree were significant sources of variation (p<0.001) for the proportion of heartwood in the cross-section of the stem (area/area), the proportion of heartwood in relation to the volume (volume/volume), and the sapwood thickness (

The sites LT and MCh (the latter of which had the youngest trees of all the sites studied) did not differ in heartwood content (either as a proportion in the cross-section or as volume/volume ratio) and had the highest proportion of heartwood (

The heartwood content in the cross-section was highest in the basal region of the stem up to breast height, and then decreased towards the apex. Sapwood thickness was highest at the upper end of the stem, although statistically, this value did not differ from that recorded at the base, and thus could be considered largely constant along the stem (

Based on the average data of the 20 trees studied, the heartwood radius decreased uniformly, closely following the shape of the stem. In contrast, sapwood thickness remained constant (

Sapwood thickness was not correlated with the height of the stem (TH, r = 0.06; p = 0.57), while the heartwood radius showed a negative and significant relationship with the height of the stem (r = -0.46; p < 0.001).

The heartwood radius and the radius of the cross-section (over and under bark, and at all sampling height) were positively and significantly correlated (r= 0.96; p < 0.001 for the section over bark and p< 0.05 for the section under bark). Thus, the linear models predicting the diameter of the heartwood based on diameters of the stem over and under bark showed high fitting (R
^{2}
= 0.92 and p< 0.001 in both cases) (

The proportion of heartwood, determined with the three calculation procedures used, was positively correlated with TH, DBH over bark and the volume of the stem. The correlation was greater when the heartwood content of the whole tree involved in the proportion was estimated as the sum of the volumes of heartwood in each log (
_{ob}
). Eleven 19 – 27 years old trees sampled at the others three sites showed a great variation of heartwood content (33-65%) and DBH (16.6 – 32.3 cm) (

^{3}
) as dependent variable and DBH, TH, tree age and site as predictor (covariates) or categorical independent variables (factors). Heartwood content could be predicted by DBH or TH alone. The site or tree age did not affect the adjustment of the model. Trees of the LC site had the lowest DBH dispersion (15-17 cm), with large age range (19-32 years) and low heartwood content (35-40%). In contrast, LT trees showed the highest range of DAP (16.6 - 33.2 cm), with tree age like LC trees site (17-32 years) and with high heartwood content (55-65%) (

The 20 trees studied were characterized by an average height to the crown of 57% of the total height (Table S1 [supplementary]). In addition, all the trees presented heartwood along the stem, which is an important aspect for the potential use of this wood destined to sawing (Searle & Owen, 2005; Bradbury

Our results showed that the development of the heartwood had a conical shape (correlation between heartwood radius and height r= -0.46) and closely followed the contour of the stem (correlation between radius of the heartwood and cross-section radius r= 0.96), which can be considered as a regular pattern (Climent

The local plantations studied presented a proportion of heartwood which could be considered within the range of values reported in other countries for ordered plantations (

It is accepted that the heartwood area is usually greater at the base of trees and decreases towards the apical region (Hillis, 1987; Taylor

The sapwood thickness in our material (2.18 cm) was lower and showed greater individual variation (

The high variation of heartwood content among trees found in the present study could be related to the non-uniformity of ages and/or different growth rates that characterized the material within the sites (

It has been indicated that the better the conditions of the site, the greater the formation of heartwood, but it has also been mentioned that this analysis can be confusing (Taylor

The estimation of heartwood volume in
^{2}
= 0.79 - 0.89). Therefore, the proposed hypothesis is accepted. The heartwood content was affected differently by the tree age and the growing site due to the forest structure of non-uniform age and without management. In our sampled trees, the absolute amount of heartwood was driven by growth rate. However, it is possible to fit a general linear model for heartwood prediction.

The

The age of the trees did not affect the amount of heartwood, whereas the environmental conditions of the sites did affect this parameter. These positive heartwood/size and heartwood/growth rate ratios previously reported for the species were also recorded in the local resource studied. It is then expected that a forest management aimed to achieve a structure of regular mass, with major rotations and interventions on plant density, will lead to higher and more uniform development of the trees. According to the results of this research, these better developments will be associated with higher heartwood content, which is the most important quality attribute of this wood. On the other hand, the heartwood volume in this forest resource can be estimated through linear equations whose predictive variables are DBH and TH.

We would like to particularly thank to Ing. Ftal. Marcelo Elizalde and Ing. Ftal. Fabio Achinelli for their professional assistance in wood sampling. We want to thank Ma. Victoria Gonzalez Eusevi for translating the manuscript and Dr. Enrique Portiansky for his disinterested help with the proofreading of the English writing.