Abstract
This article investigates the relationships between ring width (expressed as ring width class) and cambial age (expressed as chronological class) with specific gravity, modulus of rupture (MOR), compressive strength, and shrinkage. On those stands located on volcanic soils, it was found that when moving from the first ring width class (≤2 mm) to the seventh class (≥7 mm), a total decrease in specific gravity of 12.7% was observed, accompanied by a 19.5% decrease in compressive strength and a 22.8% decrease in MOR. With an increase in tree age, as expressed by the chronological class, there was a general decrease in the values of specific gravity, MOR, and compressive strength. It was therefore determined that chronological class is related to ring width, while specific gravity can predict MOR and compressive strength values for trees grown at volcanic sites. The results for a stand grown on calcareous soils showed a different trend. Furthermore, it was confirmed by cross-variance analysis that there was a correlation between ring width and chronological class.
Download PDF
Full Article
Wood Quality of Chestnut: Relationship between Ring Width, Specific Gravity, and Physical and Mechanical Properties
Manuela Romagnoli,a,* Daniela Cavalli,b and Stefano Spinaa,c
This article investigates the relationships between ring width (expressed as ring width class) and cambial age (expressed as chronological class) with specific gravity, modulus of rupture (MOR), compressive strength, and shrinkage. On those stands located on volcanic soils, it was found that when moving from the first ring width class (≤2 mm) to the seventh class (≥7 mm), a total decrease in specific gravity of 12.7% was observed, accompanied by a 19.5% decrease in compressive strength and a 22.8% decrease in MOR. With an increase in tree age, as expressed by the chronological class, there was a general decrease in the values of specific gravity, MOR, and compressive strength. It was therefore determined that chronological class is related to ring width, while specific gravity can predict MOR and compressive strength values for trees grown at volcanic sites. The results for a stand grown on calcareous soils showed a different trend. Furthermore, it was confirmed by cross-variance analysis that there was a correlation between ring width and chronological class.
Keywords: MOR; Shrinkages; Specific gravity; Compressive strength; Wood quality
Contact information: a: Department of Science and Technology for Agriculture, Forestry, Nature and Energy (DAFNE), University of Tuscia, 01100 Viterbo, Italy; b: Department for the Innovation of Biological, Agro-food and Forestry Systems (DIBAF), University of Tuscia, 01100 Viterbo; c: Freelancer; *Corresponding author: mroma@unitus.it
INTRODUCTION
Wood quality is a complex concept that incorporates various aspects related to wood defects, wood anatomy, and the physical and mechanical properties of wood. These aspects can be interrelated. For example, the physical and mechanical properties are often highly related to wood defects such as ring shake in chestnut (Spina and Romagnoli 2010; Romagnoli and Spina 2013). For a long time, foresters have attempted to identify easily measured parameters that could be good predictors of wood quality. Among these, wood density is considered a very important attribute because it is known to influence several physical and mechanical properties of the material (Genet et al. 2013). In particular, wood density affects the wood workability and mechanical strength (Giordano 1984; Tsoumis 1991), as well as other characteristics such as wood health and durability (Humar et al. 2008). Exploring the links between wood density and the growth rate of trees is therefore an important research topic.
The growth rate, which can be determined from an assessment of ring width in cores extracted by a Pressler increment borer, could prove to be an advantageous indicator of wood quality. Furthermore, it may be possible to use ring width to predict wood density and other parameters related to growth rate, such as the microfibril angle (Auty et al. 2013), along with mechanical properties such as modulus of elasticity (MOE) and modulus of rupture (MOR) (Zhang 1995; Dunham et al. 1999; Lasserre et al. 2009; Schneider et al. 2008). Correlations between wood density and ring width are not readily predictable. Research on conifers suggests that there may be possible correlations between wood density and cambial age (Guller et al.2011, 2012), and that these parameters are related to site conditions and climate (Drew et al. 2013; Ikvovic et al. 2013). Cambial age, therefore, may play a major role in the relationship between wood density and ring width. In much of the literature on conifers, wider growth rings are associated with lower wood densities because the percentage of earlywood was higher; hence, in forest plantations, the annual rings of logs should not be too widely spaced (Kubojima et al. 2008). In diffuse porous hardwoods, both positive and negative correlations were found between ring width and density; yet in some cases, no correlations were found (Zeidler 2012; Skarvelis and Mantanis 2013).
Many other studies have shown that density is related to additional variables associated with wood anatomy, which are not reflected in either ring widths (Wimmer 1995; Rathgeber et al. 2006) or the percentage of latewood (Adamopoulos et al. 2009). The growth rate of ring-porous hardwoods, which is expressed as ring width, has been positively correlated to wood density, i.e., wider growth rings are associated with higher wood densities (Zobel and Van Bujten 1989; Dobrowolska et al. 2011). These findings were attributed to the fact that the earlywood zone was nearly constant from year to year and the wider rings contained higher contents of dense latewood with fewer vessels (Zobel and Sprague 1998; Zobel and Van Bujten 1989; Adamopoulos et al. 2010a). However, a few studies suggest that this relationship is sometimes neither significant (Adamopoulos et al.2010b) nor predictable because of the influence of silvicultural management, variations due to the position inside the tree (Adamopoulos and Voulgaridis 2002), differences in cambial age, and other factors that can modify the expected correlation. Nevertheless some authors (Guilley et al. 2004, Bouriaud et al. 2004), were able to elaborate a model.
The object of the present investigation is chestnut, an economically important species in the Lazio region in Italy. The goal is to study the correlations of ring width (RW) and cambial age (expressed by chronological class) with specific gravity, shrinkages, and mechanical strength, and the correlations between density and shrinkages and mechanical properties. From this research, it may be possible to identify those parameters related to the growth rate of a tree that can be directly used in forests to predict wood quality. It is also possible that such research may be able to assess the effect of growth rate on more than just the microfibril angle (Auty et al. 2013) or the ratio of modulus of elasticity to modulus of rupture (MOE/MOR) (Dunham et al. 1999; Lasserre et al. 2009; Schneider et al. 2008, etc.).
Because chestnut is a ring-porous species, the results are subsequently compared with the few references available regarding either chestnut wood (Fioravanti 1999; Adamopoulos et al. 2010b), or similar hardwood species in comparable study situations.
EXPERIMENTAL
Materials and Methods
The study was carried out on chestnut trees collected from seven sites in Lazio, Italy. These sites belong to four different regions of high productivity: Cimini Mountains (A), Sabatini Mountains (B), Castelli Romani (C), and Lepini Mountains (D). The site characteristics have been described in detail in previous reports (Spina and Romagnoli 2010; Romagnoli and Spina 2013). A notable difference in these sites is that the Cimini Mountains, Sabatini Mountains, and Castelli Romani are all located on volcanic soils, whereas the Lepini Mountains are located on a calcareous substrate.
After four months of seasoning, important physical and mechanical characteristics of the wood were measured in accordance with the UNI ISO 3130, 3131, 3133, 3787, and 4469 protocols (1985). Disks were prepared with strips along the length of the radius being marked in 2 cm increments. Specimens measuring 2 × 2 × 3 cm were used to measure shrinkage, specific gravity by the gravimetric method, and compressive strength in the axial direction. For measurement of the MOR (i.e., bending strength), samples measuring 2 × 2 × 30 cm were used. Several physical and mechanical characteristics representative of each site and of the total region have been reported in a previous investigation (Romagnoli and Spina 2013); these characteristics were distinguished in both sound and shaken trees. For this investigation, only the wood specific gravity at a moisture content of 12% (ρ12), tangential and radial shrinkages (βt, βr), axial compressive strength (σ12), and static MOR were determined.
For each of the specimens prepared for measuring physical and mechanical properties, the number of rings present was counted within a 2 × 2 cm surface area. The ring width was the estimated by the ratio between the radial size of the specimen (2 cm) and the number of rings visible on the cross-cut. Specimens were excluded from further analyses when the rings could no longer be regarded as almost complete, and thus the analysis was carried out only on those samples without defects. The specimens were next divided into seven classes according to their average ring width, and the cambial age of each specimen was then measured. Afterwards, the samples were divided according to their cambial age class, as reported in Table 1.
Table 1. RW Class Number and Corresponding Interval of Ring Width, along with the Chronological Class and Corresponding Interval of Tree Age
The decision was made that rings below 10 years of cambial age would not be taken into account, in order to limit juvenile wood effects. Hence, the cambial age class begins with Class 3 rather than Class 1 in the following analysis. The ring width and chronology class were treated as independent variables in statistical analyses. Analysis of variance (ANOVA) was used to test whether or not the dependent variables of density (ρ12), tangential and radial shrinkage (βt and βr), compressive strength (σ), and MOR were significantly related to ring width or chronology class. Fisher test values (F) and the corresponding values of p<0.05 (*), p<0.01 (**), and p<0.001 (***) were determined, and the Kolmogorov-Smirnov test was used to verify which ring width class was significantly different from the others.
Cross-variance testing, which takes into account the interactive effect of chronology class and ring width, was also performed using ANOVA analysis. This analysis also took into account whether the samples were nested within trees according to their ring width class within a given chronological class. However, these results proved to be insignificant, and thus this variable was not considered in further analysis. A general linear model was also performed in order to estimate each physical and mechanical parameter according to the ring width and chronological class.
To transpose the results obtained from these specimens to the typical parameters used by foresters in the field, trees of comparable ages (25 to 30 years) were considered, and their diameter at breast height was measured. This measurement, however, was not possible to perform at any site other than Site A. The trees were then divided into seven diameter classes: 0–20, 21–25, 26–30, 31–35, 36–40, and >40 cm. The average values of specific gravity, tangential and radial shrinkages, compressive strength, and MOR for each of the trees were compared. A variance analysis of the physical-mechanical characteristics of the wood was taken into account, using the diameter class as an independent variable. The results of this analysis may therefore be considered a further approach to estimate the growth rate of a tree.
RESULTS
Effect of Ring Width and Chronological Class on Physical and Mechanical Parameters
The values for the physical and mechanical properties within each region have been previously reported (Romagnoli and Spina 2013). These results show that the physical and mechanical properties vary according to the site of origin. In Table 2, the average specific gravity at 12% moisture, tangential and radial shrinkage, compressive strength, and MOR are reported, along with the number of specimens represented in each average ring width class and the chronological class for each site.
The highest density value of 666 kg/m3 was found in the Lepini Mountains (Site D), while the lowest density value of 586 kg/m3 was found in the Cimini Mountains (Site A). The highest MOR value of 117 MPa was found in the Sabatini Mountains (Site B), and the lowest MOR value of 76 MPa was found in the Castelli Romani (Site A). On average, the highest shrinkage was found in the Sabatini Mountains (Site C), and the lowest value was found in the Lepini Mountains (Site D). Although there are no average ring widths for these samples, the specimen data used to measure specific gravity by gravimetric methods were inserted in Table 1 so as to place each specimen in a chronological class.
As can be seen from Table 2, the fifth ring width class was lacking at all sites, while the most represented ring width classes were the third and fourth. The Castelli Romani site had the highest number of specimens located in the sixth class.
The results from the ANOVA analysis (Fig. 1) show the effect of ring width class on density values. A significant decrease in density can be observed with increasing ring widths at Sites A and C; however, a different behavior was observed at Site D, which is located on a calcareous substrate. At Site D, increases in density were associated with an increase in ring width up to a threshold of ring width class 6 (6 to 7 mm), but above class 6, a decrease in density is observed.
Table 2. Mean Values (mv) and Standard Deviations (sd) of Each Physical and Mechanical Property
Numbers in brackets represent the number of investigated specimens for each RW class and each chronological class.
Fig. 1. ANOVA analysis of ring width (RW) class and chronological class showing the effect on specific gravity at Site A-Cimini Mountains, B-Sabatini Mountains, C-Castelli Romani, and D-Lepini Mountains. The numbers above the graphs are F values with significant probability: * p<0.05, ** p<0.01, *** p<0.001.
In this study, an increase in the ring width class is associated with a decrease in compressive strength and MOR at Sites A and C, while significant decreases were only observed in compressive strength at Site D (Fig. 2).
Fig. 2. ANOVA analysis of the effect of ring width (RW) class on compressive strength and MOR. The numbers above the graphs are F values with significant probability: * p<0.05, ** p<0.01, *** p<0.001.
Most studies of ring-porous hardwoods have shown frequent positive correlations between ring width and density (Zhang et al. 1993; Zobel and Van Bujten 1989; Guilley et al. 1999). Although divergences from this trend have been observed in the past (Zobel and Van Bujten 1989), further studies have shown that ring width relationships can depend on many other factors. Foremost among these is management, and both Sites A and C are characterized by high growth rates initiated via coppice management.
The results of this study are somewhat in agreement with data obtained on plantation-grown Quercus rubra; while in mature natural stands of the same species, some positive correlations were observed (Genet et al. 2013). In chestnut from sites in Greece (Adamopoulos et al. 2010a) and from other coppice stands in the Tuscany region in Italy (Fioravanti, 1999), the data shows a decreasing trend in density over a threshold of 6 mm in ring width. This is exactly the result that was obtained from Site C on volcanic soil, where the most evident decreasing trend in density, which is also significant from a statistical point of view, started from e ring width class of 6 mm. The effects of forest management practices were similarly not negligible in diffuse-porous hardwood. For example, studies of coppiced Betula pubescenstrees revealed a higher growth rate than seed-originated birches, though the management technique only caused minor differences in density that were insignificant with regards to industrial use (Luostarinen et al. 2009). Similar results, though lacking in statistical significance, were also observed in hornbeam (Samariha 2011), cherry (Nocetti et al. 2010; Nocetti et al. 2012), and yellow poplar (Mettanurak et al. 2010).
Coppice management cannot be the only reason for this high growth rate, because the relationship between ring width and density shows little difference in behavior at Site D that is also coppice managed, but stands on a calcareous substrate. The results therefore reveal a complexity in the relationship between ring-width and density in chestnut: the site characteristics, in this case related mainly to the soil and high growth rate of coppice management, play an important role in determining such correlation.
If one looks at the decreasing trend in density with increasing growth rate, one cannot exclude the possibility that it is influenced by the analytical methods used. In the literature, the relationship between density and ring width are typically investigated by means of X-ray densitometry, which allows detection of annual and inter-annual variability. Conversely, the methods used in this study were more focused on detecting large variations in wood quality that could affect its practical utilization, than on single ring widths or annual growths; therefore, it is expected that the information obtained differs from that obtained by X-ray densitometry.
The unexpected negative correlations between ring width and wood density observed at some sites of this investigation may possibly be explained by the effect of morphological anatomical parameters that can cause variations in wood density. Consequently, ring width may not be sufficient to explain variations in wood density, even when it is related to the width or percentage of earlywood and latewood (Wimmer 1995). That has been observed by Fioravanti (1999) to be the case in chestnut, in which decreasing density values explained anatomically by a lower cell wall thickness and the proportion of latewood alone could not explain the density variation.
The effect of chronological class on density was less pronounced than that due to ring width class, and general trends were difficult to detect. The most significant results were again seen at Site A, where decreases in density were associated with increases in the chronological class. A similar behavior was observed at Site D, but here additional increases in the higher chronological class were observed. Chronological class, i.e., tree age, takes into account the process of wood maturation, which is crucial when evaluating juvenile and mature wood (Zobel and van Bujten 1989; Zhang and Zhang 1991; Adamopoulos et al. 2011). Many of the difficulties that can be associated with age variations due to juvenile wood were avoided in this study by excluding specimens with cambial ages of less than 10 years from analyses, and this explains why only a minimal influence of chronological class on density was observed. At Site D, however, an increase in the density of the highest chronological class was observed because of wood maturation.
The correlation between ring width and mechanical properties may be related by the effect of ring width on density, especially when we consider the agreement among the experimental results of ring width-density-mechanical properties for Sites A and C. In site D, the ring width seems to have had a more direct effect on mechanical properties than on the density values. This result can be compared with that obtained in poplar trees, where the effects of growth suppression and release can be seen more readily in the mechanical properties than in the specific gravity (Mettanurak et al. 2010).
The effects of ring width class on shrinkage were significant only with respect to the tangential direction at Site A. Compared to the results obtained by ANOVA for density and mechanical properties, the correlation between ring width and shrinkage was less significant, and no specific trends were readily apparent (data not shown). This can be compared with results obtained with Carpinus betulus, where it was shown that some environmental factors that can interact significantly with growth rate do not have any significant effect on the physical properties of wood, such as shrinkage (Samariha et al. 2011).
Chronological class was found to have a significant effect on compressive strength and MOR at Site A (Fig. 3). It also had a significant effect on MOR at Site D, whereby increases in the chronological class were associated with lower MOR values until class 7, after which the mechanical properties actually increased. These trends were particularly apparent in class 4 and above (≥4 mm).
Fig. 3. ANOVA analysis of the effect of chronological class on compressive strength and MOR. The numbers above the graphs are F values with significant probability: * p<0.05, ** p<0.01, *** p<0.001.
Overall, the trends of the compressive strength and MOR curves were more similar for all sites other than Site D, where the trend of both compressive strength and MOR was every different. No relationships were found between chronological class and shrinkage values. Fluctuations in density due to chronological class have also been shown to be related to whether the tree stands were natural or managed (Genet et al. 2013).
In Table 3, the results of the Kolmogorov-Smirnov test show that the physical and mechanical parameters of each ring width class differed significantly from each other. At Site A, the most remarkable difference was observed for parameters starting at Class 2 and above. At the second ring width class, the density decreased significantly; meanwhile, above the second ring width class, a significant decrease in compressive strength and MOR were also observed. At Site B, changes were apparent by Class 4 when the density is compared to Class 2. At Site C, a significant decrease in mechanical properties by Class 4 was observed with respect to Class 1 and 2, while density changes were significant for Class 3. Much greater variability was observed for the correlations with ring width at Site D, although increases in density with ring width were also observed.
The analyses carried out on the wood samples were comparable with experiments carried out through the measurement of a standing tree, where in the diameters were grouped by diameter class. In Fig. 4, decreases in density and other mechanical properties are apparent as the diameter of trees comparable in age (25 to 30 years) is increased (i.e., a higher growth rate).
Table 3. Matrix of Significant Differences According to Kolmogorov Smirnov Test among each RW Class for Each Site
Parameters differing significantly at p<0.05 (*), p<0.01 (**), and p<0.001 (***) are reported for each RW class