Wood samples from H. brasiliensis were obtained from five clones (LCB510, RRIM600, IAN873, IAN717and GT1) from plantations located in the municipality of Selvíria, Mato Grosso do Sul State, Brazil. The locations have a mean annual precipitation of 1440 mm/yr and an annual average temperature of 23 ºC (Flores et al. 2016). The soil is classified as dystrophic Red Latosol (LVd) according to Santos et al. (2018).

The northward position of each selected tree was identified to standardize the collection of wood samples. Then, we felled five randomly selected 33-year-old trees per clone and cut discs 10 cm in thickness from each tree at breast height (D, 1,3 m from the ground). Tree height and DBH of selected trees are shown in Table 1. In each disc, we cut five samples in each strip from pith to bark. For a total of 25 samples per clone: 0 % (close to the pith), 25 %, 50 %, 75 %, and 100 % (close to the bark).

Table 1
Mean and standard deviation of DBH and tree height (HT) of Hevea brasiliensis.

Values in parentheses are standard deviation

Basic density was determined by the method of maximum moisture content ABNT 11941 (2003). Samples of 2 cm x 2 cm x 3 cm were saturated by treatment with a vacuum system for 72 h to obtain saturated volume of wood. In sequence, the samples were dried in a laboratory kiln to determine the oven-dried mass at 103 ºC ± 2 °C.

Volumetric shrinkage was determined according to the ABNT 7190 (1997). The samples were saturated in water, measured with a caliper, and oven-dried at 103 ºC ± 2 °C. The dry volume of each sample was then determined. The difference in percentage between the two measurements is the volumetric shrinkage.

Wood samples (2 cm³) were softened by cooking in water and glycerin in a proportion of (4:1) until they presented ideal conditions for sectioning. Histological sections 20 μm in thickness were obtained using Leitz 1208 and Zeiss-Hyrax S50 slide microtomes. Sections were clarified by washing in 60 % sodium hypochlorite to remove cell contents; stained with safranin; Provisional slides were mounted in 60 % glycerin for measurements (Johansen 1940).

In addition to the histological sections, dissociated wood was prepared according to Franklin method (Berlyn and Miksche 1976). Thin sticks were cut and placed in wheaton containers, containing 100 volume hydrogen peroxide solution and glacial acetic acid (1:1). The containers were sealed with adhesive tape and remained 48 hours in an oven at 60 ºC. Subsequently, the material was washed with running water and stained with 1 % alcoholic safranin. The terminology and characterization of wood followed the IAWA list (IAWA 1989). All anatomical measurements were obtained with a microscope Olympus CX 31 equipped with a camera Olympus Evolt E330 and a computer with image analyzer software Image-Pro 6.3. (Software Media Cybernetics 2021).

To evaluate the effect of clone x radial variation within the tree on physical and anatomical properties, the variance homogeneity test was initially performed through Hartley's test and later the F test of variance analysis according to the experimental design of randomized blocks. The F test was applied (P > 0,05), and the means were compared using Tukey’s test. The relationship between variables was evaluated using Pearson’s correlation. Statistical analyses were conducted using the SAS statistical program (SAS 1999).

The average value for basic density was 580 kg/m³ (Table 2). Our results showed that H. brasiliensis clones are considered moderately heavy, ranking it in the class C20 by ABNT 7190 (1997). Wood basic density varied according to clone type from 560 kg/m³ at GT1 to 600 kg/m³ at LCB510 (Table 3). The wood density averages (Table 3) are higher than those calculated by Chukwuemeka (2016). The basic density value obtained for clone RRIM600 is similar to that found by Raia et al. (2018). However, Santana et al. (2001) reported lower values for 40-year-old H. brasiliensis clones. These differences between density values, when compared to the literature, can be explained by such factors as genetics, different tree ages, and/or the local characteristics of each plantation (Chaendaekattu and Mydin 2018, Rungwattana et al. 2018).

Table 2
Analysis of variance of basic density (BD), volumetric shrinkage (ε_v), fiber length (FL), fiber wall thickness (FWT), vessel element length (VEL), vessel diameter (VD), vessel frequency (VF), ray width (RW), ray height (RH) and ray frequency (RF) of 33-year-old Hevea brasiliensis.

** significant at 1% level of significance; n.s. = not significant and CV_e= coefficient of experimental variation

The average value obtained for volumetric shrinkage of 8,32 % (Table 2) is considered low for rubber trees (Mainieri and Chimelo 1989). These values are lower than those verified by Raia et al. (2018).

Clones IAN717 and GT1 had the smallest volumetric shrinkage, and clone RRIM600 had the highest (Table 3). Our results differ from those of Santana et al. (2001), who reported lower values for volumetric shrinkage of clones IAN717 and GT1 compared to our study. The volumetric shrinkage values of clone RRIM600 are similar to those obtained by Lara Palma (2010).

Anisotropy values found for rubber trees are usually high; therefore, this species is considered very unstable for production of wooden furniture (Raia et al. 2018). However, some Eucalyptus species, which are already being used commercially, also present values of anisotropy very close to that of the rubber tree and, hence, may not be an obstacle to the use of this wood for certain purposes (Batista et al. 2010, Santana et al. 2001). Minimal variations in shrinkage and swelling along the stem in rubber wood were reported by Owoyemi et al. (2018). Rubber wood to be used in internal flooring must undergo thermal modification processes to improve dimensional stability (Emmerich and Militz 2020).

Table 3
Average of basic density (BD), volumetric shrinkage (ε_v), fiber length (FL), fiber wall thickness (FWT), vessel element length (VEL), vessel diameter (VD), vessel frequency (VF), ray width (RW), ray height (RH) and ray frequency (RF) of 33-years-old Hevea brasiliensis.

Values in parentheses are standard deviation. Means followed by different letters on the same column indicate different mean values for the Tukey test (at 5 % level of significance).

Only ray width did not differ significantly among the clones (Table 2). Fiber length of clone RRIM600 (Table 3) has similar values to the ones obtained by Ramos et al. (2018).

Clones LCB510 and RRIM600 showed the highest values of fiber wall thickness, which differed statistically from that of clone IAN873, which had the lowest value (Table 3). Fiber wall thickness (Table 3) was lesser than that found by Teoh et al. (2011) and Norul Izani and Sahari (2008) in their studies.

In general, clones in the present study showed fiber dimension values below the average found in the literature. However, for fiber length and fiber wall thickness, similar values were obtained by Ramos et al. (2018). In contrast, longer fibers with thicker walls were found in 20-year-old Hevea brasiliensis studied by Faria et al. (2019b) who reported that the rubber tree has potential for use in cellulose and paper production. However, wall fraction index was high, and flexibility coefficient was low which could drastically interfere in cellulose and paper production.

If there is a negative correlation between growth and fiber length according to Chaendaekattu and Mydin (2018), it may not be possible to simultaneously attain vigorous growth and longer fibers. Clone RRIM2020, which grew in a wider spacing, had shorter fiber length than trees growing in higher population density (Saffian et al. 2014). This growth relationship was tested with 100% radial position anatomy results and no significant relationship for our data was found. The population density of trees in our study was 500 tree/ha, and this density is the most used in H. brasiliensis planting in Brazil.

Vessel element length differed among clones, with clone LCB510 having the highest mean (769 µm) and clone IAN873 the lowest (670 µm) (Table 3). Vessel diameter was the same in clones LCB510 and RRIM600, but narrower in clone IAN873 (Table 3). These values are within the standard (70 µm - 224 µm) for rubber tree (Reghu 2002). Hevea brasiliensis wood usually presents large vessel diameter values (Schoch et al. 2004).

Vessel frequency was higher for clone IAN873 and lower for clone IAN717 (Table 3). Ray width did not differ among clones (Table 3). Ray height was taller in clone LCB510 and shorter in clone RRIM600, while ray frequency was higher in clone IAN873 and lower in clone LCB510 (Table 3). Values of 47 (µm), 523 (µm) and 8 (nº/mm¹), respectively, for width, height and ray frequency for H. brasiliensis wood submitted to latex exploration were obtained by Ramos et al. (2016). These values are, on average, higher than those observed in our study.

In general, we observed that clone LCB510 showed higher values for basic density and fiber wall thickness, but lower ray frequency. These values suggest high resistance in comparison to the other clones in this study. On the other hand, clone IAN873 had lower basic density and fiber wall thickness and higher vessel frequency and ray frequency, making it a material with less strong wood features. Thicker wall fibers and higher wood density strike a positive correlation previously found in Pittosporum undulatum wood by Longui et al. (2011). To explain, fiber cells are more frequent than other wood cells; thus, fibers are positively correlated to higher density owing to mass increase (Fujiwara et al. 1991).

Basic density and volumetric shrinkage tend to increase from pith to bark (Table 3). While this variation was not enough for significant differentiation to occur for basic density, it did occur for volumetric shrinkage (Table 3). No significant difference was noted between juvenile and adult wood for basic density by the presence of high levels of extractives in juvenile wood, which is usually closer to the pith (Severo et al. 2013).

Most tropical species have a tendency to present smaller cellular dimensions in the pith region compared to wood close to the bark. The same pattern occurs in rubber wood, with the exception of vessel frequency, where the reverse phenomenon normally occurs (Table 3). However, this trend was not significant for ray width and ray frequency. This same trend also occurred in Astronium lecointei (Melo et al. 2013).

Narrower and fewer vessels are observed in the pith region, while wider and fewer vessels are found in the region close to the bark (Figures 1a and Figure 1b). The pattern of variation in cell dimensions was very similar to that found by Lima et al. (2011) in Cariniana legalis and Melo et al. (2013) in Astronium lecointei.

Figure 1
Photomicrographs of Hevea brasiliensis wood, transversal sections (a): from the pith region and (b): from the bark region. Note thesmaller diameter and higher frequency of vessel (a) and the larger diameter and lower vessel frequency (b). Scale bar = 100 µm.

To better explain the relationships among basic density, volumetric shrinkage and cell dimensions, relative to radial position, Pearson's correlation analyses were performed (Table 4). Only ray dimensions failed to show a significant relationship with radial position. In addition, vessel diameter had a strong negative relationship with radial position (Table 4).

Based on these significant correlations, we performed regression analyses to verify the best models to explain these relationships. Almost all regression models showed positive correlations with radial position, except volumetric shrinkage. This model represented only 72 % of correlations among these variables (Figure 2 and Figure 3).

Table 4
Pearson’s Correlation Coefficient (PCC) obtained for correlations among the variables studied and radial position.

** significant at level 1% of significance; * significant at level 5% of significance and n.s. = not significant.

Basic density, fiber length, volumetric shrinkage and fiber wall thickness had very similar behavior, i.e., they showed a tendency of stabilization at a distance of 9 cm from the log radius. This indicates the occurrence of a transition between juvenile and adult wood (Figures 2a, Figure 2b, Figure 2c and Figure 2d).

Figure 2
Correlation between basic density and radial position (a), fiber length and radial position (b), volumetric shrinkage and radial position (c), and fiber wall thickness and radial position (d) of 33-year-old Hevea brasiliensis.

Ferreira et al. (2011) found the stabilization point around 4,0 cm - 5,5 cm in the log radius when analyzing fiber length of 50-year-old H. brasiliensis. This may have been the result of age difference among the analyzed trees since the presence of cells with larger dimensions close to cambium is related to tree aging (Wilkes 1988). Wood density of H. brasiliensis do not have any significant correlation with tree growth and fiber length, as these properties have strong genetic control (Chaendaekattu and Mydin 2018, Rungwattana et al. 2018).

Vessel element length and vessel diameter progressively increased toward the bark, while the inverse occurred for vessel frequency (Figure 3a, Figure 3b and Figure 3c). This occurred because juvenile wood in the pith region exhibits greater physiological activity, thereby producing a greater number of narrower vessels.

Consequently, narrower vessels in juvenile wood are related to tradeoff lower hydraulic conductivity and higher embolism resistance, since that ability of vessels to conduct water increases proportionally with diameter, but large vessels in adult wood, which results in higher hydraulic conductivity can be submitted vessel to embolism under high water potentials (Wheeler et al. 2005, Lachenbruch and McCulloh 2014, Santiago et al. 2018, Simioni et al. 2020).

Also, as the transition from juvenile to adult wood occurs, the bark region presents fewer wide vessels (Wilkes 1988, Melo et al. 2013). However, according to Santos et al. (2019), cell dimensions of rubber wood can vary, depending on the different types of wood (tension, reaction and normal) in each radial position sampled.

Based on the results, the evaluated rubber tree clones have the necessary potential for use in some industrial activities that do not require much physical resistance in civil construction or the manufacture of different types of decorative objects and handcrafts. This was also the conclusion of Raia et al. (2018) for H. brasiliensis wood based on the homogeneity of its physical properties along tree height. They also found that the rubber tree presents characteristics similar to those of other species already used commercially.

Figure 3
Correlation between (a) vessel length and radial position (b) vessel diameter and radial position (c) vessel frequency and radial position of 33-year-old Hevea brasiliensis.

^♠Corresponding authors: limailde@gmail.com