Location and description of the agroforestry system

Two populations of Cedrela odorata (Spanish cedar) planted in an agroforestry system were studied in the community of Peralta, Turrialba, Cartago, Costa Rica (Fig. 1a). The denominated populations 1 and 2 were planted in 2009, and the sampled trees were felled in 2016. At the sampling time, the trees were 7 years old. The seeds of both populations came from the same provenance. The area where the trees originated is the Caribbean slope of Costa Rica (humid zone of Costa Rica), where precipitation levels are on average 3759 mm. The area does not present a well-defined dry season, and the months of March and April present the least amount of precipitation. The agroforestry systems consist of a plantation of Theobroma cacao (cacao) at 3x3 m with C. odorata trees as shade. The plantations are between coordinates 9° 58' 1.06"N - 83° 36' 43.97"W and 9° 57' 13.72"N - 83° 37' 12.45"W. The two plantations are easily identifiable by the type of bark: population 1 has a smoother bark (Fig. 1b) than population 2, which has a more cracked bark (Fig. 1c).

Fig. 1 (a) Agroforestry system of Cedrela odorata trees with Theobroma cacao, (b) tree from population 1, and (c) tree from population 2. 

Sampling

Randomly, 9 trees from population 1 and 11 trees from population 2 (N total = 20) were selected to study morphological characteristics and wood properties, according to the methodology proposed by ^{Moya and Muñoz (2010)}. The samples were obtained from these trees. The dasometric characteristics of the two populations are detailed in Table 1. The selected trees had straight trunks, normal branching and had no symptoms of disease or plague. A sample of a cross-section of approximately 3 cm thickness was cut from the base of the tree at 1.3 m height. From each tree selected, two logs of 1.3 m were obtained from the base of the tree until the DBH, and another from the DBH to a 25% height. Subsequently, cross-sections were taken at 25%, 50%, 75% and 100% of commercial height (Fig. 2A). The North-South direction was marked on each cross-section.

Fig. 2 Tree sampling patterns (A), cross-section measurements obtained from each tree (B) and sampling to determine the physical properties and amount of wood shrinkage (C-D). 

Determination of the percentages of heartwood, sapwood, bark and pith

The total diameter, diameter without bark, pith diameter and heartwood diameter (Fig. 2B), were measured on a transverse line drawn in two directions (north-south and east-west). The averages of all diameters were calculated using the two measurements from each transverse sample. The total heartwood area and pith from the cross-section were calculated as the geometric area of the circle. Bark thickness was obtained by calculating the difference between the total area and the area without bark. Then, its percentage was obtained by calculating the difference between total area and the area without bark, expressed as a percentage for the total area of the cross-section. The eccentricity of pith was determined by using transverse lines from the north-pith and east-pith and was applied using the method proposed by ^{Moya, Araya and Vílchez-Alvarado (2008)}.

Determination of the green density, specific gravity, shrinkage and green moisture content

A 3.0 cm-thick disk (including the pith) at different heights was cut from the cross-section, which was divided into two subsamples to study the physical properties of the wood (Fig. 2C- D). Each piece was weighed, and its volume measured. Volume was measured with water displacement method, according to ASTM D2395-14 standard (^{American Society for Testing and Materials, US [ASTM], 2014a}). Then, pieces were dried at 103 ºC for 24 hours. Afterwards, their weights and volumes were measured again. The green density was calculated as the green mass divided by the green volume, and the specific gravity was calculated as the oven dried weight divided by the green volume.

Volumetric, tangential and radial shrinkage were determined. From all transverse disks of 3.0 cm thickness, after cutting the central part (Fig. 2C), specimens to measure the shrinkage in the tangential and radial directions (Fig. 2C) were obtained. Their lengths were measured in green condition; then, the pieces were placed in the oven for 24 hours at 103 ºC and measured again to determine the amount of shrinkage that occurred. The total volumetric shrinkage was determined for the same transversal pieces obtained from the central part of the cross-section using the dimensional difference between green and dried conditions (0% moisture content), which was determined by measuring the displacement of water in both moisture conditions. For the moisture content, the specimens were weighed in the green condition, then dried at 103 °C for 24 hours and weighed again to calculate the moisture content.

Mechanical properties

From the logs obtained from the trees (Fig. 2A), 1-2 specimens were obtained to determine the modulus of elasticity (MOE) and the modulus of rupture (MOR) in bending, tension stress (parallel and perpendicular to the grain), the shear stress parallel to the grain, the longitudinal compression stress and the Janka hardness (axial and lateral direction). The samples were prepared following the norm ASTM D143-14 (^{ASTM, 2014b}), and from each tree, 1-2 specimens (2 specimens × 9 or 11 trees = 18 or 22 specimens) were extracted.

Statistical analysis

The assumptions for homogeneous variances (Levene Test) and normal distribution through the Shapiro-Wilks test were verified for all the properties evaluated. A t-test was used for two samples assuming equal variances with a 95% confidence level. Regression analyses were applied to the characteristics of the evaluated wood (heartwood, sapwood and bark percentage, pith eccentricity, green density and specific gravity) according to the tree height, where tree height was the independent factor and the properties of wood the dependent variable.

Morphological properties

The populations analyzed have similarities in some of the morphological characteristics of 7-year-old trees evaluated at DBH (Fig. 3). The diameters of population 2 trees were greater than those recorded for population 1, while the bark percentage and pith percentage were higher in the trees of population 1. The percentage of sapwood and heartwood and the eccentricity of the pith did not present differences between the trees from both populations in the agroforestry system (Fig. 3).

Fig. 3 Averages of the morphological properties of two populations of 7-year-old trees at the diameter at breast height in Cedrela odorata agroforestry systems with cacao. 

Legend: Different letters indicate significant differences at the 95% confidence level for the same height. Values in parentheses represent standard deviations.

The analysis of the percentage of height shows that the heartwood percentage (Fig. 4a) and eccentricity (Fig. 4e) at 7 years old do not present significant differences between the populations at any height. The percentage of sapwood (Fig. 4b) and pith (Fig. 4d) only presented differences in height at 100%, being higher in population 2. The bark percentage shows differences along the entire height profile between both populations at 7 years old, being statistically higher in population 1 than population 2 (Fig. 4c).

Fig. 4 Variation of the percentages of heartwood (a), sapwood (b), bark (c), pith (d), and pith eccentricity (e) with the height (%) of the two populations of 7-year-old Cedrela odorata trees growing in agroforestry systems with cacao. 

Legend: Different letters indicate significant differences at a 95% confidence level for the same height.

The mathematical modeling of heartwood, sapwood, bark, pith, and pith eccentricity by height revealed that the heartwood percentage of the trees from both populations of 7-year-old trees decreases logarithmically (Fig. 5a), but the sapwood percentage and bark percentage (Fig. 5b, 5c) showed a polynomial tendency of grade 2, as well as the percentage of eccentricity (Fig. 5e). However, the eccentricity showed no differences between heights. Pith trends differed in the trees from both populations. In trees from population 1, the tendency was polynomial grade 2, while for trees from population 2, the trend was logarithmic (Fig. 5d).

Fig. 5 Trends in the morphological properties of two populations of 7-year-old Cedrela odorata trees growing in agroforestry systems with cacao. 

The best-fit equations are presented in Table 2. There is great variation within the data for all the properties analyzed in relation to height, since the percentages of R² are low for all equations obtained. The f-value indicates that morphological characteristics and physical properties are statistically significant because the f-value is lower than the α value (0.05) with differences between heights, except for the percentage of bark in population 1 and the percentage of eccentricity in both populations.

Physical properties of wood

The physical properties measured at DBH showed that the volumetric shrinkage and green moisture content are statistically higher in trees from population 1 than population 2 at 7 years old, but the tangential and radial shrinkage showed no significant differences between the trees from both populations (Table 3). The shrinkage ratios were lower than 2 in both populations.

At the height of the DBH in trees at 7 years old, there is variation between the populations in green density (Fig. 6), which was statistically greater in trees from population 1 than population 2; however, for specific gravity, there are no differences between the populations (Fig. 6).

Fig. 6 Average of the green density and specific gravity at the diameter at breast height of two populations of 7-year-old Cedrela odorata growing in agroforestry systems with cacao. 

Legend: Different letters indicate significant differences at the 95% confidence level for the same mechanical property. Values in parentheses represent standard deviations.

The variation of the green density and specific gravity of the 7-year-old trees in relation to height showed an irregular behavior (Fig. 7). The green density was higher in the lower parts of the trees, decreasing in value to the height at 25%. The green density increased with the height in both populations starting from this height (25%) (Fig. 7a). The specific gravity had variations in trees of both populations with height, with a tendency to increase proportionally with the height (Fig. 7b). The green density showed no differences at a height of 75% between the populations, but the trees from population 1 had greater density at the other heights (Fig. 7a). Likewise, the specific gravity showed no significant differences between populations at 7 years old.

Fig. 7 Variation with respect to the percentage of height of the green density and specific gravity in two populations of 7-year-old Cedrela odorata trees growing in agroforestry systems with cacao. 

Mathematical models showed variations in green density and specific gravity, with a polynomial tendency grade 2 for both populations at 7 years old. There is an irregular pattern for both variables regarding height, having lower values at intermediate heights of the trees and a tendency to increase moving upwards from the intermediate part of the tree. The best-fit equations are presented in Table 4 from the previous models of the green density and specific gravity within each population, which showed that the models are statistically significant (f-value lower than α value), with the exception of the specific gravity of population 2, but they have low correlation values (R²).

Mechanical properties

Table 5 shows the mechanical properties for 7-year-old trees from two populations of C. odorata in an agroforestry system. MOE and MOR in bending, the maximum longitudinal compression stress was statistically higher in population 2. The axial and lateral Janka hardness and shear stress values did not present differences.

Morphological characteristics

Trees of C. odorata at 7 years old tended to grow better in agroforestry systems than pure plantations due to the intensive management, type of soil, topography and type of fertilization received (^{Navarro, Montagnini, and Hernández, 2004}). Therefore, greater development at 7 years old is expected in the diameter of trees growing under these conditions. ^{Leyva and Erguera (2014)} showed that the diameter varied between 10 and 15 cm for 14-year-old C. odorata trees growing in agroforestry systems; these values are lower than the range of values obtained in this study (13.9 cm - 19.7 cm). Greater growth was observed in population 2 (Fig. 3), and this could have occurred because this population is probably more likely to grow in diameter than population 1, as usually occurs in different varieties due to genetic considerations (^{Navarro et al., 2002}).

As for the percentage of sapwood in 7-year-old trees, values between 56.35 and 60.20% were obtained. The heartwood percentage determines the quality of the wood since chemicals and extractives that increase the wood quality are synthetized in mature trees during the heartwood formation process (^{Ypushima, Salcedo, Manríquez, Silva, Zamora and Hernández, 2014}; ^{Flórez et al., 2014}). Therefore, it is important that trees present a high percentage of this type of wood and thus guarantee the quality of the final product when the wood will be used in furniture especially since juvenile trees have a low heartwood content.

As for bark percentage, the minimum percentage reported for tropical species is 2% and the maximum is 25% (^{Hegde, Kalkoor, Jha and Thakur, 2014}). Population 1 of C. odorata at 7 years old achieved a bark percentage of 21.3%, and population 2 of 12.8%, therefore these values are within the previous range and hence, considered normal. The bark percentage is mainly influenced by temperature and is inversely proportional to precipitation (^{Rosell and Olson, 2014}). Population 2 showed a smaller area of bark (Fig. 3), which would indicate it is the population better adapted to the area of Turrialba which is considered a rainy area, unlike population 1 that achieved a higher percentage of bark which is less well adapted to the weather conditions of the place. ^{Fasola, Olagunju and Robert (2014)} indicate this may be due to morphological differences within the species. Moreover, the bark percentage tends to vary in tree height and it does not vary based on diameter (^{Serrano, Moya, Berrocal, Trejos and Foglia, 2015}). This pattern of height is observed in both populations, however there are differences between the diameters when an increase in the bark occurs along the stem (Fig. 4)

The decrease of heartwood with height is common in trees (^{Moya, Salas, Berrocal, Valverde, 2015}; ^{Tenorio et al., 2016}) and the trend found in 7-year-old C. odorata trees growing in an agroforestry system with cocoa (Fig. 4a) is congruent with this. Meanwhile, the decrease in sapwood percentage with height (Fig. 4b, 4c) is congruent with data found for trees growing in forests (^{Parolin, Müller and Junk, 2008}). The results found for pith percentage have a variation of 0.4% to 0.7% (Fig. 4d), which are similar to those found for different tropical species growing in pure plantations (^{Tenorio, Moya, Salas, and Berrocal, 2016}).

Prediction models

In both 7-year-old populations it was observed that there is a low relationship between height and morphological characteristics or physical properties (specify gravity and green density), lower value of R² (Tables 2 and (4). This could have occurred because there is high variation within each tree, or because another variable must be included in the equation that explains the low variability within each population. The morphological and physical differences within each population are evident since in several cases both populations follow a different trend (Fig. 4d, 4e), therefore each population has its own behavior.

Physical properties

^{Christoforo, Almeida, Almeida, Santos, Panzera and Lahr (2016)} report volumetric shrinkage values of 8.68%, tangential shrinkage of 5.02% and radial shrinkage of 3.42% for mature Cedrela sp. trees in pure plantations. These values agree with those obtained in this study for population 1 at 7 years old but not for population 2 (Table 3).

The values of C. odorata trees at 7 years old are different from the values of another species from the same genus. Cedrela fissilis presents values for radial and tangential shrinkage of 12.8% and 17% respectively (^{De Avila,Gatto, Stangerlin, de Cademartori, Weinert and de Avila, 2013}), a much higher percentage than those presented in the two populations in this study. Tangential shrinkage (4.06%) of trees in agroforestry systems was higher than in S. macrophylla (1.4%) and C. alliodora (5.6%) (^{Anoop, Jijeesh, Sindhumathi and Jayasree, 2014}) for trees growing in plantations.

Cedrela fissilis reported a volumetric shrinkage of 13.2% (^{De Avila et al., 2013}), above the value obtained in both populations of C. odorata. Population 2 shows lower values than those previously reported by ^{Christoforo et al. (2016)} in volumetric shrinkage (6.82%) and radial shrinkage (2.53%), but the tangential shrinkage is similar to those found in the literature (4.92%).

The specific gravity obtained in the present study (Fig. 6) for both populations at 7 years old is within the range reported by ^{Van der Hout (2015)} of 0.25 - 0.50 for adult trees. A disadvantage of low specific gravity is that it can be categorized as soft and this type of wood does not have good mechanical resistance, however the wood is appreciated for its aesthetic value (^{Flórez et al., 2014}).

The green density tends to become smaller as height increases (Fig. 7a). The increase in density in medium zones of the tree may be due to the fact that there are more branches in that area (^{Trautenmüller et al., 2014}). The increase in specific gravity with height (Fig. 7b) does not present a common pattern in most species (^{Tenorio et al., 2016}); however, some tropical species have this type of pattern (^{Moya and Muñoz, 2010}, ^{Tenorio et al., 2016}).

Mechanical properties

The evaluation of the properties helps to make decisions regarding different objectives of use for wood in the industry (^{Hegde et al., 2014}). Among the possible uses is light construction (^{Van der Hout, 2015}). The results obtained for both C. odorata populations at 7 years old growing in an agroforestry system with cocoa could indicate their possible use in light construction. For C. odorata, 65.9 MPa is reported for MOR (^{Bárcenas-Pazos, Ortega-Escalona, Ángeles-Álvarez, and Ronzón-Pérez, 2014}) and 6700 MPa for MOE (^{Venson, Guzmán, Talavera and Richter, 2008}); in this study, with lower MOR and MOE values were found for both populations (Table 5).

The longitudinal compression stress reported for C. odorata is 26.87 MPa (^{Bárcenas-Pazos et al., 2014}). The values obtained in this study are lower than those. The value for the shear stress (Table 5) is lower than those reported for C. odorata by Bárcenas-Pazos et al. (2014). In all cases, this is certainly due to the fact that the populations studied here are of juvenile wood from young trees and those found in the literature are from mature trees.

Variation within the Cedrela odorata populations

The vegetative material used for this agroforestry system was material obtained from a seed stand, that is, seeds from the same source were used to establish the plantation. The lack of knowledge about the dynamics of the initial population does not allow to explain the difference in the development of trees in the plantations studied, which is expressed in differences between individuals. In this case, for example, there is evidence of differences between the bark of different trees (Fig. 1).

The variation between two populations of 7-year-old C. odorata is mainly explained by genetics. For example, the migration of genetic material (pollen and seeds) changes the species due to adaptation to an external change and loss of information due to excessive use of trees. In addition to this, the size of the sampled population has an influence in the results obtained since small populations harbor less variability than large populations; furthermore, fragmentation leads to cross-breeding between related individuals and loss of genetic information. This is why different populations occur, as was the case in this study. This has repercussions on the quality of the wood, mainly morphological aspects such as heartwood, pith and bark. Differences in population growth in a single site with evidence of distinction in the physical characteristics of the material is due to differences in the genetic base (^{Cavers, Navarro, and Lowe, 2004}). The variability of the species and their differentiation within the same population in a geographic area can occur due to the environmental changes that produce modifications in the trees at the genetic level (^{Barbour, O'Reilly-Wapstra, Little, Jordan, Steane, Humphreys, Bailey, Whitham and Potts, 2009}). This occurs over time through the adaptation of a population to a specific site (^{Navarro et al., 2002}). Variations in the environment, such as over-exploitation, reduce the size of the population and therefore produce loss of the species’ own characteristics. The genus Cedrela tends to present diversification within the populations due to historical changes that have produced fragmentation of the initial population (^{Soldati, Fornes, Van Zonneveld, Thomas and Zelener, 2013}), and consequently, this diversity is transmitted to agroforestry systems, as in the present study. The maintenance of the characteristics causes differences between populations.

Several studies have shown morphological changes in individuals within the same species, and these studies concluded that we can discuss different varieties for populations (^{Fajardo and Piper, 2011}). The adaptation occurs through changes in the soil in the same site, where morphological differences have been found within a population (^{Misiewicz and Fine, 2014}) and in the quality of the wood because of the differences previously presented. The most marked differences were in diameter, bark, pith, volumetric shrinkage, green density, MOE, MOR, longitudinal compression and perpendicular tension.