Results and Discussion

Changes of vessel diameter in reaction wood

The vessel diameter decreased significantly more in reaction wood of Euonymus alatus (p ≤ 0.01), Ilex crenata (p ≤ 0.05), Clerodendron trichotomum (p ≤ 0.05), and Daphne odora (p ≤ 0.01) than it did in opposite wood among the nine species tested. The highest decrease rate (DR) (16.7%) was observed in Euonymus alatus. Almost no significant differences in the vessel diameter were observed between reaction and opposite woods in the other species studied (Table 1). In general, the diameter of a vessel decreases with the formation of reaction wood in many angiosperms [5,11]. Our previous report revealed that the vessel number decreased in all nine species; therefore, the pore distribution in the transverse section of reaction wood of Daphne odora (figured porous) was discontinuous and had oblique to dentritic vessel arrangement [6]. In the present experiment, the diameter of the vessel in Daphne odora also decreased significantly and was accompanied by reaction wood formation (Figure 1). These observations indicate that discontinuous vessel arrangement in reaction wood of Daphne odora might be caused by the reduction of the vessel diameter as well as the vessel number. A similar trend was reported by Hiraiwa, et al. in reaction wood of Osmanthus fragrans (figured porous) [11]. The decrease of the vessel diameter and number in reaction wood is a general trend of the cambial response to a gravi-stimulus. Various DRs of vessel diameter in reaction wood were observed in nine species, irrespectively of differences in the pore distribution pattern.

Table 1: Changes in the morphological characteristics of reaction wood cells on the upper side of leaning branches of nine Japanese hardwood species.

*, significance at 5% level (p ≤ 0.05) and **, significance at 1% level (p ≤ 0.01) by t-test; ns: No Significance; DR: Decrease Rate; IR: Increase Rate.

Figure 1: Vessel distribution in the microphotographs of transverse sections of Daphne odora. A) An oblique to dentritic vessel arrangement in opposite wood. B) Discontinuous vessel arrangement in the reaction wood. Scale bars, 500 µm.

Changes of fiber diameter in reaction wood

With the formation of reaction wood, the fiber diameter significantly increased in Rhododendron wadanum, Daphne odora, and Enkianthus subsessilis and decreased significantly in Euonymus alatus and Ilex crenata. The highest Increase Rate (IR) and DR were observed in Daphne odora (22.6%) and Euonymus alatus (12.9%), respectively (Table 1). In the four other species, Euscaphis japonica, Viburnum dilatatum, Ilex macropoda, and Clerodendrum trichotomum, no significant differences in the fiber diameter were identified between reaction and opposite woods. Chow reported that tension wood fibers were narrower than those of normal wood in beech [12]. However, Onaka pointed out a larger tangential diameter of tension wood fiber in some species [1]. Almost no differences were found in the diameter of the fiber tracheid among the upper, lateral, and lower sides of inclined trunks in two Magnolia species [5]. The present experiment suggested that variation in the change of wood fiber diameter depends on the species because the unique nature of the trend was not observed in the fiber diameter changes in the reaction wood of all nine species studied here.

Changes of fiber wall thickness in reaction wood

The thickness of the fiber wall significantly (p ≤ 0.01) increased in reaction wood of Euscaphis japonica and Ilex macropoda, while it significantly (p ≤ 0.01) decreased only in Daphne odora with the formation of reaction wood. The highest IR and DR were observed in Euscaphis japonica (77.4%) and Daphne odora (30.4%), respectively. In other species, significant changes in the thickness of fiber wall were not observed in reaction wood (Table 1). Conspicuously thickwalled fiber in reaction wood with G-fiber was found in arborescent angiosperms [13]; in contrast, thick-walled fiber was observed in normal wood of Fagus sylvatica when reaction wood fibers contained a G-layer [12]. On the other hand, almost no differences in the thickness of fiber tracheids without a G-layer were found among the upper, lateral, and lower sides of two Magnolia species [5]. In the present study, the fiber wall thickness of all nine species without a G-layer varied from species to species. Based on the observation of the present investigation, it is suggested that changes of fiber wall thickness are not always associated with either the presence or absence of G-layer in reaction wood fiber.

Changes of length of vessel element and wood fiber in reaction wood

The length of vessel element significantly decreased with 20.7% only in reaction wood of Daphne odora among nine species examined (Table 1). The contrast observation was reported in Populus euramericana by Jourez, et al. [14], who observed that the vessel elements were longer in tension wood than in opposite wood.

The length of wood fiber significantly increased in Enkianthus subsessilis and Clerodendrum trichotomum with the formation of reaction wood, the highest increased rate (13.3%) being found in Enkianthus subsessilis. On the other hand, the length of wood fiber decreased significantly in Euscaphis japonica and Daphne odora, and the highest DR of 38.2% was observed in Daphne odora. No significant differences in the fiber length between reaction and opposite woods were observed in the other five species (Table 1). In a comparison of normal wood fiber length, tension wood fibers were sometimes reported to be longer, sometimes shorter, and sometimes with no marked difference [14-16]. On the basis of an experiment on fiber length in tension wood of some Eucalyptus species, Wardrop reported that eccentric growth is generated with more formation of shorter length wood fibers [17]. In a previous report, we demonstrated that the growth increment was the highest in Daphne odora [6]; on the other hand, the shortest wood fiber formation in reaction wood of this species was observed in the present experiment. Therefore, our observation in the case of Daphne odora is in agreement with the findings of Wardrop [17].

Changes of microfibril angle of S2 layer in reaction wood fiber

In a previous study, we demonstrated that the microfibril arrangement of the S3 layer in both reaction and opposite wood fibers was a flat S-helix [6]. In the present study, we examined the changes of the MFA in the S2 layer of wood fibers owing to reaction wood formation. The MFA of the S2 layer in wood fibers significantly (p ≤ 0.01) decreased in Euscaphis japonica and Ilex crenata. The highest DR (28.3%) was observed in Ilex crenata. In other species, almost no significant differences in the MFA change of S2 layer were observed between reaction and opposite wood fibers (Figure 2).

Figure 2: Decrease rate of the MFA in the S2 layer of reaction wood fibers in nine Japanese hardwood species. **, significant at 1% level (p ≤ 0.01) by correlation coefficient test; ns: No Significance.

It has been reported that the microfibrils of a G-layer oriented parallel or nearly parallel to the longitudinal axis of a cell in many angiosperm trees [18]. The tendency of decreasing MFA has also been observed in the S2 layer of reaction wood where G-layer is absent. This trend is in agreement with the observation of Yoshizawa, et al. in some primitive angiosperm trees lacking either G-layer or S3 layer in the secondary wall of reaction wood fiber [5]. This tendency showing the decrease of MFA in the S2 layer of reaction wood fibers corresponds to a phenomenon in which tensile stress at longitudinal direction increases in reaction wood that maintains or recovers the position of the inclined stems or branches to appropriate orientation [7]. The present results confirmed that the MFA of the S2 layer in wood fibers might not always decrease due to reaction wood formation in some species having an S3 layer but lacking a G-layer in reaction wood fibers. Therefore, the mechanisms are unknown for these species to maintain an appropriate position for a leaning stem or branch. Therefore, a more detailed investigation is needed to clarify the changes in the MFA in the S2 layer in comparison to those in the other species with a three-layered secondary wall in wood fiber without a G-layer.

Relationship among reaction wood characteristics

In the nine species studied here, we previously examined growth eccentricity, change in vessel number, and lignin content by the formation of reaction wood [6]. In the present study, we investigated changes in other complementary reaction wood characteristics, such as vessel diameter and length, fiber diameter, wall thickness and length, and MFA in the S2 layer of wood fiber. Using all these reaction wood characteristics of nine species, relationships were established in a) growth eccentricity and other reaction characteristics and b) the decrease rate of lignin and other reaction characteristics (Table 2).

Table 2: Relationships among reaction wood characteristics of nine Japanese hardwood species.

*, significance at 5% level (p ≤ 0.05) and **, significance at 1% level (p ≤ 0.01) by correlation coefficient test; ns: No Significance; MFA: Microfibril Angle; DR: Decrease Rate; IR: Increase Rate.

Relationships between growth eccentricity and other reaction characteristics: A significant negative (r = -0.635, significant at 5% level) relationship was found between eccentric growth and IR of fiber wall thickness (Table 2). There are some reports that indicated a relationship between fiber wall thickness and growth stress. Fang, et al. revealed that the correlation between double cell wall thickness of fiber and growth stress was negative, which has also been confirmed by another study [8]. Although several studies pointed out the relationship between fiber cell wall thickness and growth stress, the present result revealed a negative correlation between growth eccentricity and cell wall thickness of wood fiber.

Relatively high positive correlations were found between the growth eccentricity and DR of the vessel diameter (r = 0.613) and length (r = 0.486). Correlations between growth eccentricity and fiber diameter and fiber length have been observed [14,17], but not in the nine species without G-fiber in their reaction wood, which showed relatively high relationships between the growth eccentricity and DR of the vessel diameter and length.

A positive but not significant correlation was also observed between the growth eccentricity and DR of lignin content (r = 0.564). This fact is a common phenomenon in the reaction wood anatomy of many angiosperms. For example, Yoshizawa, et al. reported that tension wood showed a relatively low lignin content when a growth increment was observed on the upper side of a leaning stem or branch of Magnolia obovata and M. kobus [5].

On the other hand, as shown in Table 2, a high negative correlation was observed between the growth eccentricity and DR of MFA (r = -0.617), although it was not significant. This result was opposite to a report by Yoshizawa, et al. [5], who speculated that eccentric growth occurred on a leaning stem of Magnolia species having neither an S3 layer nor a G-layer in their reaction wood fiber but MFA decreases in the S2 layer of reaction wood fiber, which means that the relationship was positive between eccentric growth and DR of MFA in the S2 layer. In the present study, this opposite trend of reaction wood characteristics is considered to occur with the presence of the S3 layer in reaction wood fiber.

Relationships between DR of lignin content and other reaction characteristics: A significant positive correlation coefficient (0.869, p ≤ 0.01) was observed between the DR of lignin content and that of vessel number (Table 2). In previous studies, there was a common phenomenon in which tension wood contained a lower proportion of vessels than normal wood [5]. The present study suggests that the decrease in the number of vessels due to reaction wood formation is a reason for the decreasing lignin content. Xu, et al. reported that the vessel wall and middle lamella are more highly lignified than the secondary wall of wood fiber; therefore, it is possible that the decrease in the number of vessels resulted in the decreased lignin content in reaction wood [18].

In the present study, a relatively high negative correlation coefficient was recognized between the DR of lignin content and DR of MFA (r = -0.516, not significant). The opposite result was observed in the relationship between the DR of lignin content and that of the MFA in reaction wood of Liriodendron tulipifera with G-layer-like fiber [6,8]. These researchers observed that the lignin content in wood fibers of tension wood decreased with decreasing MFA, i.e., a positive correlation exists between those of two characteristics.