臺灣欒樹的抗張材在傾斜苗木與枝條的解剖構造與生物力學功能

(1)

國立臺灣大學生命科學院生態學與演化生物學研究所博士論文

Institute of Ecology and Evolutionary Biology College of Life Science

National Taiwan University Doctoral dissertation

臺灣欒樹的抗張材在傾斜苗木與枝條的解剖構造與生物力學功能

Anatomical Structure and Biomechanical Function of Tension Wood in Inclined Seedlings and Branches

of Koelreuteria henryi Dummer

洪麗分 Li-Fen Hung

指導教授：黃玲瓏博士

Advisor: Ling-Long Kuo-Huang Ph.D.

中華民國 106 年 6 月

June, 2017

(2)

(3)

致謝

能在不惑之年結束之前完成博士學位，真的要歸功於老天爺的安排以及身邊師長親友的鼓勵與陪伴。首先要感謝我的指導教授黃玲瓏老師，我們的師生緣起於 1991 年，有幸成為黃老師的第一個碩士生，感受到老師亦師亦母的指導與照顧。

畢業之後，不忘師恩教誨，努力在學業、家業與事業中取得平衡。因緣際會之下，

2008 年回到實驗室，成了黃老師的第一個博士生，亦師亦友的黃老師除了傳道、

授業與解惑外，更帶著我成長並分享生活中的喜怒哀樂，衷心感謝您對我的愛護與付出。我也要謝謝黃彥三老師在我論文研究期間的諄諄教誨與耐心指導，感謝您帶我進入生物力學的世界，讓我的論文研究得以順利進行。謝謝高文媛老師、邱少婷老師與簡慶德老師擔任我的資格考口試委員，提供寶貴的意見，引導我建立完整的研究計畫；更要感謝邱少婷老師、蕭淑娟老師、簡慶德老師、王兆麟老師與林法勤老師在百忙之中，擔任我的口試委員，細心審查並協助我修正論文的細節，讓我的博士論文得以完成。也謝謝陳淑華老師在課業上的教授與生活上的關懷，李鳳鳴老師在我擔任普植實驗助教時的協助與鼓勵。

感謝植物解剖研究室的大學姊陳香君老師，您給予的不只是研究技術與材料的支持，還有生活上點點滴滴的協助。有了您，生活變得更容易。謝謝萬能學長與秋容的鼓勵、金梅的技術協助，以及學弟妹們的支援與陪伴：沒有謦竹，我的論文難以產生，謝謝你一路陪著我做研究，寫論文；感謝瑋育和育銘陪著我採樣與切片；

謝謝資棟、瑋庭、彥佑、靖玟、哲瑜、綱祐、恬君和名偉，是你們讓我跟得上時代，

保持青春與活力。感謝高家的學弟妹們，雅倫、孟穎、泰中、佳娟、顯淳、怡清、

譯泯、麗智、渼晨、昆松、彥治…讓我有地方串門子，調劑生活。

要特別感謝所辦月鈴助教協助打理從入學到畢業過程中的學業與生活瑣事。

也要感謝博班的同學們，妤馨、欣怡、孟穎、冠霆、敬舒、勵婉，相伴一起走過博班生活的酸甜苦澀。感謝和我一起帶實驗課的所有助教們，特別是月鈴、聖傑、英超、馨儀、玉玲和小米，我從你們的教學方式中獲得啟發，勇於嘗試新方法；也謝謝亞臻、以君和佩穎支援生物電顯實驗；更要謝謝被我帶過實驗課的學生們，是你們活化我的教學細胞，讓我發揮所長，享受工作的樂趣。

感謝我的家人，謝謝爸爸媽媽無為而治的教育方式，讓我得以適性發展；謝謝大哥大嫂的厚愛，也謝謝麗雀、麗文兩個妹妹的陪伴與分享。我把最深沉的感激留給與我相依相伴的先生耀銘和兒女俐安與柏宇，是你們無限的包容、支持與愛，讓我得以實現自我，謝謝。最後，給在天上的弟弟源倉，願與你一起分享這份成就與喜悅。

(4)

中文摘要及關鍵詞

抗張材在維持被子植物的生物力學穩定上扮演重要的角色，因此在研究傾斜主幹恢復直立以及枝條角度變化的機制時，必須同時關注抗張材的生物與物理特性。本論文以台灣欒樹為材料，研究在恢復直立的傾斜樹苗主幹中以及樹冠不同角度的枝條內，抗張材的形成、構造與分布等生物特徵以及應變分布的物理特徵，以期了解抗張材在傾斜主幹與不同角度枝條中所扮演的角色。

在人為傾倒的兩年生台灣欒樹樹苗主幹中，抗張材的生成與恢復直立的過程歷時約三個月。透過插針法，我們確認：當樹苗傾倒時，主幹上側的形成層區包含

形成層與發育中的木質部纖維細胞均感受到力學變化（相對位置的改變），開始形

成抗張材，其中的膠質纖維可產生強烈的收縮應變，將主幹拉回到直立的位置。在傾斜主幹之基部形成的抗張材比在半株高處者多，產生的收縮應變也較大，顯示在台灣欒樹樹苗傾斜主幹恢復直立的過程中，主幹基部扮演較關鍵的角色。此外，傾斜主幹伴隨上側抗張材形成的偏心生長有助於主幹恢復直立，而在主幹下側部分測量點所量到的壓縮應變亦可靠著推力協助主幹恢復直立。針對樹冠之不同角度枝條的研究則顯示，台灣欒樹枝條內存在著和傾斜主幹不同的應變分布、抗張材分布與偏心生長。枝條之生長應變參數隨著木材細胞次生細胞壁的成熟而有季節性的變化；角度大的傾斜枝條上下側可能具有收縮或壓縮應變，然而角度較小的近直立枝條的上下側多為收縮應變，顯示這兩種枝條可能具有不同的功能。抗張材可能分布在枝條的各個位置，和抗張材主要分布在傾斜樹苗主幹的上側不同，因此，抗張材有助於枝條角度的動態調整。枝條的偏心生長位於枝條下方，可能阻礙枝條的上揚，甚至促進枝條的下壓。枝條上多樣的應變分布與抗張材分布，顯示各枝條可能為了因應環境因子如重力和光線的差異，而有不同的生物力學需求。

關鍵詞台灣欒樹、生長應力、生長應變、生物力學、抗張材、枝條、偏心生長、

插針法、傾斜主幹、彎曲傾向

(5)

英文摘要及關鍵詞

Abstract

Tension wood plays a role in maintaining the mechanical stability of angiosperm trees. Both biological and physical aspects of tension wood are essential in understanding the mechanism of trunk or branch reorientation. In this dissertation, we first worked on both the tension wood formation and its biomechanical function in artificially inclined 2- year-old Koelreuteria henryi seedlings. The tension wood formation and reorientation process of the trunk last for about 3 months. With pinning method, we confirmed that at the beginning of inclination the cambial zone including the vascular cambium and the developing normal wood fibers on the upper side of the inclined trunk perceives the onset of mechanical change and starts to produce G-fibers that generate a strong contractile released growth strain (RGS) for gravitropic correction. Stronger contractile RGS and more tension wood were found at the trunk base than at the half-height, suggesting that the trunk base plays a key role in trunk uprighting of K. henryi seedlings. The eccentric cambial growth in the tension wood side increases the efficiency of gravitropic correction and the compressive strains measured in the opposite wood of some inclined seedlings also help the upright movement. Then we further discriminated the biomechanical behavior of branches from leaning trunks. We thus investigated the development of growth strains, distribution of tension wood, and eccentricity on the branchwood of K.

henryi. The results revealed the unusual distribution of released growth strain and tension wood as well as growth eccentricity. The growth strain parameter showed seasonal changes, possibly due to the maturation of secondary cell wall. Both sides of the plagiotropic branches exhibited either contractive or extensive growth strains, whereas the orthotropic branches exhibited mostly contractive strains on both sides, which implied

(6)

different physiological function of the two branch types. The tension wood arcs may occur in any direction of the branchwood which is different from the inclined trunk with tension wood on the upper side, suggesting dynamic adjustment in branch reorientation.

In contrast to trunks, the hypotrophic eccentric growth in branches functioned in obstructing upward movement and even facilitates downward movement, probably because the dissociation between tension wood and eccentric growth. Diversified growth strain and tension wood distribution on the branches may reflect the individual biomechanical requirements for each branch depending on the environmental factors, possibly gravitropic and phototropic stimuli.

Key words bending dynamics, bending tendency, biomechanical model, branch, G-fibers, gravitropic correction, growth eccentricity, growth strain, Koelreuteria henryi Dummer, tension wood

(7)

Index of Figures

Figure 1. Experimental design for studying reaction wood formation in inclined

Koelreuteria henryi seedlings. ... 13 Figure 2. Sample collection of the Koelreuteria henryi branch. ... 16 Figure 3. The 3 wood discs of the branch B5 showing the measurement of eccentricity

and the hypotrophic growth on the lower side. ... 19 Figure 4. Dynamics of the uprighting process of the inclined seedlings. ... 25 Figure 5. RGS distribution at the half-height (a, c) and the trunk base (b, d) of the

control (a, b) and inclined (c, d) Koelreuteria henryi seedlings in the 2009- and 2012-experiment. ... 27 Figure 6. Radial wood growth increments at the half-height (a, c) and the trunk base (b, d) of the control (a, b) and the inclined (c, d) Koelreuteria henryi seedlings in the 2009-experiment. ... 31 Figure 7. Histochemical stainings for normal wood (a-c) and tension wood (d-f). ... 32 Figure 8. TEM (a-d) and SEM (e, f) photographs of opposite wood fibers (a, c, e) and

G-fibers (b, d, f). ... 33 Figure 9. Light micrographs of cross sections of the trunk of inclined Koelreuteria

henryi seedlings showing the tension wood formation. ... 34 Figure 10. Light micrographs of cross (a, c) and radial longitudinal (b, d) plastic

sections of the lower (a, b) and the upper side (c, d) of the trunk of the T3 seedling. ... 35 Figure 11. The relationship between tension wood ratio and RGSs. ... 36 Figure 12. Relationship of growth strain parameter () and inclination duration at the

trunk base (a) and at the half-height (b). ... 38 Figure 13. Scatterplot of the RGS on the upper (gu) and the lower side (gl) of the

(11)

plagiotropic branches (closed symbols) and the orthotropic branches (open symbols). ... 41 Figure 14. Seasonal RGS distribution on the upper side (gu) (a) and on the lower side

(gl) (b) of the plagiotropic and the orthotropic branches. ... 41 Figure 15. Pith eccentricity on the wood discs of the measuring sites of the plagiotropic

branches (a) and the orthotropic branches (b). ... 45 Figure 16. Anatomical features of the branchwood of Koelreuteria henryi. ... 46 Figure 17. Distribution of tension wood on the wood sections at the 3 measuring sites

of every plagiotropic branch sampled in Nov. 2008 ... 47 Figure 18. Distribution of tension wood on the wood sections at the 3 measuring sites

of the orthotropic branches B11-18 sampled in Nov. 2011 ... 48 Figure 19. Seasonal SBS parameter () (a) and RGS parameter () (b) of the

plagiotropic and orthotropic branches of Koelreuteria henryi. The sampling date was listed in the order of seasons ... 50 Figure 20. Relationships between  and  (a),  and gu (b), and  and gl (c) for the

plagiotropic branches collected in Nov. 2008 (closed symbols) and the

orthotropic branches in Nov. 2011 (open symbols) ... 53 Figure 21. The relationship between dCs/dR, dCg/dR, dC/dR and the distance from the

measuring sites to the trunk for the plagiotropic branches (a) and the

orthotropic branches (b). ... 55

Figure A 1. Diameter measurement. ... 86 Figure A 2. The height growth of the control (a) and the inclined (b) seedlings. ... 87 Figure A 3. Diameter growth (the vertical and the horizontal diameter) of the control

seedlings C7~C12 and the inclined seedlings T7~T12. ... 88

(12)

Figure A 4. The experimental design and the practical procedure of pinning method. . 90 Figure A 5. Pinning marks in the cross sections of the inclined K. henryi stems. ... 92 Figure A 6. Selected serial sections of a pinning mark in the normal wood from the

pinning center to the edge. ... 93 Figure A 7. Cross sections of a pin mark in the normal wood showing the wound tissue

in the center (a, b) and at the edge of a pinhole (c, d). ... 94 Figure A 8. Cross sections of a pinning mark in the tension wood. ... 95 Figure A 9. The measurement of radial growth increment on the wood section marked

by pinning. ... 96 Figure A 10. Three independent measurements of radial increment via serial sections of

one pinned sample. ... 97 Figure A 11. Wood sections of the control upright seedlings (C) at the stem base (b) and the half-height (h). ... 99 Figure A 12. Wood sections of the inclined seedlings (T) at the stem base (b) and the

half-height (h) showing tension wood distribution before and after the

experiment. ... 100 Figure A 13. Cross sections of cambial zone in the A and B side of the control upright

seedlings (C) ... 102 Figure A 14. Cross sections of cambial zone in the upper side (U) and the lower side (L) of the inclined seedlings (T) ... 103 Figure A 15. Plastic sections showing tension wood formation in K. henryi seedling. 104 Figure A 16. Seasonal cambial activity of inclined seedlings and upright seedlings. . 105 Figure A 17. The branches were tied with rope to manipulate the branch angle. B19 and

B21 were bended down and B20 and B22 were bended up. ... 107 Figure A 18. Wood sections double stained with safranin O and alcian blue showing the

(13)

pinning mark (white arrows) and tension wood distribution. ... 109 Figure A 19. Illustration showing the tension wood distribution on the tied branches. 110

(14)

Index of Tables

Table 1. Measured data of the control and the inclined seedlings of Koelreuteria henryi ... 28 Table 2. Experimental data of the inclined seedlings of Koelreuteria henryi for

prediction of bending dynamics ... 39 Table 3. Eccentricity, spring-back strains (SBS), and released growth strains (RGS) on

the branches of Koelreuteria henryi ... 42 Table 4. SBS parameters () and RGS parameters (), bending tendency, and the

rate of curvature change of Koelreuteria henryi branches ... 51 Table 5. Measured strain data of the natural and artificial defoliation branches of

Koelreuteria henryi ... 57 Table 6. Experimental data of Koelreuteria henryi used for calculating the curvature

change of branches ... 58

Table A 1. Experimental data of the branch angle manipulation study of Koelreuteria henryi ... 108

(15)

Introduction

Trees keep the mechanical stability by adjusting the position of their trunk and branches in response to various environmental disturbances and gravitational stimulus (Ewart and Mason-Jones 1906; Fisher and Stevenson 1981; Gardiner et al. 2014; Sinnott 1952; Wilson and Archer 1979). Leaning tree trunks and branches could reorient to an equilibrium position by producing reaction wood accompanying increased cambial growth. The characteristics of the reaction wood including anatomical and physical aspects are different in gymnosperms and angiosperms. In gymnosperms, compressive growth stress induced by compression wood is generated on the lower side to push the leaning trunk and branches upward. However, in angiosperms, tension wood generally develops on the upper side and produces tensile stress to pull the leaning trunk or branches upward (Onaka 1949; Scurfield 1973; Sinnott 1952; Wilson and Archer 1977).

Anatomical characteristics of reaction wood Macroscopic Appearance

Compression wood in gymnosperms is usually darker in color, from brown to dark reddish brown depending on species, which leads the name “Rotholz (red wood)” in the German literature. It is an extremely hard and brittle wood with a higher density than normal wood (Timell 1986a; Wardrop and Dadswell 1950). The occurrence of compression wood is associated with hypotrophic eccentric growth, i.e., the promoted growth increment on the lower side and the pith toward the upper side of the leaning trunk and branches. Therefore the growth ring is normally wider on the lower side. (Gardiner et al. 2014; Timell 1986b)

When sawed in the green condition tension wood is likely to leave a rough, woolly

(16)

surface (Clarke 1939; Gardiner et al. 2014). Because of the silvery appearance tension wood in temperate hardwood is called “Weissholz (white wood)” in the German literature.

However, dark streaks are sometimes visible in the tension wood of tropical species. The occurrence of tension wood is generally accompanied with epitrophic eccentric growth, i.e., the promoted growth increment on the upper side and the pith toward the lower side of the leaning trunk (Gardiner et al. 2014; Timell 1986b). Nevertheless, the association of tension wood and growth eccentricity is ambiguous in branches of angiosperms (Kučera and Philipson 1977a; Patel et al. 1984; Tsai et al. 2012; Wang et al. 2009) and awaits further investigation to understand the effect of growth eccentricity on the bending of branches.

By the shiny appearance under a low-angle natural light on the fleshly saw disks of poplar cultivars, tension wood areas can be quickly identified, and made it possible for a large scale research (Badia et al. 2005). Besides, brushing the surface of a wood disc with zinc chloro-iodide solution (Herzberg’s reagent) is an efficient and popular method for investigating tension wood areas. Because the chlorine destroys hydrogen bonds between macro-polymers of cellulose and thus facilitates the accumulation of iodine molecules which colors tension wood to be light purple to violet (Doğu and Grabner 2010;

Grzeskowiak et al. 1996; Wardrop and Dadswell 1948).

Cellular Structure

Comparing to the more rectangular to hexagonal normal wood tracheids without intercellular space, the tracheids of compression wood are shorter in length and round in cross section with intercellular spaces. The cell wall of a normal wood tracheid or fiber is composed of a primary wall (P) and a secondary wall consisting S1, S2, and / or S3 layers.

The secondary wall of compression wood tracheids possesses a relative thick S2 layer

(17)

with larger microfibril angle (MFA) and a higher lignin content but lacks S3 layer. The thick and heavily lignified cell wall of compression wood tracheids often shows cracks.

(Timell 1986a; Wardrop and Dadswell 1950).

The structure of tension wood is not as consistent as that of compression wood.

Typical tension wood is featured by the gelatinous fibers (G-fibers) with the S2 and/or S3 layers replaced by a special gelatinous layer (G-layer) which consists of highly crystalline cellulose fibrils that are almost parallel to the fiber axis, i.e., with smaller MFA (Scurfield 1973; Wardrop and Dadswell 1948). The G-layer has been presumed to be composed of nearly pure crystalline cellulose (Norberg and Meier 1966), however, some authors (Araki et al. 1983; Pilate et al. 2004; Scurfield and Wardrop 1963) have reported a slight deposition of lignin in the G-layer, especially in transitional regions between normal and tension wood. Besides, in addition to cellulose, the G-layer may also contain other polysaccharides including pectin and hemicellulose (Bowling and Vaughn 2008).

In angiosperms, not all of them produce typical tension wood with G-fibers. Onaka (1949) found that 79 % of 346 angiosperms produce G-fibers. Hӧster and Liese (1966) examined 266 trees/shrubs and found only ca. 50% of the investigated species exhibit G- fibers in branchwood and ca. 25 % in root wood. Fisher and Stevenson (1981) detected G-fibers on the upper side of the branches in 46% of 122 dicotyledonous species in 46 families. Besides, Buxus microphylla var. insularis Nakai even produces compression wood on the lower side of the leaning trunks, exemplifying the considerable diversified reaction anatomy in angiosperms (Yoshizawa et al. 1993a; Yoshizawa et al. 1993b).

Among species producing G-fibers, there are variations existing in their cell wall structure. G-fibers may contain cell walls consisting P+S1+G, P+S1+S2+G, or P+S1+S2+S3+G (Scurfield 1973; Wardrop and Dadswell 1955). Besides, a special polylaminate secondary wall structure with an alternate of lightly lignified thick layers

(18)

and more lignified thin layers was found in tension wood of Casearia javitensis (Clair et al. 2006c) and Laetia procera (Poepp.) Eichl (Ruelle et al. 2007b) in Flacourtiaceae family. In the species producing tension wood without G-fibers, the decreasing of vessel frequency seems to be a general feature among quantitative anatomical criteria (Ruelle et al. 2006).

The role of plant hormones in reaction wood formation

Plant hormones including auxin, ethylene, and gibberellin were involved in the formation of reaction wood. The regulatory role of hormone in reaction wood formation have been studied by both locally applying the hormone and/or its inhibitors and by quantifying endogenous hormone during reaction wood formation (for review, see Du and Yamamoto 2007; Gardiner et al. 2014; Hellgren et al. 2004)

The results of application experiments suggested that a high level of IAA is associated with the induction of compression wood in gymnosperms and a low level of auxin for tension wood in angiosperms (for review, see Du and Yamamoto 2007; Gardiner et al. 2014). On the other hand, the studies concerning endogenous IAA quantification often yielded contradictory results (Funada et al. 1990; Wilson et al. 1989). These early experiments can hardly be explained or compared due to coarse and incongruous sampling criterion. Until the high spatial-resolution quantification of endogenous IAA was carried out across the cambial region, Hellgren et al. (2004) disproved the regulatory role of IAA level in reaction wood formation. They suggested that other factors such as auxin perception mechanisms or additional signals are involved in reaction wood formation.

While there is no report of G-layer induction by ethylene treatments (Gardiner et al 2014), the hormone has been associated with compression wood differentiation. Savidge

(19)

et al. (1983) measured the endogenous ethylene precursor (ACC, 1-aminocyclopropane- 1-carboxylic acid) in the vascular cambium of Pinus branches. The results showed that the endogenous ACC was only detected on the compression wood side but not the opposite wood side. Applying ethrel, an ethylene–generating compound, to stems stimulated wood production in both woody gymnosperms (Yamamoto and Kozlowski, 1987) and angiosperms (Yamamoto et al. 1987); however, the anatomical features of the induced wood tissue are different from compression wood and tension wood (Gardiner et al. 2014). The regulatory role of ethylene in reaction wood formation need to be confirmed with further investigation.

In gymnosperms, the possibility of a role of gibberellins in compression wood formation has not yet been demonstrated (Gardiner et al. 2014). However, GAs have been suggested to play an important role in tension wood formation. Applying gibberellin promotes the formation of well-defined tension wood in artificially-tilted angiosperm seedlings (Baba et al. 1995; Nugroho et al. 2012; Yoshida et al. 1999). Moreover, the exogenous gibberellin induces the tension wood formation on vertical stems without gravitational stimulus (Funada et al. 2008). Besides, the application of inhibitors of gibberellin biosynthesis suppresses the increases in the thickness of gelatinous layers and the elongation of gelatinous fibers (Nugroho et al. 2013). These results suggested that GAs, at least, mediate the development of gelatinous fibers.

The onset and formation of reaction wood and the pinning method

Scurfield (1972) and Jourez and Avella-Shaw (2003) have proposed that at the onset of tilting, the differentiating xylary fiber cells in different stages may react to gravitational stimulus and add a G-layer in the innermost cell wall. The quick transformation from developing cells to G-fibers allows the plant to promptly recover to an equilibrium

(20)

position. However, the authors did not provide direct evidence to confirm the proposal.

For studying the onset and formation of reaction wood in a microscopic view, the pinning method could be helpful. This method was first introduced by Wolter (1968) and improved by Yoshimura et al. (1981) and is a precise method for measuring xylem growth by marking the position of cambial zone at the time of pinning. Practically, a pin or a needle is inserted through the bark, the cambium zone to the xylem tissue for inducing the wound reaction of the cambial cells by the minute mechanical injury (Seo et al. 2007;

Wolter 1968; Yoshimura et al. 1981). Since it can be difficult to gain good sections from the center region of the tiny wound, some authors (Kuroda and Kiyono 1997; Veenin et al. 2006) used knife incision instead of needle or pin insertion, and thus called it as the wounding method.

The pinning technique has been employed to monitor seasonal dynamics of cambial activity or wood formation in Picea abies (L.) Karst. (Gričar et al. 2007; Mäkinen et al.

2003; Nocetti and Romagnoli 2008), Pinus sylvestris and Betula spp. (Schmitt et al. 2004), Hevea brasiliensis (Ohashi et al. 2001) and Eucalyptus camaldulensis Dehnh (Veenin et al. 2006); as well as intra-seasonal growth of Toona ciliata (Heinrich and Banks 2002). It was also used to investigate the secondary growth via successive cambia in mangrove species Avicennia marina (Schmitz et al. 2008). The pinning method has been also applied in the study on compression wood formation of Japanese larch (Yamaguchi et al. 1983) and slash pine (Nix and Brown 1987). In the research of tension wood formation, Mukogawa et al. (2003) marked the vascular cambium with a knife instead of a pin to observe the eccentric growth of leaning trees in a macroscopic view. To study the phototropic and negative gravitropic bending, Matsuzaki et al. (2007) used the pinning method to mark the radial direction and measured possible torsion of the main stem. To distinguish the tension wood produced after the inclination treatment, Jourez et al. (2001)

(21)

bent the pot and the shoot in the opposite direction of natural incline. In this study, we attempted to identify the origin of G-fiber by the pinning method.

Physical characteristics of reaction wood Growth stress and growth strain

The growth stress originates in the cambial zone when growing wood cells contract in the longitudinal direction and expand in the transverse direction during differentiation and maturation. Since the contraction is impeded by older cells, the new cells generate longitudinal tensile stress; and the obstruction of the lateral expansion by neighbor cells results in tangential compressive stress (Kubler 1987; Münch 1938). In a vertical trunk, there is a longitudinal stress gradient from the periphery inward along the radius, i.e., tensile stresses in the newly formed wood and compressive stresses near the pith (Archer 1986; Dinwoodie 1966; Jacobs 1945). In gymnosperms, a strong compressive stress occurs on the lower side of the leaning stem; whereas in angiosperms, a strong tensile stress on the upper side (Onaka 1949; Scurfield 1973). These growth stresses perform essential biomechanical functions related to the curvature change of the trunk and branches, which ultimately affecting the tree architecture and its survival strategy.

(Alméras and Fournier 2009; Huang et al. 2010; Moulia et al. 2006; Washusen et al. 2003b;

Wilson and Archer 1977). However, when trees are felled or wood is processed, these growth stresses are released and the resultant splits, cracks, and warp cause the loss of potential economic value (Clarke 1939).

The growth strain, a consequence of relief of growth stresses, is the dimensional change per unit original length. Within the proportional limit of elasticity, level of growth stress is a function of growth strain and the elastic modulus of the wood (MOE). Growth

(22)

stresses, the force per unit area, are usually impossible to measure directly, while growth strains can be readily measured. The growth strains can be released and measured on tree or log surface by three major methods (Archer 1986; Yang and Waugh 2001), i.e., the Nicholson method (Nicholson 1971), the French method, and the strain gauge method (Yoshida et al. 1999). By measuring the strain, the corresponding stress can be calculated (Yang and Waugh 2001).

Generation of reaction stresses

The dispute about the generation of growth stress is based on the lignin swelling (Boyd 1972, 1973) and the cellulose tension (Bamber 1987, 2001) hypotheses. Boyd (1972, 1973) proposed that the lignification and associated transverse swelling of the cell wall induce the strains and growth stresses in normal wood and compression wood.

However, the hypothesis cannot explain the generation of high tensile stress in tension wood, especially in G-fiber tension wood. As an alternative, Bamber (1987, 2001) proposed the cellulose tension hypothesis in which the stress of the both reaction woods arises from the spring-like cellulosic component of the wood. In compression wood, the S2 layer acts as compressed helical springs and thus exerting an extensive force to push the stem upright or to stabilize it. In tension wood, the microfibrils of the secondary wall and/or the G-layer are considered as stretched linear springs and thus developing a tensile stress to pull the stem upright or stabilize it. The role of lignin is to consolidate the compressive strength of compression wood rather than contribute to the generation of growth stress (Bamber, 2001). Besides, an idea that unifies the lignin swelling (Boyd, 1972) and the cellulose tension (Bamber 1987, 2001) hypotheses was proposed by Okuyama (1994).

In the past decade, many researches have been focused on the stress generation in

(23)

tension wood (Clair et al., 2006b; Goswami et al., 2008; Washusen et al, 2003a, b;

Yoshida et al, 2002). The presence of G-fibers on the upper side of the inclined stem is correlated with high tensile growth strains or stresses (Fang et al. 2007; Ruelle et al. 2011;

Washusen et al. 2003a). The mechanism of stress generation is related to the features of G-fibers including small MFA (Donaldson 2008; Okuyama et al. 1994), increased cellulose lattice spacing (Clair et al. 2011; Clair et al. 2006b) and high content of xyloglucan (Baba et al. 2009; Mellerowicz and Gorshkova 2012; Mellerowicz et al. 2008;

Nishikubo et al. 2011). Besides, the magnitude of growth stress is influenced by the thickness of G-layer and the ratio of G-fiber area (Fang et al. 2008).

The study of growth strain distribution in angiosperm trees

Growth strains on naturally leaning or artificially tilted trunks (Alméras et al. 2005;

Clair et al. 2006a, c; Dassot et al. 2012; Kuo-Huang et al. 2007; Ruelle et al. 2007b; Wang et al. 2009; Yamamoto et al. 2005; Yoshida et al. 2002) and branches (Kuo-Huang et al.

2007; Tsai et al. 2012; Wang et al. 2010; Wang et al. 2009; Yoshida et al. 1999; Yoshida et al. 2000b) have been investigated in several angiosperms. These studies revealed that on the upper side of both tilted trunks and branches exhibited contractive strains; however, on the lower side of the leaning trunks exhibited mostly contractive strains but the branches mostly extensive strains.

Biomechanical models

To explain the mechanism of upward bending process, Fournier et al. (1994a) presented a simple biomechanical model according the maturation strain asymmetry of a leaning trunk. Coutand et al. (2007) applied the model to investigate the gravitropic

(24)

response of poplar trunks and found it limited to explain the total variance. For providing the theoretical predictions on biomechanical design and long-term stability of trees, Alméras and Fournier (2009) expanded the original model with the gravitational disturbance, eccentric growth, and heterogeneous stiffness. Based on Fournier’s original model, Huang et al. (2010) developed an equivalent model by using spring-back strain (SBS) to evaluate self-weight bending moment and made the model practical and easy to follow. Huang’s model has been successfully applied in the prediction of bending dynamics of broadleaf tree branches (Tsai et al. 2012).

Goals of this dissertation

The investigation of strain distribution, tension wood formation, growth eccentricity, and/or the physiological function in angiosperm trunks have received more attention and reached a consensus (Alméras et al. 2009; Coutand et al. 2014; Mukogawa et al. 2003;

Washusen et al. 2003a, b; Yoshida et al. 2000a). In contrast, there are substantial controversies in studies of branches. Comprehensive studies concerning physical and anatomical properties in both trunk and branches of angiosperm trees are rare in the literature.

Koelreuteria henryi Dummer (synonym: K. elegans ssp. formosana) in Sapindaceae family is a deciduous tropical tree species with meandering branches. In Taipei, the tree starts to change leaf color and to lose leaves from late December and then flushes in next February in Taipei. Flowering and fruiting of K. henryi occurs for a relatively brief period from September to December. The terminal panicle florescence (up to 25 cm long) contains many flowers. The fruit is an inflated papery capsule (4 cm long) containing 6 globose seeds (5 mm across). It is endemic to Taiwan but is listed as an invasive species

(25)

in south-east Queensland, Australia (Batianoff and Butler 2002) and in Florida, USA by Florida Exotic Pest Plant Council (2015). The findings of unusual eccentric growth and released growth strain (RGS) distribution in its branches have led us to conduct studies for having a comprehensive insight into the role of reaction wood formation in the leaning trunk and branches during the tree form adjustment.

In the dissertation, we first monitored the reorientation process of the artificially inclined trunks of K henryi seedlings. The growth strain distribution and related eccentric growth were surveyed. Besides, we used the pinning method to directly identify the onset of G-fiber formation. The cell wall structure and distribution of G-fibers were also investigated. Huang’s model (Huang et al. 2010) was applied for analyzing the bending dynamics of the inclined trunks. Then we measured the seasonal strain distribution of branches. To verify the physiological function of branches in building the tree architecture, we further investigated the bending tendency and its relationships with growth eccentricity and tension wood in orthotropic and plagiotropic branches. We proposed that strains on the branch may change seasonally. Furthermore, the branches of different angles or locations may have different bending tendency and the consequent physiological functions that are distinguished from the trunk.

(26)

Materials and Methods

Plant material and experimental design

Study on the reorientation process of the artificially inclined seedlings

To study the trunk reorientation process of Koelreuteria henryi seedlings, two greenhouse experiments were conducted. On Feb 18 2009, twenty-four 2-year-old seedlings germinated from seeds were purchased from a nursery in Changhua (central Taiwan). The naturally defoliated seedlings were transplanted in a 5-liter pot with Yang- Ming-Shan soil, perlite, vermiculite, and peat moss (1:1:1:1) and acclimated in the greenhouse of National Taiwan University (2500N, 12127E). On Apr. 23, the average height of seedlings was 82.9 ± 18.8 cm and the average basal diameter was 8.44 ± 0.94 mm. Twelve seedlings (T1~T12) were randomly selected and artificially inclined at an angle of about 30° (28.8  2.7 (SD)) from vertical by placing the pots on fixed PVC pipes (Fig. 1a) and the other 12 seedlings (C1~C12) were kept upright as controls. The height and diameter growth of all seedlings were monitored during the experiment (Appendix Ⅰ). The trunk base (at 10 cm above the soil) and the half-height of all seedlings were pre-marked with white-out for subsequent RGS measurement. The bark of each seedling at 1 cm below the measuring site was pinned by an insect needle (0.4 mm in diameter) on the upper and the lower sides of the inclined seedlings and the corresponding A and B sides of the control seedlings. By pinning, the cambial zone was wounded and triangular callus tissue was formed and thus marked the location of vascular cambium at the pinning time (Yoshimura et al. 1981) (Appendix Ⅱ). We monitored the uprighting process of the inclined seedlings. Seasonally, we measured RGSs of the seedlings and investigated the related anatomical structures. We also applied Huang’s model (Huang et al. 2010) to analyze the bending dynamics.

(27)

On Apr. 25, 2012 we started a similar experiment with monthly sampling for three months. The nine studied seedlings (C13~15 for control and T13~T18 for inclination) were obtained through pruning the stems of the seedlings above the soil in 2011 and left one epicormic shoot to become the main stem. The average height was 103.2 ± 12.6 cm and the average basal diameter was 9.54 ± 0.92 mm. One control and two inclined seedlings were sampled monthly for measuring RGSs, examining the tension wood formation, and then predicting the bending tendency.

Figure 1. Experimental design for studying reaction wood formation in inclined Koelreuteria henryi seedlings.

a Photograph of the inclined seedlings at the beginning of inclination. The experiment was conducted in the green house and each pot was placed on a pair of fixed PVC pipes to set the tilting angle. b Picture showing the RGS measuring sites (the base and the half- height of the trunk) and the positions (the upper and the lower sides of the inclined trunk).

Pictures of each inclined seedling were taken weekly to monthly and from which the proximal angles () (the angle of the line joining the trunk base and the half-height to the vertical) were measured. c Illustration of the locations of pinning marks and RGS measurement. d Illustration of tension wood ratio measurement: dotted line outlining the measured whole wood area and grey area marking the area of tension wood

(28)

To study the dynamics of uprighting process, we monitored the angle change.

Pictures of each inclined seedling of the 2009-experiment were taken with a digital camera (Nikon D3) weekly for the 1^st season, biweekly for the 2^nd season, and monthly for the 3^rd and 4^th seasons (Fig. 1b). Photographs of each seedling were assembled with Photoshop CS5 to examine the trunk shape evolution during the uprighting process.

Because the radius of curvature is large in the inclined trunk of K. henryi seedlings, the proximal angle () (Fig. 1b) of each seedling was measured from the pictures using Image J and then used to analyze the uprighting process.

Study on the distribution of growth strain and reaction wood of the branches

The studied 13 plagiotropic and 8 orthotropic branches were randomly selected from Koelreuteria henryi trees (with a DBH from 12 to 24 cm) on the main campus of National Taiwan University (2500N, 12127E). For studying the seasonal biomechanical behavior of branches, 3 to 4 plagiotropic branches, with an angle of 60 to 90 from vertical near the lower tree crown, were sampled in mid-August (summer; branches B1- 4), mid-November 2008 (autumn; B5-7), mid-February (winter; ND1-3), and mid-May 2009 (spring; B8-10). In this study, the data in winter were discussed separately because the strains of branches are complicated by defoliation effects. The mean length of the 10 plagiotropic branches was 329 (SD 48) cm and the mean diameter at 10 cm from the trunk was 5.1 (SD 0.8) cm.

For comparing biomechanical behavior of branches with different angles and sizes, another 8 orthotropic branches (B11-18) with an angle of 5 to 45 from vertical near the upper tree crown, were sampled in November 2011 (autumn), as all secondary xylem had already matured and growth strains were supposed to be stable. The mean length of these

(29)

(SD 0.3) cm. The plagiotropic branches sampled in November 2008 had some dry capsules on the tip; but only few on some of the orthotropic branches of 2011. Comparing with the heavy compound leaves, we consider the load of these papery capsules small.

The spring-back strains (SBS) and longitudinal released growth strains (RGS) were measured during sampling, and the wood discs were collected for further examination of growth eccentricity and tension wood distribution. The bending tendency was evaluated with Huang’s biomechanical model (Huang et al. 2010).

Released growth strain (RGS) measurement

RGS at the stem base and the half-height of the control and artificially inclined seedlings was measured seasonally (7/23, 10/20 in 2009 and 1/28, 4/21 in 2010) in the 2009-experiments and monthly (5/25, 6/25, 7/25) in the 2012-experiments. RGS on the plagiotropic branches was measured seasonally (August and November, 2008; February and May, 2009) and the orthotropic branches in November 2011. For each branch, strains were measured every 30 cm along the branch with the first site at 10 cm from the trunk (Fig. 2a). Knots were carefully avoided.

To measure RGS, the bark of the marked measuring site was removed. The strain gauge (FLA-5-11-5LT, Tokyo Sokki Kenkyujo Co., Ltd) was glued to the surface of secondary xylem using cyanoacrylate adhesive (Tokyo Sokki Kenkyujo Co., Ltd) and connected to a data logger (TD 102, Tokyo Sokki Kenkyujo Co., Ltd) (Figs. 1c and 2b).

After the gauge being zeroed, the seedlings or branches were firstly cut off from the base and held straight to read spring-back strain (SBS; εsu / εsl). Then, RGS (εgu / εgl) was measured after the cross-cutting at the positions 3 mm apart from the upper and the lower rims of the gauge (Fig. 2c) (Huang et al. 2010).

(30)

In order to realize the effect of defoliation, the three plagiotropic branches (ND1–

ND3) sampled in February 2009 were naturally defoliated; the spring-back strain (ε1su / ε1sl) and growth stain (εgudef / εgldef) of three or four sites were measured. The branch B7 sampled in December 2008 was artificially defoliated. Before the branch was cut down, the defoliation spring-back strains (εsudef / εsldef) were measured after all leaves on the branch were removed.

Figure 2. Sample collection of the Koelreuteria henryi branch.

a The plagiotropic branch B8 sampled in May 2009 with 3 measuring sites. b The branch was held vertical to record spring-back strain (SBS). c The released growth strain (RGS) was measured after the wood disc was cut

Wood anatomical structure and morphometry of the artificially inclined seedlings Stem segments of 3 cm long with the marked pinholes were fixed and stored in FPGA (Formalin: Propionic acid: Glycerol: 95% Alcohol: distilled water = 1: 1: 3: 7: 8).

By using sliding microtome (ERMA optical works, Ltd), 20 m thick wood sections containing pinning-induced callus were collected and stained with 1% toluidine blue O in 1% sodium borax (TBO) or double stained with 0.5% safranin O and 0.1% alcian blue

(31)

stained with zinc chloride-iodine. G-layers were stained purple-red with zinc chloride- iodine, purple with TBO, and light blue with alcian blue.

Whole stem sections were collected and photographed by using a digital camera (Sony DSC-T200). The wood diameters of the vertical and horizontal axis were measured with Image J and then the average radius (R) was calculated. The area of tension wood formed after inclination by each inclined seedling was also measured on the cross section using image J. The ratio of tension wood on the upper side (hereafter tension wood ratio) was calculated as tension wood area over whole wood area formed after inclination.

Concerned the effective regions of the strain gauge measurement, we confined the whole wood area as 2 mm in width and between the pinning-induced callus and the recently matured wood (Fig. 1d). To measure the wood growth increment in the trunks, the upper and the lower sides of the trunk sections containing cambial marks and new formed wood tissue were photographed by Nikon D3 on Leica Diaplan Light Microscope. The variation of radius, i.e. the radial growth increments during the period of experiment including Ra

and Rb for the A side and the B side of the control seedling; Ru and Rl for the upper side and the lower side of the inclined seedling were measured from the margin of the cambial mark to the cambial zone with Image J. The average growth increment R was then calculated, R is equal to (Ra + Rb)/2 for the control seedling and (Ru + Rl)/2 for the inclined seedling.

For studying the cambial activity and the cell wall ultrastructure, stem blocks were cut from 20 cm above the soil. The wood samples (1 x 1 x 2 mm³) of the upper and the lower sides were fixed with 2.5% glutaraldehyde (in 0.1M phosphate buffer) and then followed by 1% osmium tetroxide (in 0.1M phosphate buffer), dehydrated with acetone series. For TEM, the samples were infiltrated and embedded in Spurr’s resin (Spurr 1969) and cut by using an ultramicrotome (Ultracut E). Semi-thin sections (1μm) were stained

(32)

with TBO and photographed by Nikon D3 under Leica Diaplan Microscope. Ultra-thin sections were stained with a fresh mixture of 5% UA and 1% KMnO4 (1:3) and observed with Hitachi H-7650. For SEM, the fixed samples were further dehydrated by critical point drying method (Hitachi HCP-2) and coated with gold (Hitachi E101) and observed by FEI Inspect S. The MFA of fibers was measured from the SEM photographs using Image J.

Eccentricity calculation, radius and radial wood growth increment measurement of the branchwood

Eccentricity of the branchwood discs was calculated according to Japan Material Society (1982). On the surface of each wood disc, the vertical diameter (DV), horizontal diameter (DH), and the vertical distance from the pith to the lower edge (DA) were measured with a caliper (Fig. 3). Eccentricity (Ec) is defined as eccentric distance (DE, the vertical distance between the pith and the geometric center) over horizontal diameter:

𝐸𝑐 = 𝐷_𝐸⁄𝐷_𝐻 = (𝐷_𝐴−𝐷_𝑉 2 ) 𝐷⁄ _𝐻

A positive sign indicates that the pith is above the geometric center, i.e., hypotrophic growth (promoted radial growth on the lower side); a negative sign means the reverse, i.e., epitrophic growth (promoted radial growth on the upper side).

The average radius of the wood disc (R) was calculated as (DV + DH)/4. And the radial growth increments during the last growing season, Ru for the upper side and Rl

for the lower side, were also measured with a caliper (Fig. 3). The average growth increment (R) equals (Ru + Rl)/2.

(33)

Figure 3. The 3 wood discs of the branch B5 showing the measurement of eccentricity and the hypotrophic growth on the lower side.

P pith, GC geometric center, DH horizontal diameter, DV vertical diameter, DE the vertical distance between the pith and the geometric center, DA the vertical distance from the pith to the lower edge, Ru the radial growth increment on the upper side, Rl the radial growth increment on the lower side

Tension wood distribution and branchwood structure

Wood discs obtained from the 3 plagiotropic and the 8 orthotropic branches sampled in autumn were examined for tension wood distribution. K. henryi contains living fibers and the massive starch grains in living fibers, making it difficult to use iodine-zinc chloride solution to detect G-fibers. Therefore, cross sections of wood discs of 20 m were cut using sliding microtome (Reichert-Jung Hn 40) and doubly stained with 0.5%

safranin O and 0.1% alcian blue (in 0.3% acetic acid). For wood discs that were too big for sliding microtome, they were sawn into several pieces first and then sectioned, doubly stained, and finally assembled. The whole sections of wood discs were photographed using Nikon D80. The outline of wood discs, annual rings and tension wood sectors were

(34)

traced using CorelDRAW X5 (Corel Corporation, 2010). For studying the anatomical features, the wood sections and macerated wood samples prepared by Schultze’s methods (Johansen 1940) were observed and recorded by light microscope (Olympus CX41) and photographed by Cannon EOS 700D.

Prediction of the bending dynamics of trunks and branches

Huang’s biomechanical model (Huang et al. 2010) was applied to predict the bending dynamics of the trunk and branches. The rate of change in curvature associated with growth increment is expressed as

d𝐶

d𝑅

=

^d𝐶^s_d𝑅^+d𝐶^g

=

^{4(𝛽−𝛼)}_𝑅₂

(1)

where dCs/dR (= 4R²) is the rate of change in curvature due to spring-back (self-weight) associated with growth increment, and dCg/dR (= –4R²) is due to asymmetric growth strain; SBS parameter () associated with gravitational force is defined as half the difference of the SBS between the upper and the lower side: = (su – sl)/2; RGS parameter () associated with gravitropic correction is defined as half the difference of the RGS between the upper and the lower side: (gu – gl)/2. When dC/dR is positive, the branch tends to bend upward; when negative, the branch tends to bend downward.

That is, the bending tendency can be predicted by simply examine the sign of –

positive for upward bending and negative for downward bending. The bending tendency can also be predicted by the value of gravitropic performance (PG =

greater than 1 for upward bending, smaller than 1 for downward bending (Fournier et al. 1994a).

(35)

Equation (1) can be further rectified with eccentric radial growth by substituting  with '

𝛼

^′

= (𝜀

_gu^𝑑𝑅_𝑑𝑅^u

− 𝜀

_gl^𝑑𝑅_𝑑𝑅^𝑙

)/2

⁽²⁾

where dRu is the wood growth increment on the upper side, dRl the lower side, and dR the average increment. The dRu/dR and dRl/dR were estimated with the measured Ru/R and

Rl/R, respectively. The corrected rate of change in curvature is expressed as dC'/dR.

d𝐶^′

d𝑅

=

^d𝐶^s^+d𝐶_d𝑅 ^′^g

=

^{4(𝛽−𝛼}_𝑅₂ ^′⁾

(3)

Modified model for defoliating action of deciduous trees

During the defoliating season, β1 is the half difference of spring-back strain between the upper and the lower side measured on the leafless branch, and α1 is that of growth strain. With the spring-back strain during the defoliating process

expressed as εsudef and εsldef, respectively, we obtain

α

1

=

^(𝜀^gudef₂^−𝜀^gldef⁾

=

^(𝜀^gu^−𝜀^sudef^)−(𝜀₂ ^gl^−𝜀^sldef⁾

=𝛼 − 𝛽

_𝑑𝑒𝑓

(4)

β

1

=

^(ε^su^−ε^sudef^)−(ε₂ ^s^l−ε^sldef⁾

=𝛽 − 𝛽

_𝑑𝑒𝑓 (5)

𝛽

1−

𝛼

1

= 𝛽

−

𝛼

(6)

where βdef = (εsudef − εsldef)/2 is the β during defoliating action, and εgudef and εgldef are,

respectively, the measured growth strains on the upper side and the lower side of the branch after defoliation. It is clear that growth strain is the sum of measured growth

(36)

strain after defoliation and spring-back strain during defoliation. This means that growth strain is superimposed by the spring-back strain during defoliation. So that dC/dR can be expressed as

𝑑𝐶

𝑑𝑅

=

^4(β−α)_𝑅₂

=

^4(β¹_𝑅^−α₂ ¹⁾ (7)

The rate of change in curvature of a defoliated branch associated with growth increment dC1/dR and gravitropic performance P1G can be expressed as

𝑑𝐶₁

𝑑𝑅

=

^4(β¹^−α_𝑅¹₂^−β^𝑑𝑒𝑓⁾ (8)

𝑃

_1𝐺

=

^α

β₁

=

^α¹^+β^𝑑𝑒𝑓

β₁ (9)

In the case of eccentric growth increment on the cross-section, a similar relation can be obtained as below:

𝑑𝐶′1

𝑑𝑅

=

^4(β¹^−α′)

𝑅² (10) 𝑃_1𝐺 =_β^α′

1 (11)

Statistics

For the inclined seedlings, all data were tested for normality by Shapiro-Wilk test.

When the data conformed to normal distribution, the means of RGSs of the upper and the lower sides were compared with paired t-test, for which t-value and p-value were provided. The means among different inclination durations (3, 6, 9 and 12 months) were compared with one-way ANOVA, for which F-value and p-value were provided. For heteroscedastic data or data deviated from normal distribution, Kruskal-Wallis rank sum

(37)

test was used instead, for which chi-square and p-value were provided. Either ordinary linear regression or polynomial regression was used to analyze the relationship between strain magnitude and inclination duration, and between RGSs and tension wood ratio, for which, r² and p-value were provided. Statistical tests were performed using Origin (OriginalLab, Northampton, MA) and R version 3.1.1 (the R Core Team, 2013). All statistical relationships were considered to be significant at p<0.05.

For the branches, all data were checked for normality by Shapiro-Wilk test. To compare between the plagiotropic and orthotropic branches, the means of parameters (gu,

gl, and  were compared with t-test or pooled t-test when the variances of the two samples are unequal. The means among seasons were compared with one-way ANOVA followed by Tukey’s honestly significant multiple comparison if the difference is significant. The residual plots were examined for homoscedasticity; in the case of heterogeneous variance, we performed Kruskal-Wallis rank sum test followed by Dunn’s test. The relationships between  and ,  and gu, as well as  and gl were described by ordinary linear regression and Pearson’s correlation (r²), for which p-value of the slope was provided. Statistical tests were performed using R software (R Core Team 2013). All statistical relationships were considered significant at p < 0.05.

(38)

Results

Study on the reorientation process of the artificially inclined seedlings of Koelreuteria henryi

Dynamics of the uprighting process of the inclined seedlings

In the 2009-experiment, we found that the uprighting process occurred mainly in the first 3 months after inclination in spring season. One week after tilting, five out of the 12 inclined seedlings sagged a little, while others showed only little change in orientation (Fig. 4a). Two weeks after tilting, all the 12 inclined seedlings curved upward. The proximal angle change after inclination was 2.7 (SD = 1.6, n=12) in the 1^st month, 2.7

(0.8) in the 2^nd month, and 2.1 (SD 1.0) in the 3^rd month; i.e., the total proximal angle change was 7.5 (SD = 2.52, n=12) in the 1^st season. Thereafter, the uprighting process was much slower: the proximal angle changed 1.5 (SD = 1.2, n=9) in the 2^nd season, 2.2 (1.1, n=6) in the 3^rd season, and 1.5 (SD = 0.9, n=3) in the 4^th season (Fig. 4a). At the end of this experiment, the average tilting angle for the last 3 seedlings was 19.9 (SD 3.7). The apical region of the seedlings curved up soon after the inclination, they were straight and vertical after 6- to 8-weeks and mostly stayed vertical and only some of them showed overcorrection (Fig. 4).

(39)

Figure 4. Dynamics of the uprighting process of the inclined seedlings.

a Changes of the proximal angle (positive for uprighting and negative for sagging) of the inclined seedlings in the 2009-experiments. b Tree shapes of the T3 seedling from April 23 to July 23, 2009. The proximal angle changed from 25.1 to 15.8^o in these 3 months. c Seasonal tree shapes of the T9 seedling (April, July, October 2009 and January 2010).

The tree shapes on the last two seasons were similar, however fewer leaves were observed on January 2010. d Seasonal photos of the T10 seedling (April, July, October 2009 and January, April 2010). Arrow heads point out the half-height of the trunks.

(40)

Released growth strain distribution

Since the stress is proportional to the strain within a proportional limit (Archer 1986), we estimated the pre-stress by measuring the RGSs of green wood. Figure 5a and b shows that the control seedlings exhibited either contractive (−) or extensive () strains on both sides of the trunk base and the half-height (Fig 5, Table 1). In the 2009-experiment, the RGSs measured on the A and B sides of the control seedlings showed no significant difference (t = −0.343, p = 0.735), indicating that in the green house there was no perspective effect. The average RGSs were small: −120.21  at the trunk base and

−76.67  at the half-height, while strain values were more stable at the trunk base (SD

= 145.19) than at the half-height (SD = 232.48) (F = 0.379, p = 0.012), because two stronger contractive values (−610 and −652 ) were measured at the half-height on the last sampling date. The RGSs of the trunk base of the control seedlings in the 2012- experiment fell into the range of those in the 2009-experiment (Fig. 5b).

In the 2009-experiment, the inclined seedlings exhibited only contractive strains on the upper sides (gu: −1690 to −64 ), while on the lower side either contractive or extensive strains were measured at both the trunk base and the half-height (gl: −1276 to

495) (Fig. 5c, d, Table 1). The gu and gl showed no significant change during the whole experiment period (gu:² = 3.43, p = 0.329 for the trunk base, ²= 2.897, p = 0.408 for the half-height; gl: ² = 1.256, p = 0.740 for the trunk base, ²= 1.051, p = 0.789 for the half-height), whereas the value of gu was significantly smaller than gl

(paired t = −2.653, p = 0.011 for the trunk base, t = −2.956, p = 0.007 for the half-height).

Similar results were observed in the 2012-experiment; however, the RGSs on the lower side are more extensive than those in the 2009-experiment (Fig. 5c, d). When the data of the two experiments were pooled, a negative relationship was found at the trunk base

(41)

between gl and inclination duration (r² = 0.29, p = 0.022); the correlation coefficient is improved when a potential outlier was removed (r²= 0.59, p < 0.01)

Figure 5. RGS distribution at the half-height (a, c) and the trunk base (b, d) of the control (a, b) and inclined (c, d) Koelreuteria henryi seedlings in the 2009- and 2012-experiment.

g RGS of both A side (ga)and B side (gb) for the control seedlings, gu andgl, RGS for the upper and the lower side of inclined seedlings

臺灣欒樹的抗張材在傾斜苗木與枝條的解剖構造與生物力學功能

國立臺灣大學生命科學院生態學與演化生物學研究所 博士論文

Institute of Ecology and Evolutionary Biology College of Life Science

National Taiwan University Doctoral dissertation

臺灣欒樹的抗張材在傾斜苗木與枝條的 解剖構造與生物力學功能

Anatomical Structure and Biomechanical Function of Tension Wood in Inclined Seedlings and Branches

of Koelreuteria henryi Dummer

洪麗分 Li-Fen Hung

指導教授：黃玲瓏 博士

Advisor: Ling-Long Kuo-Huang Ph.D.

中華民國 106 年 6 月

June, 2017

致謝

中文摘要及關鍵詞

英文摘要及關鍵詞

目錄

Index of Figures

Index of Tables

Introduction

Materials and Methods

=

=

𝛼

= (𝜀

− 𝜀

)/2

=

=

α

=

=

=𝛼 − 𝛽

β

=

=𝛽 − 𝛽

𝛽

𝛼

= 𝛽

𝛼

=

=

=

𝑃

=

=

=

Results

國立臺灣大學生命科學院生態學與演化生物學研究所博士論文

臺灣欒樹的抗張材在傾斜苗木與枝條的解剖構造與生物力學功能

指導教授：黃玲瓏博士