A comparison of root architecture and tree stability in bare root- and Jiffy-trees (Picea abies), eight meter tall

Transcription

1 A comparison of root architecture and tree stability in bare root- and Jiffy-trees (Picea abies), eight meter tall Anders Tærø Nielsen Anders Tærø BA. Oecon Nielsen and Stud. Silv. Faculty of LIFE Sciences, Copenhagen University Study number: SBK08008 Supervisor: Carl Christian Nørgaard Nielsen December 17 th 2010

2 Abstract Root architectural and storm stability measurement were conducted on 16 year old Norway spruce trees at the height of eight meters, with one bare root plant sample in comparison with one containerized (Jiffy) sample. The investigation was conducted at the flat and sandy out-washed plain of western Jutland, which is the part of Denmark where wind throw is most likely to occur. Problems with bare root plants are mainly in the planting process, which leads to root deformation that in the long run can destabilize the trees. Containerized plants were developed in the seventies and through three generations of developing this plant type, an air root pruning system is now developed. This system developed better symmetry in the root system in youth and is hypothesized that this will in the long run lead to better stability of the trees. In this report root architecture and tree stability is investigated, by the use of the following methods. Stability data was produced by the method of winching down the tree and analysis of covariance was used for testing differences between the two samples. The root systems from the winched trees were dug up and measurements were taken. By the use of the database Treearch, root architectural variables were produced to do the testing for differences between the two samples. The investigation shows that the containerized plant type Jiffy made a better architecture of the structural root system and it was found that the Jiffy plants also had a higher resistance to wind throw. The higher resistance to wind could partly be explained by the better root architecture. Keywords: Tree stability, root architecture, modeling root systems, Jiffy, bare root, Norway spruce, wind throw, containerized plants. Resume Rodarkitektur og træstabilitetsmålinger blev udført på 16 år gamle Rødgran træer, som var 8 meter høje, med en stikprøve, der indeholdt barrods planter, samt en stikprøve, der indeholdt dækrodsplanter planter (Jiffy). Undersøgelsen blev udført i Vestjylland på et af de sandede, udvaskede områder, som er den del af Danmark, hvor der er størst risiko for stormfald. Problemet med barrodsplanter ligger primært i plante-processen, der kan fører til rod-deformation, som i det lange løb kan destabilisere træerne. Dækrodsplanter blev udviklet i halvfjerdserne og gennem tre generationers udvikling af denne plantetype, blev et luft rodbeskærings-system udviklet. Dette system udvikler bedre symmetri i rodsystemet når træerne er unge. Hypotesen er, at det i det lange løb vil føre til en bedre stabilitet af træerne. I denne undersøgelse er udviklingen i rod arkitektur og træ-stabilitet undersøgt og sammenlignet, ved brug af følgende metoder. Stabilitetsdata blev lavet ved hjælp af nedtrækning af træerne med spil, og kovarians analyse blev brugt til at teste, om der var forskel mellem de to stikprøver. Rodsystemerne fra de væltede træer blev gravet op og målt. Ved hjælp af databasen Treearch, blev der konstrueret variable, der beskrev rodarkitekturen. Rodarkitekturen kunne så testes for forskelligheder mellem de to stikprøver. Undersøgelsen viser, at dækrodsplanterne (Jiffy) havde en bedre arkitektur i det strukturelle rodsystem. Samtidig blev det blev vist, at dækrodsplanterne også havde en bedre stormstabilitet. Den højere stormstabilitet kunne til en vis grad forklares af den bedre rodarkitektur. Nøgleord: Træ-stabilitet, rodarkitektur, modellering af rodsystem, Jiffy, dækrod, barrod, stormfald, Picea abies 2

3 Content PREFACE... 6 Definition of variables, (Appendix 1) INTRODUCTION LITERATURE REVIEW AND REASONING OF THE HYPOTHESIS Planting and development of the structural root system Symmetry and method of planting Storm stability and components of the root system Theoretical basis for modeling stability Wind induced damage on stand level Wind induced damage on single tree level Sum up MATERIALS AND METHODS Introduction The experimental field The sample Root architecture Data collection Transformation of data Stability: Maximum applied force, first root breakage and first large series of root breakages Experimental design Data gathering Redesigning the experiment for second sampling round Correction for crown weight in 9 trees Statistical method Other field measurements (lever arm) Modeling the relation between root architecture and stability Theoretical approach for modeling root system and the relation to stability Empirical approach for modeling root system and influence on stability Distribution analysis and assessment of very poor performing trees External factors: Competition index and soil quality Competition index Soil quality

4 4. RESULTS Root architecture Analysis of variance Stability: Applied force measurements Analysis of variance Regression analysis Analysis of covariance Relating root architecture to tree stability Theoretical approach Empirical approach Scoring the root systems Analysis of distributions in assessment weak points of (very poor performing trees) External factors: Competition index and soil differences Competition index Soil analysis DISCUSSION The sample The experimental design Age of the material Differences in the distribution of diameter of breast height Discussion on root architecture results Cross sectional area number of roots and distribution of the CSA between roots Radial distribution of roots and symmetry parameters Generalization Discussion on the stability results Discussion on competition index as a stability influencing factor Influence of soil quality on stability Discussion on root architecture related to stability Discussion on theoretical vs. empirical modeling of root systems influence on stability Discussion on sporadic wind throw and weak points in the forest Conclusions versus indications and pitfalls in concluding

5 6. CONCLUSION Conclusions on the present dataset Root architecture: Answer to question Stability: Answer to question Relation between root architecture and stability: Answer to question Influence of external factors: Answer to question Approving or rejecting the hypothesis Generalized conclusions Indications from the results PERSPECTIVES Acknowledgements LITERATURE LIST APPENDICES APPENDIX 1: VARIABLE IDENTIFICATION KEY Appendix 2: Documentation of the investigation Appendix Appendix Appendix Appendix Appendix Appendix Appendix Appendix Appendix

6 Preface The aim of this paper is to investigate the difference between bare root plants and containerized plants (Jiffy), with regard to root architecture and tree stability. To give the reader an indication of the structure of the investigation, a reader s guide has been made. The guide will not give an insight in the results or conclusions, but merely an overview of the connections between the different chapters. Figure 1: Readers guide. Definition of variables, (Appendix 1) The reader is provided with a list of the variables and abbreviations. See appendix 1. This list is especially important in the parts Material and method and Result, where it will serve as a variable identification key. Appendix 1 is made as a flip out page. 6

7 1. Introduction Containerised forest plant material was introduced during the 1970 ies in Europe. Problems with root deformation arose with the Paper pot container and became obvious already one decade later (Lindström 1998, Lindström and Rune 1999). Problems partially consisted of the deformed stems from tilted trees (Rune 2002) and partially the expectations of poorer storm stability, when the stands grew old. The history of the containerized plant is divided into three generations. The first generation of containerized plants were hard pots and paper pots. These pots gave rise to root spiralling, which is when the roots meet the pot edge; it started spiralling around the edge of the container. When it was transplanted into the forest, the trees had difficulties escaping the spiral tendency and the mentioned problems arose (Nielsen 2004). The second generation of containerized plants were also hard pots, but in order to prevent the roots from spiralling, the pots were equipped with vertical ribs. Instead of spiralling this lead to so called root knees, where the roots started growing up and down in the container, and problems as mentioned before were also made by this type of container (Nielsen 2004). The third generation of containerized plants uses root pruning, mechanical, chemical or air pruning. The root pruning is done to stop the roots from growing when they meet the walls of the container (Nielsen 2004). The Jiffy pot is of the third generation of containerized plants and is relying solely on air pruning, meaning that the roots will grow until the edge of the pot, where they will stop due to lack of moisture. The sphagnum plug of Jiffy is containerized into a nylon net, which the roots can penetrate anywhere, when it is transplanted into the forest. In the nursery the Jiffy containers stands separated so the roots do not grow into the container next to it e. g. air pruning. Jiffy plants were introduced and put into production in Denmark by Peter Benfeldt with the nursery company Danverde A/S. A bare root plant is grown in the nursery seed bed during the first growth season. Then it is excavated and transplanted into an outside transplant bed, when the second growth season takes place. The plant can be in the transplant bed for different amounts of time (Nielsen 2004). The problems with the bare root plants are to a large degree in the planting methods, but also in the excavating process from the seed and transplant beds. There are mainly two ways of planting bare root plants, namely planting with spade and planting with a planting machine. Planting with a spade leads to root disorientation and root cutting just before planting, in order to fit the plant to the planting hole. Moreover this may lead to death of the initial root system, and the plant must rely on adventives roots to build the root system. Planting with a machine leads to a root direction opposite the planting direction and thereby root deformations (Nielsen and Ditlevsen 1997, Ditlevsen 1998). 7

8 In the excavating process, damage made to the roots is dependent on how well the soil is loosened before excavating. If the soil is poorly loosened, the roots can be damaged (Nielsen 2004). In cooperation with Forest and Landscape on the Faculty of Life Science (LIFE) and the Christmas Tree Union a field experiment was established in 1994 on Danverde A/S. The experiment contained the first prototypes of Jiffy plants, in comparison with bare root plants (See appendix 2). An initial investigation was conducted in 1998 by Kristian Sönnichsen and Christian Nørgaard Nielsen (Sönnichsen 2002). The investigation indicated a better root architecture in the Jiffy plants, but the plant material was still very young. The investigation done by Kristian Sönnichsen indicated that the root architecture of Jiffy plant were better than the bare root plants especially regarding root symmetry. In theory this would lead to a higher storm resistance, when the trees grow older (Nielsen 1999). 1.1 Hypothesis It is believed, that the root system of Norway spruce (Picea abies) grown in a Jiffy pot, will develop better and stronger root architecture than bare root Norway spruce and hence makes the Jiffy trees more resistant to wind throw. The hypothesis will be tested both for resistance to wind and for root architectural differences. All other factors will as far as possible be excluded. Two external factors can although not be excluded; namely soil differences and competition status of the tree. These factors will be included in the investigation. 1.2 Statement of intent The problem is how to test the hypothesis in a comprehensive way. This will be done first by reviewing the literature on the subject so far to assess the level of knowledge on the subject. On basis of the obtained knowledge in the literature review, ways of testing the hypothesis will be developed in Materials and Methods. The aim of the tests is to assess differences between bare root plants and Jiffy plants. The tests will be divided into four groups, in order to answer the four questions stated below: 1. Are there any differences in architecture of the structural root system between bare root and Jiffy trees? 2. Are there any differences in stability of the trees between bare root and Jiffy trees? 3. Is there a relation between root architecture and stability and does bare root and Jiffy trees behave different with regards to this relation? 4. Is there a difference between the two samples (Jiffy and bare root) with regards to the two foremost external factors, namely soil quality and competition status of the tree and if so, does this have any influence on results answering question 1 to 3? On basis of the results, answering these questions and a discussion of the results, an approval or rejection of the hypothesis will be made. 8

9 2. Literature review and reasoning of the hypothesis As described in the introduction, there is substantial difference in the method of planting between a bare root plant and containerized plants (in this case Jiffy container). Where a bare root mostly is planted by spade or by machine, containerized plants are mostly planted by a planting tube. This leads to the indication and hypothesis, namely that this difference in planting method would make the containerized plant, develop a better and stronger root system and hence increase stability of the containerized plant (Nielsen 2004, Nielsen and Ditlevsen 1997, Sönnichsen 2002). Theoretical arguments of this statement will be analyzed in this chapter. 2.1 Planting and development of the structural root system Coutts (1983a) describes that the differentiation of the roots, meaning at which age the structural root system is developed, to a large degree happened within the first six years after planting in the forest. Some of the developed roots were from the nursery root system, and some of them were adventitious roots developed from the stem base after planting in the forest (Ibid). This means that the structural root system of the tree to some degree is developed as a consequence of the quality of the planting. Better planting leads to more surviving root from the nursery. Some of the structural roots were although adventitious roots which were developed not from the nursery root system, but after planting. This indicates that the tree to a certain degree has the ability to adapt to an insufficient planting (Coutts 1983a, Nielsen 2004). Coutts (1983a) finds that there is a large variation in how the cross sectional area (size) of the roots were distributed between the individual roots. The largest roots that accounts for 80% of the total root CSA, is considered by Coutts (1983a) to be dominant roots and these roots are those of the structural root system. It was found that the distribution of these 80% of CSA varied from between 3 to 11 dominant roots. Nielsen (1990), Coutts (1982 & 1977) found that the explanation of differentiation of roots is found in the internal competition for carbon hydrates and in the size of the apical bud. These factors determine which roots are developed into the structural root system. The trees ability to form large apical buds must be determined by genetics more than anything else and will therefore not be treated in this study. The internal competition for carbon hydrates can be determined by more factors. The two foremost factors is the local environment around the individual root, meaning that a root in nutrient rich soil is a stronger competitor in the internal competition for carbon hydrates, than a root in nutrient poor soil. Excluding this factor the size and overall wealth of the individual roots after planting is a determinant of the differentiation of the roots, since a larger root provides more nutrients to the tree than a small root and the same for a healthy over an unhealthy root (Ibid). 2.2 Symmetry and method of planting Sönnichsen (2002) indicated that containerized Jiffy plants developed a more symmetric root system, than bare root plants. Relating this to the earlier mentioned differentiation of the root system, gives rise to the question: Does better initial root symmetry provide a better symmetry in the structural root system later on? The reasoning for this question is as follows. The better symmetry indicated by Sönnichsen (2002) gives rise to a more equal internal competition for carbon hydrates just after transplant to the forest and therefore more roots will be serious participants in the competition of carbon hydrates. This should in theory lead to a more symmetric structural root system (Nielsen 1998). 9

10 Another question arises: Will this then lead to a larger root system, since more roots could develop into structural roots? 2.3 Storm stability and components of the root system There are several forces working when a storm is raging. The first is the wind blowing at the crown, which is indicated by the horizontal arrow in figure 2.1 below. When the tree is bending the weight of the crown is also creating a force since it is not placed directly over the gravitational center of the tree anymore and the weight of the tree is now pulling down. These forces are distributed via the stem to the root system. When the force of the tree load and the wind load exceeds the forces of anchoring, the tree will be thrown over (Coutts 1983b & Nielsen 2004). Figure 2.1: Illustration of the forces by wind, acting in the crown of the tree. 10

11 Figure2.2: Illustration of the forces working below ground In the root system there are several forces working as well. First of all there is the soil tension, which is the force it takes to break the soil apart between the root ball and the surrounding soil (Coutts 1983a). The size and weight of the root ball is also a determining factor for stability. The root ball is strongly determined by soil type, maximum root depth, intensity of the vertical roots and the symmetry of the root system. See figure 2.2 above, Nielsen (1999) & (2004). Second there are the tension roots, which are the roots that continue horizontally and vertically out of the root ball into the surrounding soil on the windward side of the tree. These roots are enhancing stability by holding the root ball down. See figure 2.2 left side, (Nielsen 2004) The supporting roots, also called the lever arm roots, are the thick dominant roots around the stump on the lee side of the tree, which provides support due to large bending strength. The spot where these roots start to bend is known as the turning axis, or the hinge. The lever arm roots contribution to the moment of anchorage strongly depends on mass, vertical diameter, branching and taper of the roots close to the stump on the lee side of the tree. Symmetry of the root system also has importance for the supporting roots and its contribution to the tree stability (Nielsen 1999 & 2004). Winching of a tree, a static method, is an appropriate method to simulate storm, in order to test for stability, since there was found similar results using dynamic methods (Coutts 1983b). The method will be used in this investigation to simulate wind load and weight of the crown, as described above. 2.4 Theoretical basis for modeling stability A variety of modeling has been done to estimate the effects of the above mentioned root architectural traits. Coutts (1983b), figure 2.3 finds that the largest stability enhancing effect is of the tension roots (wind ward roots) at the maximum moment of anchorage, which is the maximum force the tree can 11

12 resist without being thrown over (Stability), this is shown in figure 2.3 by the maximum of the total resistance curve. Figure2.3: Coutts 83b findings of different root component importance. Figure: 2.4: Nielsen 90 findings of the importance of the different root components, with different competition statuses. The weight of the root soil plate follows as the second largest force and the soil shear is the third. Soil tension does not have any effect when maximum moment is reached and the supporting roots only have a minor effect. Nielsen (1990), figure 2.4 finds that there is a large variation in which of the architectural trails, are the strongest stability determining factor. If the tree is dominant, results similar to Coutts (1983b) are found, but if it is a dominated tree the distance to the supporting roots now plays a more important role than the root ball weight. Nicoll et al (2008) models the anchorage in two different ways. The first is a resistance model where the maximum moment of anchorage is estimated as a function of the distance from the center of gravity to the windward edge of the root plate. The second model of Nicoll et al (2008) was the same as the first, but the weight of the root ball was added to the model. The first model was the best performing (Ibid). Since the distance from the gravitational center to the windward edge was not measured in this investigation, this modeling approach cannot be used. Nicoll et al (2008) although found a proportional relation between the lateral diameter of the root ball on the wind ward side of the tree and diameter at breast height (DBH). This coincides with what Nielsen (1990) found, namely that the root ball was strongly determining stability for dominant trees e.g. trees with a large diameter, but the root ball did not have a large effect on stability for dominated trees e.g. small diameter trees. Furthermore both Nicoll et al (2008) and Nielsen (1990), states that the maximum rooting depth also was a strong determinant for the size of the root ball. The influence of the tension roots were estimated by Coutts (1983b), where this effect was separated from the other root component effects. Measurements like this were not made in this investigation. Instead tension roots will here be represented as a sum of the cross sectional area of the horizontal 12

13 root, together with the sum of cross sectional area of the vertical roots including any tap root, as a function of distance to the hinge and average distance from the stump. In estimating the supporting roots/lever arm, the direct measure of the distance to the hinge from the field, will be used in the model (see Material and Method part 3.4.1). 2.5 Wind induced damage on stand level It is known that when the wind is blowing on the forest edge it starts to make turbulence above the forest canopy (Skovsgaard, Finnigan and Brunet 1995). This results in wind gusts hitting the forest canopy, with very large strength. It is also well known that these gusts will cause sporadic wind throw or root ball loosening in the forest (Nielsen 1990). It is therefore highly likely that the sporadic wind throw starts where the trees are weakest. The weak trees could be trees with an undesirable architecture of the structural root system. When the sporadic wind throw has taken place, there are holes in the canopy, which allows the wind to get a better grip on the surrounding trees. Furthermore the social stability in the area of the sporadic throw will be significantly reduced, because the trees can no longer lean towards each other. Therefore a basis for a larger wind throw now exists (Nielsen 2004). In relation to this experiment it is of interest if the two plant types behave in different manners regarding the above mentioned weak points in the forest, where sporadic wind throw first occurs. 2.6 Wind induced damage on single tree level There exists three kinds of damage wind can make to trees. The first is breakage of the stem, which occurs when the force of the wind and the root system is greater than the bending strength of the stem. This will not be analyzed in this investigation. The second is root ball loosening, which is when the force of the wind is not greater that of the root system, but substantial damage is made to the root system. This results in a weakened root system, which has difficulties in repairing itself due to swaying of the tree that will make the root ball move. Any small roots that grow outside of the root ball may be broken even in smaller wind gusts when the root ball is moving. This will reduce the growth ability of the tree and increase the risk of rot entering the root system and will thereby reduce the mechanical adaptation and individual stability of the tree (Nielsen 2004). The severity of root ball loosening is differential and will in this experiment be expressed in a mild and a severe form. The mild form is represented as the first larger root breakage and the more severe form will be expressed as the first a larger series of root breakages occur. The last form of damage is when the force of the wind and crown weight exceeds the maximum moment of anchorage and will be expressed as the maximum force the tree can resist (maximum applied force) without being thrown over (Nielsen 2004). 2.7 Sum up The theoretical review above revealed that there could be a substantial difference in how the root system of a bare root and a Jiffy is developed. Relating this to stability revealed that there is basis to assume that the better root architecture will lead to a better stability. Several models on how the root system influences tree stability were reviewed in order to build models in the Materials and Methods. Also it was found that both the competition status and the soil quality, in previous investigations, had a significant influence on root development and stability. Last it was reviewed which damages wind can induce on a tree and how this would affect the stand as a whole. 13

14 In the following two chapters Materials and Methods and Results will be presented. The theory reviewed above serves as a basis for developing models to test the hypothesis. In the Materials and Methods and Result chapters, appendix 1: Variable identification key, is an important tool to assess the meaning of the different variables used in the investigation, starting in part Materials and Methods Introduction Before going in the field, an experimental design must be carefully planned. There has to be questions that needs to be answered, a clearly defined hypothesis and one must produce the strongest evidence possible to challenge the hypothesis (Ruxton 2003). Challenge of the hypothesis and relation to the statement of intent As described in the introduction, one must challenge the hypothesis as much as possible. In order to do this the experiment must be designed the best way to reduce the noise (errors) as much as possible and only to catch the natural variation in the trait of interest. An appropriate statistical method should be used. The statistical method should utilize the information in the gathered data as much as possible. As described in the Statement of intent, four questions needs to be answered in order to test the hypothesis in a comprehensive way. In the following the data basis for the investigation will be explained and the investigation field will be described. Furthermore tests and models, which can answer the four questions raised in the statement of intent, will be developed. 3.1 The experimental field The experimental field was designed in 1994 as a randomized block design, containing; Norway spruce Picea abies, Scots Pine Pinus sylvestris, Picea omorica, Abies normandie and Abies nobilis. Six plots of every tree species were established, three using bare root plants and three were using Jiffy plant. An extra Abies Normandie and an extra Abies nobilis plot were also present in the field. This will not affect the result here since only Norway spruce is investigated. Each plot contained 49 individuals. A forest worker with no interest in the result did the planting in 1994 (see appendix 2). The experimental field which was divided into 32 plots, containing 49 trees, furthermore it Figure 3.1: The investigation field 14

15 was divided into three blocks, two of them containing 10 plots and the most southern containing 12 plots. The block division is done due to changes in the soil quality. The rooting depth and content of sand varied considerably between the blocks. Only Norway spruce (RGR in figure 3.1) is considered in the investigation. In figure 3.1 the plots of interest are shown (Jiffy colored in light green and bare root in dark green) and the block division is indicated by the bold lines. A coordinate system was developed to identify the different plots, with the north eastern corner as (1,1), within each plot the trees were also given coordinates using the same method, where the northeastern tree was (1,1). 3.2 The sample All Jiffy and bare root plots were represented in the sample. This is done in order to be able to overcome block specific characteristics e.g. Soil types, moist regime, water logging, cemented layers, east/west gradients and shelter effects from other plots, etc. Within each plot the sample trees were randomly selected. The sample size was selected at 20 trees pr. plant type. Two bare root trees were excluded from the sample due to failure of the experimenting process. 3.3 Root architecture The aim of this part of the investigation was to collect data that was applicable for program Treearch (Nielsen & Dencker 1995), which provided variables that could describe the root architecture of the trees. Ways of testing the root architectural variables obtained was here developed in order to answer question 1 from the Statement of intent: Are there any differences in architecture of the structural root system between bare root and Jiffy trees? Data collection The root systems were dug up and measurements were taken. A mini digger was hired to do the excavating of the root systems. The root systems were cleaned with high pressure water pistol and placed bottom up, before the measurements were taken. The measured parameters of interest are: 1. the diameter of the individual roots, both largest and smallest at 20 centimeters from the stump center for the horizontal roots and 25 centimeter below ground for the vertical roots; 2. the compass direction of the roots; 3. the vertical angle of the roots. To do these measurements a compass was placed on the root system in order to find the compass direction of the individual roots. A vernier caliper was used to measure the diameters of the roots. In addition to the above mentioned measurements, counts of planting damage, root turnings and root brushes were taken Transformation of data The collected data was entered into Treearch database (Nielsen & Dencker 1995). The program gave many different root characters. In this analysis the main emphasis will be put on the variables of: cross sectional area of the roots, number of roots, number of dominant roots, maximum angel between dominant roots, maximum angel between all roots and symmetry parameters. 15

16 Additionally a variable was constructed in order to assess the overall measure of root system symmetry. It will be based on SYMMINS1H1120 and SYMMAXS1H1120 for 120 degree parts expressed as a ratio. SYMMIN/SYMMAX is the relation between the smallest and the largest 120 degree part of a root system, measured in CSA. The closer this variable gets to one the better overall root symmetry the root system has. Analysis of variance was used to analyze the data and to assess any difference between the two plant types occurred. 3.4 Stability: Maximum applied force, first root breakage and first large series of root breakages The aim of this part of the investigation was to collect data that could be tested in order to answer question 2 from the Statement of intent: Are there any differences in stability of the trees between bare root and Jiffy trees? Experimental design Before the experiment was initiated all branches of the trees were removed to a height of 2.3 meter. Then the tree was cut in the height of 2.3 meters in order to exclude any stability enhancing factors from the neighboring trees e.g. social stability and crown weight. Cutting the top of tree was only done for some of the experiment. (See correction for crown weight in nine trees.) Now the trees were ready for experimenting. In the above mentioned way of sampling all biases should be reduced as much as possible. As described in the chapter Redesigning the experiment the stem should be supported by a metal pole attached to the tree by the use of clamps for the second sampling round. This was done in the second sample due to a large breakage percent especially in the Jiffy plants within the first sampling round. Here 11 out of 25 (44%) of the Jiffy trees broke and 5 out of 23 (22%) of the bare roots broke. In the second sampling 6 Jiffy and 2 bare roots trees were sampled, with 0 breakage percent. For simplicity the applied force (measured in kilos) is used in the statistical analysis instead of the moment of anchorage (Newton), because the lever arm e.g. the height where the tree was pulled down, was 2 meters for all trees and the calculation from applied force into moment of anchorage would only be a scaling and would therefore not affect the result. 16

17 Figure 3.2: Experimental design, force measurements The tree was connected to a winch and to a force measurer. The amplitude measuring device, which measured the trees displacement from the gravitational center at two meters height and goniometry which measured the angle of the base of the stem, was connected to the tree and the microphone was placed in mineral soil approximately cm below soil surface, in order to tap the first root breakage and the first series of root breakages. The tree was now ready to be pulled down Data gathering In the stability part of the experiment, several datasets were gathered; the applied force curve for the tree was gathered; the root bursts were tapped by the microphone in the soil; the angle of the stem Figure 3.3: Applied force curve 17

18 basis was measured and the amplitude of the stem at two meters was measured. There are three interesting measurements regarding the applied force. The first is the force measured at the first root breakage. This measurement tells when the tree is starting to get injured. The second measurement is the force where the first larger series of root breakage occurs, which is when a more severe damage occurs. The last and most important measurement is the maximum applied force. This variable reflects the force it takes to throw the tree over and is the ultimate stability factor. All force measurements were corrected for the angle of the pulling cable, in order to reflect the force of a horizontal wind blow as shown in figure Redesigning the experiment for second sampling round In the it was revealed, that many of the stems could not endure the force from the pull, before the root was completely loosened. This gave rise to a breaking percentage of 22% for the bare root plants and 44% of the Jiffy plants. Due to this rather large breaking percentage especially for the Jiffy plants, a large bias in the sample occurred. This failure in the first round of data sampling, lead to a an additional sampling in the plots where this phenomenon mainly occurred, namely plot 2,2 for Jiffy and random distributed for the bare roots. The bare root second sample were taken in plot 3,6. Figure 3.4: Redesign of the experiment with a metal pole supporting the stem. The rest of the experiment was the same as shown in figure 3.2 In order to overcome this serious problem, the new sample was taken, with a different design of the experiment. The basic problem was the bending and shears strength of the stem, which was not strong enough to support the force to make the root ball loosen. There are two ways to solve this problem: 1) The height of the pull cable could be reduced from 2 to 1 meter in order to utilize the presumably stiffer lower part of the stem and then hope that this will be enough. 2) The second way to overcome the problem is to carry over the experiment exactly like before, but now the stem must be supported by a metal pole. To do this, an iron pole was attached to the stem. 18

19 Crocwn weight, kg. This was done by the use of clamps, both in the bottom of the stem and in the height of 2 meter where the pulling cable was attached. To be sure of not getting the same or a lesser degree of the bias from the first sampling round, alternative two was chosen. In this way it should be possible to get measurements from even the strongest root systems Correction for crown weight in 9 trees Some of the trees were not cut before experimenting. This lead to an excessive crown weight, because the trees were taken in groups, and there were no supporting trees next to them. In block 1.6 where the excessive crown weight was an issue. A correction model for the excessive crown weight was made. Three diameters each representing three diameter classes (small, intermediate and large) were used, 7,6 (small, DBH = [0;9,05]), 10,5 (intermediate, DBH = [9,05;11,5] ) and 12,5 (large, DBH = [11,5;...]) to model the crown weight for the different diameter trees. Furthermore trees for each diameter class were weighted at different amplitudes e.g. different displacement from gravitational center of the tree at 2 meters height (measured in millimeter), which is the measure of the amplitude from the experiment described in figure 3.2. Figure 3.5 shows how much extra crown weight is added for different amplitudes and diameters. An example of this is that a trees with a DBH around 10,5 centimeter that has its maximum applied force at an amplitude at 400 millimeters displacement from the gravitational center, will be added. Example 1: Y (Crown weight added to force measurement) = 0,1475*400-29,5 = 29,5 kg The equation above is from a linier regression of crown weight vs. the amplitude measured at maximum applied force. The model then tells how much force the crown weight accounts for, and this amount will be added to the measured maximum applied force Correction model for crown weight Amplitude, mm. Figure 3.5: Correction model for crown weight. Model DBH 7,6 y = 0,0575x - 6,5 R² = 0,9888 ModelDBH 10,5 y = 0,1475x - 29,5 R² = 0,9847 Model DBH 12,5 y = 0,1375x - 10 R² = 0,9774 Dia. 7,6 Dia. 10,5 Dia. 12,5 Lineær (Dia. 7,6) Lineær (Dia. 10,5) Lineær (Dia. 12,5) 19

20 Analysis of covariance revealed that this correction did not make any significant change in the applied force result (See appendix 3) Statistical method Analysis of variance (t-test) was used to compare the two datasets both on the maximum applied force and the diameter of breast height (DBH). Simple linier regression is used to assess at the relationship between DBH and the applied force for each dataset. Y = a + bx + ε Where Y (dependent variable = applied force), X (independent variable = DBH), a is a constant, b is a coefficients and is an ε = error term. In order to quantify if the method of planting (Jiffy or bare root), has a significant difference according to applied force, an analysis of covariance (ANCOVA) was used to assess this. Here the method of planting will be given a binary value, 0 for bare root and 1 for Jiffy Other field measurements (lever arm) The experiment was done by using a cable with a winch connected to the tree 2 meters above ground and to at tractor. The tree was pulled down until the root ball was completely loosened. Then the root lever arm (distance to the hinge) was measured, from the center of the stem base, to the point where Figure 3.6: Lever arm experimental design the soil started lifting on the pulling side (lee side) of the tree, when the winch pulled (See figure 3.6). Analysis of variance was chosen to analyze these data. The variable will serve as a root architectural variable in the further analysis. 20

21 3.4 Modeling the relation between root architecture and stability In the following the models used to assess the relation between root architecture and tree stability will be developed in order to answer question three from the Statement of intent: Is there a relation between root architecture and stability and does bare root and Jiffy trees behave different with regards to this relation? Theoretical approach for modeling root system and the relation to stability As mentioned in the literature review chapter a theoretical model was build, based on the different root components found by Coutts (1983b). The model to be testes here, were developed from the variables obtained from Treearch (Nielsen & Dencker 1995) The root ball component Nicoll et al (2008) found that the size of the root ball was proportional to DBH. In the same experiment Nicoll et al (2008) also found that the root ball size was strongly determined by the maximum rooting depth. It is here assumed that the relation between parameters used to model the root ball components are proportional. Nielsen (1990) & (1999), states that the root ball size is also dependent on the symmetry of the root system. The ratio SYMMIN/SYMMAX (see appendix 1) was chosen to reflect the overall symmetry of the root system. Since some of the root systems had 0 as SYMMIN and the ratio then will become 0, 1 will be added to the variable. The relative importance of the three components used to estimate the root ball is not known and this could induce a bias in the estimation of the root ball component. A real measure of this component as Coutts (1983b) did, would give a more reliable estimate of this component, but since this was not measured here the following estimation is used: Root ball component = DBH * Max rooting dept * (1 + SYMMIN/SYMMAX) The use of DBH in the estimation of the root ball component may cause an overestimation of the importance of this component, because DBH is strongly correlated with the force results Tension roots The tension roots will here be defined as a part that is the vertical roots and a part of the horizontal roots. Vertical tension roots The influence of the vertical roots is dependent on the root mass (CSA), which quadrant this mass is placed in and the distance to the hinge (See figure 3,7). The western and the half of the northern and southern quadrants are placed on the wind ward side of the tree, and are therefore both dependent on the distance to the hinge H and the average distance from the stump center. This average distances are here termed a for the windward half of quadrants north and south and b for the western quadrant (see figure 3.7). The importance of the vertical tension roots will here be estimated as: QW = (H+b)*(CSAQW) 21

22 WWSQNS= (H+a)*(1/2CSAQN+1/2CSAQS), where WWSQNS is wind ward part of quadrant north and south, QW is the western quadrant, CSAQW is CSA of the vertical roots in the western quadrant, CSAQN and QS is the CSA of the vertical roots in northern and southern quadrants respectively and H is the distance to the hinge. The eastern and the other half of the northern and southern quadrants are placed on the lee ward side of the tree and will be estimated the same way, but with opposite signs for a and b, here termed c and d (see figure 3.7). These are the same distances as a and b respectively, just on the lee side of the tree. QE=(H-d)*(CSAQE) LSQNS =(H-c)*(1/2 CSAQN+1/2 CSAQS), Figure 3.7: Root plate seen from above: Used in the calculation of the tension roots. Point a is the mean distance for the western quadrant to the stump center on the wind ward side of the tree. Point b is the mean distance of the northern and southern quadrant, the part placed on the wind ward side of the tree, to the stump center of the tree. Point c and d is the same distances as point a and b respectively, but on the lee side of the tree. H is the distance from the stump center to the hinge, measured earlier. stump. where LSQNS is lee ward side half of the northern and southern quadrant, c and d is mean distance to the stump for the respective quadrants and H is the distance to the hinge. Where the hinge is less that c and d the lee ward side expression will be normalized to 0, since a negative effect will make no sense. The tap roots are also included as a function only of the hinge, due to the fact that the tap root by definition is placed directly below the TAP = CSATAP*H The importance of the vertical roots will be estimated as a sum of the above described equations. The effect of the distance to the hinge is assumed not to be influenced by the angle between the lever arm and the root plate, since all the roots will be broken within a few degrees of lifting the root soil plate and the angling will have only insignificant effect on the result. Horizontal tension roots The horizontal tension roots will be estimated as a function of the mean distance to the stump center for the different quadrants on the wind ward side of the tree (a and b) and the distance to the hinge (H). Only roots on the windward side of the tree will be considered, since horizontal roots on the le ward side of the tree by definition are supporting roots. 22

23 The importance of horizontal tension roots will be estimated as a sum of the two equations below: QW=(H+b)*(CSAQW) QNS = (H+a)*(1/2 CSAQS+1/2 CSAQN), where a and b are the same as for the vertical roots. QW is the western quadrant, QNS is the wind ward side of the northern and southern quadrant, CSAQW, CSAQS, CSAQN is the CSA of the western, southern and northern quadrant respectively. The reasoning for including the distance to the hinge both for vertical and horizontal roots was to give an indication on the relative importance of the roots in the different quadrants, based on physical arguments. The accuracy of the modeling for horizontal tension roots are not as precise, as it is not quantified where these roots breaks. The relative importance calculated above on basis of the hinge and the mean distance to stump center, will be disturbed because it is likely that the horizontal tension roots breaks at different distances from the stump. Here it is assumed that the breakage of horizontal roots occur the same place for all trees Supporting roots The supporting roots are the ones that make up the distance to the hinge. Therefore the direct measure, from the field, of the distance to the hinge will be used. The model is then: Max applied force y = α + β 1 root ball + β 2 tension roots + β 3 distance to inge + ε Multiple regressions will be used to estimate the coefficients for the root components Empirical approach for modeling root system and influence on stability Two empirical systems were developed to assess the importance of the root system on stability. The first was made to find the foremost root variables (from Treearch database), which determine stability and the second is to determine an overall performance of the root system and its influence on stability Assessment of stability determining root parameters To find out which of the root architectural variables is determining stability correlation tables were made. The variables that correlate with stability (max applied force) were used to model the root systems influence on tree stability. Multiple regressions were used to model Assessment of overall root system performance and influence on stability In order to assess the overall quality of the individual root systems a scoring system was made for each of the root variables. This was done due to the fact, that if all the obtained root variables from Treearch were included in a multiple regression, the R 2 increased from 0,56 to 0,94. This indicated that there was some explanatory power within the obtained root variables although most of the variables were statistically insignificant by themselves and those which were significant gave little causal explanation. 23

24 The scoring system was based on a normal distribution, where both the bare root and the Jiffy samples were included. A common mean and a 95% confidence interval (CI) for the two samples were made for each chosen variable. The score of each variable of the individual root system was determined as: 1 lower than the 95% CI; 2 from the lower 95% of the distribution to the mean; 3 from the mean to the upper 95% CI and 4 for larger than the 95% CI. For variables where the largest value represents the worst case the scale was reversed. Naturally some of the real Figure 3.8: Scoring system for root architecture variation in the different traits will be lost due to the aggregation of the root score, so the high R 2 found including all trails is not expected here due to the loss of variation Variables included In order to get the most plausible root system definition on the basis of the obtained variables from the TREEARCH database, the variable to make the scoring system was based on theoretical arguments and the use of correlation tables to assess auto correlation. If some variables auto correlate, the most theoretically plausible variable was used for scoring basis. Root size (CSAS1H1QT/DBH) The size of the root system is a stability enhancing factor. This is best described on basis of the cross sectional area of the whole root system (CSAS1H1QT). This variable is strongly influenced by the size of the tree, therefore the root/shoot ratio (CSAS1H1QT/DBH) is chosen as the preferred variable to represent the size of the root system. Number of roots (NS1H1QT) Here the total number of roots was the preferred variable, because it is an indication of the overall root system health, and the roots ability to obtain nutrients and water. It also reflects the potential for future mechanical adaptation to storm in development of more structural roots. (Coutts 1983a) Number of dominant roots (N_DOM1S1H1QT, N_DOM2S1H1QT, N_DOM1S1QT and N_DOM2S1QT) The number of dominant roots was chosen because it indicates the root systems ability, both to create strong tension roots and to create good supporting roots. Where the vertical roots are included, this variable also gives an indication on the root systems ability to hold soil, which enhances the root ball size. Furthermore dominant vertical roots functions as vertical tension roots, that holds the root ball down as shown in figure 2.2. Here both the definition by (Coutts 1983a & Nielsen 1990), are taken into consideration with equal weights. 24

25 Maximum angle between dominant roots (MAXANGCOS1H1 and MAXANGNIS1H1) This variable is chosen, because it gives an indication if there is a problem with too large angles within the individual root system. Here both (Coutts 1983a & Nielsen 1990) definitions are again taken into consideration. Maximum angle between all roots (MAXANGALS1H1) This variable gives an indication if there is potential in the root system to adapt large to large angles between dominant roots in the future. Symmetry parameters (SIMMINS1H1120 and SYMMAXS1H1120) Here the symmetry parameter based on 120 degree division of the root system is used. These variables are the strongest symmetry parameter. Both the largest 120 degree part and the smallest part were given individual scores which were aggregated into an overall symmetry score. The overall root score was included in the regression of maximum applied force versus DBH to assess the importance of the performance of the root system Distribution analysis and assessment of very poor performing trees In this part it was the aim to assess the trees that fell well out of line with the lower part of the 95% confidence interval of the different root architectural parameters. This was done in order to find out how many of the trees within each sample, Jiffy and bare root, that performed extremely poor. Furthermore it was analyzed whether the trees that performed very poor on root architecture also were the trees that performed very poor on stability (maximum applied force). The performance of tree stability was seen in relation to the size of the tree, since a small tree will not have to have as high stability as a large tree. So what was actually looked for, were the very lower part of the regressions of maximum applied force versus DBH. The analysis was based on the distributions and confidence intervals (CI) of the different root architectural parameters and the distribution and CI of the maximum applied force in relation to tree size (DBH). Figure 3.10: Distribution analysis: An example of assessment of poor performing trees. Here a tree with 4 or less dominant roots will be considered a poor performing tree. N Bare root = 18, N Jiffy =