<%BANNER%>

Effect of Pruning Type, Pruning Dose, and Wind Speed on Tree Response to Wind Load


PAGE 1

EFFECT OF PRUNING TYPE PRUNING DOSE, AND WIND SPEED ON TREE RESPONSE TO WIND LOAD By SCOTT ALAN JONES A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2005

PAGE 2

Copyright 2005 by Scott Alan Jones

PAGE 3

To tree care professionals in all their varieties.

PAGE 4

iv ACKNOWLEDGMENTS This project was beyond my individual capacit y. I would be ungrat eful if I did not acknowledge the many creative minds and he lping hands that contributed to its completion. I want to first thank Dr. Ed Gilman, my committee chair, who provided the sure foundation for the work that was done. The project clearly would not have been possible without his interest and support. He gave me liberty to manage the project as I saw fit and was invariably patient with my personality and demands. He was a fine mentor and will remain an excellent example and friend. My graduate committee was better than I could have imagined. Dr. Perry Green provided a desperately needed engineering dimension to the committee. I want to thank him for his patient tutoring as I spent many hours at his desk or in his lab designing and testing various parts of the experiment, the li ons share of which are not even represented in this publication. I want to thank Dr. Richard Beeson for his limitless generosity and invaluable experience. The data acquisition system would not have been built or functional without him. And I want to th ank Dr. Jason Grabosky for his loyalty and enthusiastic moral support. I always found his door open desp ite the geographic distance. Every member of my committee was excepti onal and I will forever remember their instruction and care. I should next thank the benef actors that funded the projec t. I want to thank Dr. Gilman again here. He and the Department of Environmental Horticulture at the University of Florida provided a research assistantship that allowed me to work on a

PAGE 5

v project of my choice. The Tree Research and Education Endowment Fund awarded the project a Hyland R Johns Grant with which we purchased much of the needed equipment and supplies. Marshall Tree Farm donated the trees that were used and they were kind enough to let me visit their farm and hand-pi ck the trees I wanted. Rinker Materials donated 5 yards of concrete, with delivery, fr om which the pier walls were built. Sundry odds and ends were donated from Dr. Greens and Dr. Beesons other research interests. A few other people deserve special recogni tion here. Chris Harchick, the manager of the Environmental Horticulture Teaching Unit, was my right arm throughout much of the project. The dirty work would not have been done without him. Alison Trachet, an undergraduate student in civi l and coastal engineering, co llaborated on the efforts to determine the Youngs Modulus of the trunks and assisted in almost all the data collection. Chuck Broward, la boratory manager in civi l and coastal engineering, facilitated the Youngs Modulus work and was al ways generous with the resources at his disposal. Dr. Kurt Gurley, an associate professor in civil a nd coastal engineering, measured the downwind wind speed profile ge nerated by the airboa t and spent several hours teaching me about wind and how to mana ge my data. Dr. Kenneth Portier, an associate professor in the Institute of Food and Agricu ltural Sciences Statistical Consulting Unit, and Marinela Campanu, one of Dr. Portiers graduate assistants, provided needed assistance with the statistical analysis. Justin Sklaroff, Patricia Gomez, Maria (Pili) Paz, Amanda Bisson, Ryan Yada v, and Jon Martin all provided extra hands for some of the mundane manual labor. I thank them all. Finally, I want to thank my family and fr iends for their love, attention, and support despite my preoccupation with AMELIA.

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES...........................................................................................................ix ABSTRACT.......................................................................................................................xi CHAPTER 1 INTRODUCTION........................................................................................................1 2 LITERATURE REVIEW.............................................................................................3 Significance..................................................................................................................3 History........................................................................................................................ ..4 Engineering Principles..................................................................................................5 Scientific Approach......................................................................................................6 Wind Forces...........................................................................................................7 Drag coefficients............................................................................................7 Wind speed.....................................................................................................9 Tree Resistance....................................................................................................11 Bending and wind snap................................................................................12 Wind throw...................................................................................................15 Tree Dynamics............................................................................................................16 Pruning........................................................................................................................17 Dose........................................................................................................................... .19 Conclusions.................................................................................................................21 3 MATERIALS AND METHODS...............................................................................23 Tree Selection.............................................................................................................23 Experimental Design..................................................................................................25 Pruning Type..............................................................................................................25 Lions Tailing......................................................................................................26 Raising.................................................................................................................27 Reduction.............................................................................................................28 Structural.............................................................................................................29

PAGE 7

vii Thinning..............................................................................................................29 Pruning Dose..............................................................................................................30 Wind Speed.................................................................................................................30 Trunk Deflection.........................................................................................................33 Experimental Procedure..............................................................................................34 Statistical Analysis......................................................................................................36 4 RESULTS...................................................................................................................65 Randomization............................................................................................................65 Pruning Dose..............................................................................................................65 Wind Speed.................................................................................................................67 Response.....................................................................................................................68 Analytical Approach...................................................................................................70 5 DISCUSSION AND CONCLUSIONS....................................................................105 6 FUTURE WORK......................................................................................................116 LIST OF REFERENCES.................................................................................................119 BIOGRAPHICAL SKETCH...........................................................................................129

PAGE 8

viii LIST OF TABLES Table page 3-1. Measurements used to determine physical similarity of trees tested.........................37 3-2. Downwind profile of wind speeds generated by an airboat......................................39 3-3. Calibration of motor rpm to wind speed....................................................................40 3-4. Wind speeds recorded during a road test...................................................................41 4-1. Foliage and stem weight for 13 tr ees harvested to quantify pruning dose................74 4-2. Percent foliage dry wei ght, stem dry weight, and tota l dry weight removed with each pruning dose.....................................................................................................75 4-3. Wind speeds (mph) recorded during testing..............................................................79 4-4. Wind speed (mph) by pruning type and motor rpm..................................................79 4-5. Wind speed, trunk movement (m54), and deflected area (dya) at 0 pruning dose and 2750 rpm............................................................................................................80 4-6. Regression coefficients and R2 values generated using all measurements................81 4-7. Regression coefficients and R2 values generated using averages..............................83 4-8. ANOVA of predicted trunk movement (p_avm54) and predicted deflected area (p_avdya)..................................................................................................................85 4-9. Least squares means of predicted average trunk movement (p_avm54) due to interaction of pruning type and wind speed type by wind speed..........................86 4-10. Least squares means of predicted average trunk movement (p_avm54) due to interaction of pruning type and wind speed wind speed by type..........................86 4-11. Least squares means of predicted average trunk movement (p_avm54) due to interaction of pruning dose and wind speed dose by wind speed.........................87 4-12. Least squares means of predicted average trunk movement (p_avm54) due to interaction of pruning dose and wind speed wind speed by dose.........................87

PAGE 9

ix LIST OF FIGURES Figure page 2.1. Wind forces affecting trees (modified from Grace (1977)).......................................22 3.1. Three examples of lions tail pruning taken from urban landscapes.........................42 3.2. Example of geometrically defined lions tail pruning...............................................43 3.3. Example of visually defined lions tail pruning.........................................................44 3.4. Two examples of raising pruning type taken from urban landscapes........................45 3.5. Example of geometrically defined raised pruning.....................................................46 3.6. Example of visually defined raised pruning..............................................................47 3.7. Two examples of reduction pruning type taken from urban landscapes...................48 3.8. Example of geometrically defined reduction pruning...............................................49 3.9. Example of visually defined reduction pruning.........................................................50 3.10. Two examples of structural pruning t ype taken from urban landscapes before (A and C) and after (B and D) respectively.............................................................51 3.11. Examples of structural pruning................................................................................52 3.12. Example of thinning taken from urban landscapes before (A) and after (B)........53 3.13. Example of thinning.................................................................................................54 3.14. Airboat used to generate wind: A) side view and B) rear view...............................55 3.15. Calibration curves of the mean wi nd speeds as measured by Campbell (West and East) and Young (Gurley) anemometers...........................................................56 3.16. The four South cable extension transducers (CET).................................................57 3.17. Southern cable extension tran sducer (CET) calibration curves...............................58 3.18. Northern cable extension tran sducer (CET) calibration curves...............................59

PAGE 10

x 3.19. Schematics (birds-eye (A) and profile (B) views of cable ex tension transducer (CET) and anemometer positions.............................................................................60 3.20. Cable extension transducer calculati ons (these are for one specific height represented in inches by subscript #).......................................................................61 3.21. Apparatus used to determine longitudinal Youngs modulus..................................62 3.22. Apparatus used to fix the trunk a nd rootball of trees during testing........................63 3.23. Data acquisition system used in the field.................................................................64 4.1. Pruning dose represented as percen tage of foliage dry mass removed......................88 4.2. Vertical profile of average generated wind speeds....................................................90 4.3. Four profiles of wind speeds gene rated by the airboat during testing.......................91 4.4. Trunk movement measured at an elev ation 54 inches above topmost root...............93 4.5. Trunk movement measured as an area of deflection.................................................94 4.6. Four time series profile s of trunk movement at an elevation 54 inches above topmost root..............................................................................................................95 4.7. Wind speed (A) trunk movement (m54) (B ) at all pruning doses by date tested......97 4.8. Three dimensional scatterplots of runk movement at an elevation 54 inches above topmost root..............................................................................................................98 4.9. Three less desirable distributions of wind speed and pruning dose...........................99 4.10. A graphical summary of the procedur e used to generate predicted response values......................................................................................................................101 4.11. Interaction between pruning type and wind speed.................................................103 4.12. Interaction between pr uning dose and wind speed................................................104

PAGE 11

xi Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science EFFECT OF PRUNING TYPE PRUNING DOSE, AND WIND SPEED ON TREE RESPONSE TO WIND LOAD By Scott Alan Jones December 2005 Chair: Edward F. Gilman Major Department: Environmental Horticulture Three to four inch caliper, clona lly propagated live oak trees ( Quercus virginiana QVTIA PP 11219, Highrise) were used to test the effect of five pruning types and four pruning doses on trunk movement at four wind speeds. Pruning types evaluated were lions tailing, raising, reduction, stru ctural, and thinning. Pruning doses were 15, 30, 45, and 60 percent foliage dry mass removed. Wind speeds were 15, 30, 45, and 60 mph. Of the affects tested, wind speed ha d the greatest impact on trunk movement. However, interactions of wind speed with pr uning type and wind sp eed with pruning dose were also significant. At high wind speeds, thinning did not redu ce trunk movement as effectively as other pruning types. Trunk move ment increased with each increase in wind speed for all pruning types except structur al; for structural, differences in trunk movement were only significant between 15 mph and 60 mph wind speeds. Trunk movement also increased with each increase in wind speed when 15% or less of the

PAGE 12

xii foliage was removed. Removal of 30% to 45% foliage prevented increases in trunk movement until wind speeds reached 30 mph. Fo rty-five mph wind speeds were required to increase trunk movement when 60% foliage was removed. At low wind speeds, 15 mph or less, trunk movement was similar among all pruning types averaged across pruning doses, and across all pruning doses averaged across pruning types. Results indicate no pruning type effectively minimi zes wind loads at currently recommended pruning doses.

PAGE 13

1 CHAPTER 1 INTRODUCTION Large landscape trees contribute an irre placeable dimension to urban landscapes, but they also present a significant, yet oddly acceptable hazard. Large trees, weighing several thousand pounds, are often found ha nging precariously over homes, roads, recreational areas and other fr equently populated sites. Ye t relatively little is known about their construction and less about their re sponse to external lo ading. Some of the most damaging external loads trees confront include static loads generated by snow and ice accumulation and dynamic loads generate d by strong winds. This investigation focused on tree response to loads generated by strong winds. Wind storms often break and topple tr ees resulting in damaged property, interrupted utility service, and personal in jury. Pruning is regul arly recommended as a method of abating wind damage in trees. Da ta supporting that reco mmendation is scarce largely because interactions between wind and trees are elaborate. Large trees are complex, dynamically built structures and wind is extremely variable in time and space. Small trees have therefore been used in gene rated wind fields to simplify the relationship for investigation. This study investigated the use of pr uning to reduce wind loads on young (5 to 6 year old), Quercus virginiana live oak trees. Five co mmon pruning practices were evaluated. Effects of pruning dose and wind speed were also included. Wind velocities were generated by an airboat elevated so the propeller was at canopy height. Results

PAGE 14

2 from this experimental study can be used by tree care professionals to better manage individual trees in our urban forests.

PAGE 15

3 CHAPTER 2 LITERATURE REVIEW Significance Trees break apart and fall over in th e wind (Allen 1992). Matheny and Clark (1994) and many others note there are factor s (internal decay, poor architecture, age, human encroachment, etc.) which cause defects th at predispose trees to failure even in normally tolerable wind events. Many of the de fects that lead to mechanical failure are visible, and efforts have been made to t each tree care professionals to recognize and proactively address structural defects (Robbins 1986; Matt heck and Breloer 1994; Smiley and Bones 2000). However, some defects ar e not visible and some wind events are severe enough to damage and destroy even healthy, structurally sound trees (Duryea, Blakeslee et al. 1996). Wind damage to trees causes tremendous lo ss. Economic loss is clearly visible in forest systems where wind damage results in lost materials that might have been harvested for lumber or paper production. Wi nd damage is believed to cost countries in the European Union approximately 15 million Euros per year, and on occasions substantially more (British Forestry Co mmission 2005). Economic loss can also be attributed to clean-up and re storation of damaged property in urban environments (Ham and Rowe 1990). Costs are associated with debris removal (Whittier, Rue et al. 1995), insurance claims, and restorati on of utilities. Repair and replacement of the urban forest are also viewed as necessary in maintaining a functioning society with an acceptable

PAGE 16

4 standard of living (Westphal 2003). Worst of all, injury and lo ss of life are often associated with wind damage in both fo rests and urban settings (Graham 1990). History In 1977 J. Grace published a monograph entitled Plant Response to Wind One of his main objectives was to review the literatur e so that botanists, foresters, agronomists, etc. could evaluate the state of info rmation about how plants respond to wind physiologically, anatomically, ecologically, a nd mechanically. The interaction between plants and wind is a complex matter. In Ju ly 1993, the Internationa l Union of Forestry Research Organizations (IUFRO) brought together scientists from many disciplines for a first conference on Wind and Wind-Related Da mage to Trees. Proceedings of that conference were published in 1995 in a volume titled Wind and Trees (Coutts and Grace 1995). In 1998, IUFRO held its second confer ence on Wind and other Abiotic Risks to Forests. Selected papers from that c onference were published in Forest Ecology and Management issue 135 (Peltola, Gardiner et al. 2000). Most recently, a third international conference, Wind Effects on Trees, was held in September 2003 and its proceedings were published in text as were the first (Ruck, Kottmeier et al. 2003) but public copies are scant (in 2005 the University of Floridas Marston Science Library was unable to purchase a copy or acquire one through interlibrary loan). All three conferences aimed at understanding the effects of wind on forest systems and most of the work presented was done on coniferous species Roodbaraky et al. (1994) remarked that by 1990, there was already a body of literature dealing with the effect of wind on woodland conifers but very little existed for broad-leaved species. In the decade following his remarks, more research was conducted using broad leaf species (Vogel 1995; Niklas and Spatz 2000; Re bertus and Meier 2001; Vollsinger, Mitchell et al. 2005),

PAGE 17

5 but information is still scarce. Arborists in America, interested in the interaction between wind and trees in urban settings, organized an other conference on tree biomechanics held in March of 2001. Proceedings of that conf erence were published by the International Society of Arboriculture (Smiley and Coder 2002). A few other texts merit attention here for their influence on the primary literature. Steven Vogels Life in Moving Fluids (1994) is an excellent re source for biologists and engineers interested in the interface between biology and fluid mechanics. Pertinent subjects discussed include: pr inciples of fluid flow, drag, drag coefficients, biological strategies to reduce drag, and complexities of fluid flow like unsteady flows, velocity gradients, and boundary layers. Struct ures: or Why Things Dont Fall Down (Gordon 1981) is both an instructive and enjoyable r ead introducing pertinent material properties like stress, strain, shear, tors ion, and fracture as well as pe rtinent design considerations like safety and efficiency but always in man-made structures. Many of those same material and design principles are discussed in relation to plants in Plant Biomechanics (Niklas 1992). Finally, Wood th e Internal Optimization of Trees (Mattheck and Kubler 1997) presents an engineering approach to developmental biology of trees including a discussion of stress transfer through li ve wood, tree response to wounding, and the proposed axiom of uniform stress. Engineering Principles Wind damage to trees (as it is considered he re) is a structural ra ther than biological issue. Biological functions, like photo synthesis, nutrient assimilation, hydraulics, growth, reproduction, etc, are all determinants of tree growth and development (Ryan and Yoder 1997). Growth and development are de termined even more significantly by a trees genetics and evolutiona ry fitness with the latter not only influenced by the

PAGE 18

6 organism but by its species (Niklas 1998; 2000b) Still, trees are subject to the same physical laws of nature as any other engi neered structure (Niklas 1992; Savidge 1996). Engineering principles should th erefore apply to tree structures just as they apply to any man-made structure, and indeed they do (S chuler and Bruhn 1973; Mattheck and Vorberg 1991; Spatz, Kohler et al. 1999; Niklas 2000a; Fourcaud and Lac 2003). Constraints governing man-made construction differ from those governing tree growth and development. As a result, trees are able to employ strategies that human engineering avoids such as bending, twisting, and reconfi guration (Vogel 1995). In order to analyze a trees mechanical design, engineering princi ples have to be expanded to deal with complications like large deflections (K emper 1968), complex loading (Morgan and Cannell 1987), and composite material prope rties (Spatz, Kohler et al. 1999). Conversely, tree structure is restrained by effi ciency in the allocation of resources. Therefore, factors of safety (load capacity / self weight) in trees are much smaller than those found in man-made designs (Niklas 1999). It should be clear that an engineering approach to tree biomechanics is useful onl y with an appropriate consideration for biological elegance. Scientific Approach Mechanically, wind damage in trees has b een classified as wind tilt, wind throw, wind prune, and wind snap, (Allen 1992). Wind tilt and wind throw are defined as the inability of the roots and soil to resist upr ooting either partially or completely when lateral forces acting on a tree reach critical limits. Wind prune and wind snap are the inability of branches and trunks respectiv ely to resist breaking under those same conditions. Catastrophic mechani cal failure has been classified in similar fashion as soil failure, root failure, or trunk failure (Som merville 1979; Moore 2000). Soil failure is

PAGE 19

7 distinguished as wind tilt or wind throw acco mpanied by extraction of a characteristic root plate. Root failure is wind tilt or wind throw with minor soil heaving at the base of the trunk, but no noticeable root plate. Tr unk failure is wind snap. An analytical approach to studying and pred icting wind damage (especially catastrophic mechanical failure) was adopted in the 1970s and consiste d of comparing forces imposed by wind to forces required to break or t opple trees (Mayhead 1973; Grace 1977). Wind Forces Wind associated forces acting on a tree ar e shown in Figure 2.1. Detailed reviews of the mechanics associated with interac tions between wind and trees are available (Blackburn, Petty et al. 1988; Wood 1995; Sp atz and Bruechert 2000). The primary wind associated force acting on a tree is the drag force as given by Equation 2.1 (1f in Figure 2.1). ) ( ) ( ) ( 2 12z A z U z C DD air (2.1) Drag force ( D ) is equal to density of the fluid ( in this case air) multiplied by the drag coefficient ( CD), velocity of the fluid squared ( U2) and surface area projected into the fluid flow ( A ) at a given elevation ( z ). Drag force is compounded by the height of the stem over which it acts as described by Equation 2.2. 1 1) ( l D M (2.2) In Equation 2.2, M1 is the drag force moment, D is the drag force as in Equation 2.1, and l1 is the height on the stem ove r which the drag force acts. Drag coefficients Drag is the sum of bluff body pressure a nd skin friction components that vary in their magnitude depending on the shape and roughness of an object (Niklas 1992). The

PAGE 20

8 drag coefficient is a dimensionless consta nt that accounts for variation in shape, roughness, and all other oddities in the beha vior of drag which are described more thoroughly by Vogel (1994). Drag co efficients for forest trees were reported as early as 1962 (Mayhead 1973). Wind tunnels were us ed to generate known wind speeds and resultant drag was measured on both indivi dual trees and model forests (Fraser 1964). Mayhead (1973) reports drag coefficients fo r eight coniferous species at wind speeds likely to cause wind throw (i.e. [68.0 m ph], 30.5 m/sec) and proposes their use in predicting critical heights of trees (the hei ght at which the give n wind speed causes wind throw). He notes however th at they are a dangerous ex trapolation and that good predictive work will require more accurate values for the drag coefficient. Yet his values have been used in risk assessment studies as late as 2001 (Moore and Gardiner 2001). Rudnicki et al. (2004) and Vollsinger et al. (2005) followed up the work of Mayhead with improvements in the determinat ion of drag coefficients by accounting for streamlining, the speed specific reorientation of branches and leaves in the wind. Using digital video to capture wind speed specific frontal area ( Ad), they showed that at the highest wind speeds tested (20 m/s) Ad decreased by as much as 54% in the conifers and 37% in the hardwoods. Still, drag coefficients for hardwood species were less than half the values typically reported for needled c onifers at equivalent speeds (Vollsinger, Mitchell et al. 2005). Also, drag coefficien ts for both conifers and hardwoods were greater and didnt decrease as sharply with increasing wind speed when calculated using Ad, as when using the still air frontal area. Both papers confirmed Mayheads findings that drag coefficients vary among species, a nd they discouraged extrapolation to other

PAGE 21

9 species. Both groups also revisited some of Mayheads unpublished work and reported a linear relationship between drag and the product of wind speed ( U ) and canopy mass ( Mc) in all conifer and hardwood species tested. Therefore, they proposed a simplified drag equation for risk assessment, Equation 2.3 that eliminates the need for calculating frontal area and errors associated with using inaccurate drag coefficients. U M Dc (2.3) With this they recognized that the trees us ed were small (saplings 3-5 m in height) and that their individual branches behaved independently. Further work is needed to define drag relationships in older and larger trees. In a commentary A. R. Ennos (1999) argued there is little unequivocal evidence of drag reduction in large trees as a result of reconfiguration. He cites McMahon (1973) and Bertram (1989) while reminding readers that mature trees have thicker stems to cope with larger gravitational loads so they are less flexible. V ogel (1989) showed that leaves and clusters of leaves will of ten streamline when oriented appropriately in a straight line wind and thus reduce their drag, but stiffer leaves did not foll ow suit. Instead of scaling proportional to the first power of wind speed, dr ag on mature trees may scale like Vogels (1989) white oak leaf, at a power even larger th an that seen in the classical drag equation (Equation 2.1). Attempts have been made to measure drag coefficients in field grown trees but they are fraught with uncerta inty and are not re liable (Ennos 1999). Wind speed Definition of the vertical wind profile and sp ectra that cause damage to trees is an almost esoteric subject that is beyond the sc ope of this work. However, Lee (2000) and Finnigan and Brunet (1995) reviewed the l iterature on this subj ect for the 1998 and 1993

PAGE 22

10 IUFRO conferences, respectively. Wind pr ofiles are reported as mean predicted wind speeds. Calculations of drag coefficients in wind tunnels were conducted with straight line wind profiles (Mayhead 1973; Rudnicki, M itchell et al. 2004; Volls inger, Mitchell et al. 2005). Wind profiles used in theoretical modeling (Hedden, Fredericksen et al. 1995; Peltola, Nykanen et al. 1997; Kerzenmacher and Gardiner 1998) typically follow the theoretical profile pr esented by Oliver and Mayhead (1974), which is an exponential profile within the canopy and a logarithmic profile above it. Niklas (2000a) measured the wind profile used when calculating safety f actors and Nilas and Spatz (2000) measured the wind profile when calculating wind induced stem stresses in an open-grown cherry tree. These profiles were best described by a third order polynomial equation. Stem stresses were then recalculated using logarith mic, constant speed (straight line), square root, and square (exponential) profiles to compare among different vertical wind profiles commonly used or seen in nature (Spatz and Bruechert 2000). They reported that stress levels generated were insensitive to the s hape of the wind speed profile (Niklas and Spatz 2000) compared to other factors. None the less, variations in wind spectra are still thought to explain the random natu re of wind damage in trees (L uley, Sisinni et al. 2002). Lee (2000) noted there was still a dearth of information on wind flow over undulating terrain, in extreme wind events, and in inhomogeneous canopies with irregular, more realistic edge transitions and forest clearings ; all of which are common in urban forests. There is also very little information a bout wind speeds within and around individual trees. Zhu et al. (2000) repor ted that vertical and horizon tal wind profile s within the canopy of a single Japanese black pine fo llowed exponential functions described by elevation and crown thickness, re spectively. They noted that average wind speeds within

PAGE 23

11 the crown were only about half what they were outside it. They also proposed equations for calculating interior wind speeds at a ny elevation in a crown based on a single measurement outside a crown, and anywhere within a horizontal plane in a crown based on a single measurement outside a crown at the same elevation. Tree Resistance The second part of the analyt ical approach intr oduced above is a determination of the force required to break or topple a tree. Before any additional force is considered, a tree must first cope with the load of gravit y or its self weight as described by Equation 2.4 and illustrated in Figure 2.1. g l m f ) (1 2 (2.4) In Equation 2.4 f2 is the force of gravity; m ( l1) is the mass of the canopy at an elevation along the trunk l1; and g is gravitational acceleration. Calculations based on scaling of trunk diameter with tree height predicted a safety factor against elastic buckling under self-weight near four (McMahon 1973). Niklas (1994) and Ma ttheck et al. (1993) independently confirmed earlier calculations through experiment ation. However, Niklas (1997a) later argued that estimates of wood dens ity used in previous calculations were unrealistically low and safety factors against el astic buckling are probab ly closer to two. He also reported (Niklas 1997b; c) that ontogenetic change s in size, shape, and wood properties occurred of necessity or imposed stresses would reach critical levels as trees grew in size. Ontogenetic development al lows for trunk and proximal branches to be rigid in support of larg er gravitational loads resulting fr om increased mass. At the same time distal branches remain flexible so the canopy maintains an ability to reconfigure and

PAGE 24

12 reduce its drag in the wind. As a matter of perspective, most trees are capable of successfully resisting gravitational loads. Bending and wind snap Wind drag increases gravitational load on a tree by generating a moment (Equation 2.5) as the trunk and stems bend (Grace 1977; Spatz and Bruechert 2000). 2 2 2l f M (2.5) In Equation 2.5, M2 is the gravitational moment; f2 is as before; and l2 is the magnitude of deflection. The magnitude of deflection is described by engineering beam theory and depends not only on the drag force exerte d on the crown, but on the geometry and material properties of the stem as given in Equations 2.6 and 2.7 (Grace 1977; Gordon 1981). EI I D l 3 ) (3 2 (2.6) In Equation 2.6, l2 and D are as before, I is the second moment of area of a cross section, and E is the modulus of el asticity (Youngs modulus). 644d I (2.7) In Equation 2.7, I is as before, d is the diameter of the stem, and is the area of the unit circle. It should be noted th at there are many equations for I and the one chosen depends on the specific cross sectional ar ea Equation 2.7 is specific fo r a circular cross section. Introductory information about mechanics of materials and beam theory is provided by Gordon (1981) and Salvadori (1980). Trunk taper accounts for the change in I over the length of the stem and influences stress distribution within the stem. Petty a nd Swain (1985) have sugge sted that taper is

PAGE 25

13 probably the most important fact or affecting susceptibility to stem breakage. Niklas and Spatz (2000) found that because of changes in taper and canopy shape, safety factors against wind induced stress likely decrease as trees mature. They also submit that wind induced stresses and related safety factors are not uniform throughout the canopy of a tree (Niklas 2000a). Niklas and Spatz (2000) f ound that material yiel d stresses decreased with increasing stem diameter for all trees examined, while stress levels were lowest at the most distal and basal portions of the tree fo r all but the oldest tree. For the oldest tree, stress levels were highest at the trunk base. Leiser a nd Kemper (1973) had earlier demonstrated that assuming homogeneity of ma terial properties, taper determines the location of maximal stress in a sapling tr ee trunk. Severely ta pered trunks had the greatest maximal stresses of all trunks analy zed with those stresses located high on the trunk near the point of loading. Untapere d trunks also had large maximum stresses but they were located at the trunk base. Both of these findings support ecological adaptations that are proposed in the literature. Distribu tion of maximal bending stresses to the distal portions of the tree may allow for preferential branch shedding (a form of self pruning) to reduce drag in high winds (Hedden, Freder icksen et al. 1995; Mattheck and Kubler 1997; Niklas 2000a; Beismann, Schweingruber et al. 2002). It may also be an evolutionary strategy facilita ting asexual reproduction (Bl ackburn, Petty et al. 1988). From a purely mechanical perspective, th e best designs distribute stress uniformly along an entire structure. W ith that in mind, Mattheck a nd Kubler (1997) presented the axiom of uniform stress based on their wo rk using the Finite Element Method to analyze notch stresses and a llometry of trunks, branches, a nd unions. Uniform stress was originally proposed by Metzger, who discove red that taper of spruce trees (height ~

PAGE 26

14 diameter3) assured a uniform distri bution of bending stresses along the trunk. Other analyses of shapes of stems (Morgan and Cannell 1994) and scaling factors of trees (McMahon 1973) provide further evidence of th e axiom of uniform stress. However, in all these studies, wood mechanical propertie s were assumed to be uniform or to change negligibly throughout the specimen. The a ssumption of uniform stress has been made repeatedly (Fraser 1962; Petty and Worrell 198 1; Milne 1991; Pelto la, Nykanen et al. 1997; Saunderson, England et al. 1999) and a ppears to be valid at least in young trees (Leiser and Kemper 1973; Petty and Worrell 198 1). Still, Wood (1995) admits there are questions as to the accuracy of the assumption of uniform stress. And those who include measures of wood anatomical features along with taper in their analysis of stress distribution report variations in the longitudinal stress di stribution within the stem (Ezquerra and Gil 2001). From Equation 2.6 it is clear that the modulus of elas ticity or Youngs modulus is another important factor determining bending stresses in stems. Youngs modulus is a measure of the stiffness of a materi al (Salvadori 1980; Gordon 1981). Tabulated values of average Youngs moduli for bot h kiln dried and gr een wood are readily available on a number of commercially importa nt species (Green, Wi nandy et al. 1999). It should be noted however, th at tabulated values are averages taken from small clear specimens of wood (Green, Winandy et al. 1999; Bailey 2000a) and it has been shown that there is tremendous vari ability in the Youngs modulus within a tree (Niklas 1997a; b; Niklas and Spatz 2000). Cannell and Morgan (1987) re ported that Youngs moduli measured in intact branches and trunks ar e lower than tabulated green wood values and suggested that a structural modul us of elasticity (the modulus of the intact living stem) be

PAGE 27

15 used to more accurately represent material pr operties of live wood. Accordingly, efforts have been made to measure the modulus of elasticity for living trees but with the assumption that it is uniform throughout the trunk (Milne 1991; Pelto la, Kellomaki et al. 2000). Foresters have long known that material properties of wood are not isotropic (Green, Winandy et al. 1999).Wood scientists have shown that 86% of the variability in the longitudinal Youngs modulus is account ed for by orientation of cellulose microfibrils (microfibril angle (MFA)) in cell walls (Evans and Ilic 2001). Barnett and Bonham (2004) recently reviewed the literature on MFA. High MFAs are associated with low Youngs moduli and are common in juvenile wood and compression wood while low MFAs convey a high Youngs modulus and are associated with adult wood and tension wood. Factors th at control the orientation of cellulose microfibrils in cell walls are still unknown (Baskin 2001). Work on MFA refutes the assumption of material homogeneity of stems but does not dispr ove the possibility of uniform stress. Wind throw Sommerville (1979) suggested that widespread wind snap poses a greater economic threat to forest systems than wind throw. However, widespread wind throw is more prevalent (Blackburn, Petty et al. 1988; Pa pesch, Moore et al. 1997; Moore 2000) and as such it has received far more attention in the lit erature. In an effort to elucidate factors involved in wind throw, Fraser (1962) introduced a pull test using a winch and cable and demonstrated that Fomes annosus root rot caused lower resistance to lodging while improved drainage had an opposite affect in Douglas fir ( Pseudotsuga menziesii ). Others used similar pull tests to demonstrate that l odging is dependent on taper (Petty and Swain 1985), rooting depth and root morphology (So mmerville 1979), soil type and moisture

PAGE 28

16 content (Moore 2000), and species (Peltola, Kellomaki et al. 2000; Moore and Gardiner 2001). Coutts (1986) used tree pulling tests to demonstrate that applied wind loads were much more significant in causing wind throw than associated gravitational loads a finding that was later confirmed by Papesch et al. (1997). Coutts also annotated the sequence of wind throw in shallow rooted Sitka Spruce (Coutts 1983) and defined components of root anchorage (Coutts 1986). Co llectively, pull tests have been used to determine maximum bending moments of trees in order to make comparisons with maximum applied loads. Those comparisons are then used to create wind throw prediction models (Peltola, Nykanen et al. 1997; Kerzenmacher and Gardiner 1998; Saunderson, England et al. 1999; Moore and Quine 2000; Talkka ri, Peltola et al. 2000) and to develop improved cultural practices (Cremer, Borough et al. 1982; Brchert, Becker et al. 2000; Gardiner and Quine 2000; Mitchell 2000). However, pull tests have mainly been used to evaluate static load s (Fraser 1962; Blackburn, Petty et al. 1988) and Oliver and Mayhead (1974) revealed that wi nd damage occurs at mean wind speeds well below those predicted by tree pulling tests. Static test s are a crude approximation of actual loads experienced by trees in wind events. Tree Dynamics Wind is a dynamic force that generates complex loads in stems as trees sway. Moore and Maguire (2004) recen tly reviewed the literature investigating tree dynamics. Most of the research has been conducted in forest systems where trees are in close proximity and exhibit strongly excurrent growth habits. Milne (1991) used pull tests to study natural frequency and damping of Sitka spruce ( Picea sitchensis ) in a forest setting. He reported that the greatest component damping a trees sw ay was interaction between crowns and branches of neighboring trees. Moore and Maguire (2005) followed up on

PAGE 29

17 Milnes work and found that interactions betw een trees were insignifi cant. They reported that aerodynamic damping of foliage had the greatest effect on tree sway. They also reported that it is inappropriate to represen t tree crowns as a series of lumped masses when calculating wind and gravitational load s, something that Mayhead had discovered earlier but never published (Moore and Magui re 2005). Still, Ma yhead (1973), Rudnicki et al. (2004), and Vollsinger et al. (2005) have all shown strong relationships between canopy mass and drag. A model describing a tr ee crown as a series of independent mass dampers was presented by Kerzenmacher a nd Gardiner (1998) and again by James (2003). Mass damping seems very likely and res earch is underway to evaluate the effect of branches as coupled mass damper s (James 2003; Moore and Maguire 2005). As a final note on tree dynamics and wind loading in general, Denny (1994) warned that the probability of wind damage calculated from extreme measurements is much greater than that calculated from an average as is typically done. Therefore, if models that are generated to pr edict catastrophic failure in trees are to be used in urban settings, consideration needs to be given to the severity of a singl e catastrophic failure. Since trees do not construct to man-made fact ors of safety, models should compensate to some degree. Pruning Mechanical pruning is co mmonly thought to reduce wind damage from strong winds because it reduces the surface area of the tree canopy (Mattheck and Breloer 1994). Duryea et al. (1996) noted that pr uned trees withstood wind damage from Hurricane Andrew better than their unpruned counterparts Ham and Rowe (1990) felt that despite losing over 4800 street trees, da mage to the city of Charlotte, NC from Hurricane Hugo was lessened by their progra m of routine [tree] maintenance.

PAGE 30

18 However, a survey of street trees in Roches ter, NY showed no difference in the frequency of storm damaged trees in pruned versus unpruned areas of the city (L uley, Sisinni et al. 2002). Pruning is one of the most prominen t tree maintenance pr actices (Accredited Standards Committee A300 2001), and it is wide ly recommended as a means of reducing wind damage to trees (Matheny and Clar k 1994; Brown 1995; Gilman 2002; Harris, Clark et al. 2004). Never th e less, research supporting recommendations for pruning trees to reduce wind damage is almost nonexistent in the pr imary literature. Moore and Maguire (2002) mention that th e natural frequency of Douglas fir trees does not appear to be significantly affected by pruning until the to pmost third of the canopy is removed. They suggest this is due to the paucity of foliage in the lower canopy and higher wind velocities at higher elevations Rudnicki et al. (2004) and Vollsinger et al. (2005) evaluated whole branch removal in their calcul ation of drag coefficients and reported that pruning did not influence streamlining in c oniferous or hardwood species tested. However, they noted that their test specime ns were small with branches that behaved independently, so effects of streamlining were inconclusive. Even with these recent reports, the primary literature appears void of evidence for pruning as a means to mitigate wind damage. There is even less in formation about how pruning should be accomplished. The American National Standards Institu te (ANSI) A-300 (2001) pruning standard and other pruning references (Lilly, Clark et al. 1993; Brown 1995; Brickell and Joyce 1996; Lang and Editors 1998; Gilman 2002; Harris Clark et al. 2004) list four primary pruning practices: cleaning, thinning, rais ing (also called lifti ng or skirting), and reduction. Other colloquial terms defining specific pruni ng practices include utility,

PAGE 31

19 structural, risk reduction, balancing, vista, restoration, topping, tipping, or lions tailing but each of those are a type or combination of the primary four. There are also specialty pruning practices such as coppicing, pollardi ng, pleaching, topiary, espalier, bonsai, and fruit tree pruning (Accredited Standards Co mmittee A300 2001) which have historically been used for specific effects, but those are not commonly used in landscapes. For definitions of the four primary pruning pract ices refer to the ANSI A-300 (2001) pruning standard section 5.6 or Gilman (2002). Dose Pruning dose is the amount of live tissu e removed at one pruning (Accredited Standards Committee A300 2001). ANSI recomme nds limits to pruning as defined by a percent of the foliage removed [from a tree] within an annual growing season (Gilman 2002; Gilman and Lilly 2002; Harris, Clark et al. 2004). However, ANSI provides no information about how to quantify the percent of foliage removed. Other resources for arborists refer back to the ANSI pruning st andard on questions of dose (Gilman 2002; Gilman and Lilly 2002). They also advise practitioners to quantify pruning dose based on the desired objective (Brown 1995; Harris, Cl ark et al. 2004), or appearance of the tree following the pruning (Waring, Schroeder et al. 1982). Thus measurement of pruning dose by practitioners is largel y qualitative and subjective. The most reliable way to quantify pruning dose is destructive. Following pruning, the foliage and stems removed and the rema ining canopy are cut into small pieces and dried to a constant weight. Pruning dose is th en calculated as a pe rcentage of total dry mass of foliage removed. Alternative, nondestructive methods for determining pruning dose could be considered using biomass pr ediction equations cal culated from sapwood area or from trunk diameter at breast height (DBH). Howeve r, while those alternatives

PAGE 32

20 may prove useful to investigators, they w ould need some modifications before finding their way to the practitioner. Sapwood cross sectional area has been used to generate prediction equations for leaf area based on the pipe model theory (Kaufmann and Troendle 1981; Waring, Schroeder et al. 1982; Meadows and Hodges 2002). Prediction equations are species specific (Rogers and Hinckley 1979), but have been generated for a number of species including Quercus velutina and Q. alba (White 1993), Q. rubra (Meadows and Hodges 2002), Q. falcata var. pagodifolia and Fraxinus pennsylvanica (Kaufmann and Troendle 1981), Picea engelmannii Pinus contorta Abies lasiocarpa and Populus tremuloides (Ohara and Valappil 1995; Meadows and Hodges 2002; Medhurst and Beadle 2002). Leaf area is first correlated to leaf biomass, and then to sa pwood area. Within a species, correlation is independent of age, site, stra ta, crown class, crown condition, and stand density (Rogers and Hinckley 1979) However, there is debate as to whether leaf area should be correlated to current sapwood area (Meadows and Hodges 2002) or total sapwood area (McDowell, Barnard et al. 2002) There are also co mplications to the models because leaf area: sa pwood area generally decreases as total height of the tree increases (Meadows and Hodges 2002) and leaf area: unit weight is larger in the lower, more shaded, parts of the canopy than in th e intermediate and upper parts (Ter-Mikaelian and Korzukhin 1997). Prediction equations need to be determined for more species before this method becomes useful to a la rge degree. Predicti on equations using DBH are more commonly used. There is a sizeable amount of information relating vari ous biomass components to DBH. Ter-Mikaelian and Korzukhin (1997) provid e an extensive review of the literature

PAGE 33

21 for 65 North American tree spec ies. Prediction equations re viewed are all of the same form but there are often multiple prediction equations for the same biomass component. As a result, the authors present all the differe nt prediction equations together, as well as various components used to generate these equa tions such as the range, sample size, and standard error of the estimate; as well as ge ographic region of the sample, and a reference to the paper where it was cited. The additi onal information is included as an aid in determining which prediction equation is most appropriate for individual circumstances. There are a considerable number of non-conife rous species listed, but species included are limited to those important in forest systems. Additional work is needed to generate similar equations for commonly used landscap e species. Another im passe to using the biomass equations based on DBH is that th ey predict the oven-dry weight of the individual component. This is fine for a researcher in pred icting pruning dose but completely impractical for a practitioner. Conclusions This review presents a summary of work published in the primary literature associated with wind damage and pruning of amenity trees. Tremendous strides are being made towards the understanding of wi nd forces on trees and trees associated mechanical response. Additional information is needed in many areas including: wind profiles and wind spectra for both forests and individual trees, dr ag coefficients of large, open grown, broad-leaved species, root and so il interactions affecting anchorage, and wood material properties. Ther e is especially a de arth of practical in formation that can be used by tree care providers in urban se ttings. The subseque nt study is aimed at providing useful information regarding pruni ng as a means for reducing wind damage in amenity trees.

PAGE 34

22 U (d1) l2l1 g m (l1) Figure 2.1. Wind forces affecting trees (modified from Grace (1977)). Wind force: ) ( ) ( ) ( 2 1 ) (1 1 2 1 1l A l U l C fd air Wind moment: 1 1 1) ( l f M Gravitational force: g l m f ) (1 2 Gravitational moment: 2 2 2l f M Magnitude of deflection: EI I f l 3 ) (3 1 2 Second moment of area: 644d I Constants are as follows: air = density of air, Cd =drag coefficient, l1 = elevation to the point of loading, U = wind speed, A = projected surface area of the canopy, m =mass of the canopy at l1 g = acceleration of gravity, l2 magnitude of deflection, E = modulus of elasticity, d = diameter of cross section, and = area of the unit circle.

PAGE 35

23 CHAPTER 3 MATERIALS AND METHODS Tree Selection Clonally propagated trees were chosen as subjects for testing to improve similarity among trees assigned to different experime ntal treatments. Live oak trees ( Quercus virginiana QVTIA PP 11219, Highrise) were us ed because they are a commonly planted species in the southeastern United Stat es and were readily available from a local nursery. Trees were selected for physical si milarity from a population of Highrise live oaks grown at Marshall Tree Farm (MTF; Mo rriston, FL, U.S.A.). Marshall Tree Farm transplanted the trees from five gallon contai ners into native soils (Orlando fine sand or Sparr fine sand) in June 2000. In Ma y 2003, clear trunk, total canopy height and diameter, trunk taper parameter, rootball diamet er, height to vertical center of a canopy, projected area of a canopy a nd canopy volume were measured on 50 trees and used as characteristics for selection (Table 3-1). English units were chosen for use throughout this thesis in order to follow American e ngineering convention. Clear trunk was the distance between the top-most root and firs t main branch. A minimum clear trunk of 54 in. was required for selection. Total ca nopy height (Tch) was determined by measuring maximum (Mxch) height (distance from t op of the rootball to top of the canopy), minimum (Mnch) height (distance from top of the rootball to origin of the lowest branch), and subtracting Mnch from Mxch Canopy diameter (Cd) was the average width of a canopy measured at its widest point in two perpendicular di rections. A visual estimate of the canopy outline was used to determine Cd single shoots extending

PAGE 36

24 beyond the canopy outline were neglected. Tr unk taper parameter (Taper) was calculated as -(R-r)/R (Leiser and Kemper 1973) where R = radius at 6 in. above the top-most root, and r = radius at 54 in. above the top-most root. Rootball diameter (Rd) was average width measured at the soil surf ace in two perpendicular dir ections. Height to vertical center of a canopy (Hvcc) was one half Tch plus Mnch. Projected area (Area) was calculated as 0.5 the vertical surface area of a cone (whos e dimensions were: height = 0.667 Tch and radius = 0.5 Cd) plus 0. 25 the surface area of an ellipsoid (whose dimensions were: radius of the long axis = 0.333 Ch and radius of the short axis = 0.5 Cd). Canopy volume (Volume) was calculated as the volume of a cone (calculated as before) plus one half the volume of an ellipsoid (calculated as before). Measurement of any characteristic that was greater than thre e standard deviations from its sample mean was cause for rejection. This eliminated obvi ous outliers according to the Empirical Rule in statistics (Ott and Longnecker 2001). On November 6, 2003 forty-four trees were moved from MTF to the Environmental Horticulture Teaching Unit (Tree Unit, Gainesville, Florida, U.S.A.). Trees met Roots Plus Growers standards (R oots Plus Growers Association of Florida 2005); meaning they were dug and completely hardened-off at the nursery with visible roots on the outside of the root ball prior to shipment. Upon arrival, trees were weighed (rootball included) using a dynamometer (Model WT-1-1000H John Chatillon & Sons, New York, NY) then healed into pre-dug holes. Holes were the same dimensions as the root balls and trees were h ealed in without removing burla p, wire, or nylon bag that secured the rootball. Trees we re irrigated three times per day with approximately four gallons of water per irrigation. After hea ling in, tree movement wa s minimized prior to

PAGE 37

25 testing. None of the trees suffered signifi cant defoliation resulting from transport to the Tree Unit. Experimental Design Three effects were evaluated: 1) pruning type, 2) pruning dose, and 3) wind speed. From those three effects, 100 treatment comb inations were constructed. Trees were randomly assigned to a pruning type following a completely randomized design. Five pruning types were included: 1) lions tailing, 2) raising, 3) reducti on, 4) structural, and 5) thinning. Each tree, within a type, wa s pruned three times to approximate targeted, orthogonally spaced, pruning dose levels. Targeted pruning dose levels were: 1) 15% foliage removed, 2) 30% foliage removed, and 3) 45% foliage removed. Within a pruning dose, each tree was subjected to a sequence of wind speeds. Four, equally spaced wind speed levels were targeted: 1) 15 mph, 2) 30 mph, 3) 45 mph, and 4) 60 mph. Data was collected on every tree, before it was pruned and at all targeted pruning dose levels, at ambient wind speeds and at all targeted wind speeds. Therefore, data was collected 20 different times on the same tree w ithin a pruning type. The first four lions tailed, raised, reduced, and thinned trees we re blocked in time by tree within type forming a complete block design. The last three lions tailed, raised, and reduced trees and all structurally pruned trees we re added in no particular order. Pruning Type Pruning types are defined in the Americ an National Standard for Tree Care Operations (ANSI 2001) and in An Illustrated Guide to Pruning Second Edition (Gilman 2002). Five types evaluated here are common in practice or are recommended as a means to reduce wind damage to trees (Matheny and Clark 1994; Brown 1995; Gilman 2002; Harris, Clark et al. 2004). One person was chosen to execute all pruning to

PAGE 38

26 maintain treatment consistency. Prior to testing, trees similar to those selected for experimentation were pruned by a team of individuals and pruning types and doses discussed. During experiment ation, parts of a canopy remove d by pruning were collected and stored in paper bags for gravimet ric analysis of actual pruning dose. Lions Tailing Four trees were lions tailed as part of the complete block design. Pruning consisted of removing primary and higher orde r branches smaller than 0.5 inches in diameter at their point of a ttachment to the trunk or parent stem. Branch diameter was determined using a digital caliper as aver age width measured in two perpendicular directions. Pruning dose levels were determined in the field from geometric dimensions of the canopy; all tissue rem oved was dried and weighed to calculate actual dose. The 15% pruning dose consisted of pruning within the lowest 15% of canopy height and the most interior 15% of canopy radius. The 30% and 45% pruning doses were applied in similar fashion. Canopy height and diameter measurements were taken on the day of testing following the procedure used for tree selection. Pr uned volume was determined as follows. The canopys main leader was ma rked at 15%, 30%, and 45% of total canopy height calculated as: canopy he ight (0.15, 0.30, and 0.45 re spectively) + min. height. Fifteen, thirty, and forty-five percent of canopy radius was calculated by multiplying 0.5 canopy diameter by 0.15, 0.30 and 0.45 resp ectively. During pruning, canopy radius was measured with a common tape measure. Foliage removed from the first four lion s tailed trees was inadequate at all dose levels. Therefore, three addi tional trees were included to better approximate targeted pruning dose levels. The three additional lions tailed trees were blocked in time with each other but not with other pruning type s. Pruning, canopy appearance, and canopy

PAGE 39

27 structure of the three additional lions tailed trees was similar to the first four, but pruning dose in the field was determined as a visual estimate of live foliage removed. All tissue removed was dried and weighed to calculate actu al dose. Examples of lions tailing from an urban landscape and pruning of test trees at each dose ar e provided in Figures 3.1 to 3.3. Seven lions tailed trees were included in the statistical analysis. Raising The first four raised trees were blocked in time with other pruning types. The first raised tree was the first of all trees tested. It was not included in the analysis because experimental procedure changed for all subseque nt trees. Pruning dose levels for the first four raised trees were determined in the fiel d from geometric dimensions of the canopy as follows. On the day of testing, canopy hei ght was measured and the main leader was marked at 15%, 30%, and 45% of total canopy height following the procedure used for lions tailing. The 15% pruning dose was appl ied by pruning from the base of the canopy up the main leader to the 15% mark. The 30% and 45% dose levels were applied in similar fashion. All tissue removed was drie d and weighed to calculate actual dose. Pruning was conducted by removing all branches from the main leader at their point of attachment. Branches originating high in the canopy but drooping be low the elevation of the mark on the main leader were also re moved to that elevation so the entire canopy width was lifted. Foliage removed at each pruning dose for the first four raised trees exceeded targeted levels so three additional trees were included to better approximate targeted pruning doses. The three additional raised trees were bloc ked in time with each other but not with other pruning types. Pruning dose for the addi tional raised trees was determined in the field as a visual estimate of live foliage removed; all tissue removed was dried and

PAGE 40

28 weighed to calculate actual dose. Pruning wa s carried out as before, but if removal of a large limb caused an excessive dose, it was trea ted as a second leader and raised as per the main leader. Examples of raising from an urban landscape and pruning of test trees are provided (Figures 3.4 to 3.6). Six of the seven raised trees were included in the statistical analysis. Reduction Reduction pruning involved making headi ng cuts (shearing) to reduce the geometric size of a canopy. Pruning did not involve drop-crotch pruning or using reduction cuts as is commonly recommended for reducing the size of a tree or part of a tree in a landscape. The first four reduced trees were blocked in time with other pruning types. Pruning dose for the first four reduced trees was dete rmined in the field by geometric dimensions of a canopy. On the day of testing, ca nopy height and average canopy diameter were determined as before. The main leader was marked at 85%, 70%, and 55% of total canopy height calculated as: max height canopy height (0.85, 0.70, and 0.55 respectively). Pruning was accomplished by fi rst removing the main leader at the designated mark followed by pruning the exteri or of the remaining canopy to re-establish each trees original three dimensional shape, but in a smaller version. No foliage was removed from interior parts of a canopy. A ll tissue removed was dried and weighed to calculate actual dose. Foliage removed from all but the first dimensionally reduced tree exceeded targeted pruning doses at all levels Therefore, three additional trees were reduced to better approximate targeted dose levels. The three additional reduced trees were not blocked in time with each other or with other pruning types. Pruning was carried ou t in as per other reduced trees but dose was

PAGE 41

29 determined in the field as a visual estimate of live foliage removed trunks were not marked prior to pruning. All tissue remove d was dried and weighed to calculate actual dose. Examples of reduction from an urban landscape and pruning of test trees at each dose are provided (Figures 3.7 to 3.9). Seven reduced trees were included in the statistical analysis. Structural Three trees were structurally pruned. St ructurally pruned trees were blocked in time with each other but were not blocked w ith other pruning types. They were all evaluated at the end of the data collection period. Stru ctural pruning involved making reduction and removal cuts to shorten and slow growth of stems competing with the main trunk, and to develop scaffold branches. Li ttle thought was given to canopy size, shape, or density. Pruning dose was determined in th e field as a visual estimate of live foliage removed; all tissue removed was dried and wei ghed to calculate actual dose. Examples of structural pruning from an urban landscap e and pruning of test trees at each dose are provided (Figures 3.10 and 3.11). Structurally pruned tree number 2 was left out of the statistical analysis because of an inadverten t change in procedure. Two structurally pruned trees were included in the statistical analysis. Thinning Four trees were thinned and a ll four were blocked in time with other pruning types. Thinning was conducted by maki ng reduction and removal cuts throughout the entire canopy, especially at the outer edge of a canopy. Pruning dose was determined in the field as a visual estimate of live foliage removed. All tissue removed was dried and weighed to calculate actual dose. Thi nning produced a uniformly dense canopy without changing the canopys initial geometric dimensi ons. Examples of thinning from an urban

PAGE 42

30 landscape and pruning of test trees at each dos e are provided (Figures 3.12 and 3.13). Four thinned trees were included in the statistical analysis. Pruning Dose Pruning dose is defined in Sections 5. 5.3 and 5.5.4 of A300 (ANSI 2001) as a percentage of foliage removed. Pruning dos e used in the statistical analyses was percentage of foliage dry weight removed. All parts of a canopy collected and stored during pruning were dried at 70C until they r eached a constant weight. Foliage was then separated from stems and both foliage and stem weight were recorded separately for each treatment combination. The remaining ca nopy (clear trunk excluded) of 13 of the 27 trees tested was also cut into small sections, dried, and measured as per pruned cuttings to calculate an average canopy dry we ight. Pruning dose was calcul ated as dry weight of an individual pruning dose (summed incrementally) average canopy dry weight 100. Pruning dose was also calculated from actual tree canopy dry weight for the 13 trees used to calculate the average canopy dry weight. Wind Speed High winds are not regularly experienced at the University of Florida Tree Unit. Winds were generated using an airboat (Fig ure 3.14) driven by a 1988 Chevy 350 engine, a 2-1-power reduction unit, and a 2-blade, Sensenich wide blade, 78-inch, left hand rotation, composite propeller (Sensenich Wood Propeller Co., Inc., Pl ant City, FL, USA). Airboat rudders were locked in an orientat ion perpendicular to the long axis of the propeller and the boat was set on two concrete pi ers. Piers were engineered to elevate the propellers midpoint to the estimated cente r of pressure on an average unpruned, undisturbed canopy. This corresponded to one third average total canopy height, or an elevation of 10 ft. 2 in. The ai rboat was lifted into place with a crane and was fixed to the

PAGE 43

31 piers with 2 in. angle iron that ran across the hull and 0.75 in. diameter threaded rod secured into the concrete with epoxy. A downwind profile of generated wind speed was used to determine the location for placement of trees during testing. Gill propeller anemometers (Model 27106R R.M. Young Company, Traverse City, MI, USA) a nd a proprietary data acquisition program were used to measure and record wind sp eeds. Orthogonal anemometers were mounted on a steel tower 16 ft. 5 in. and 33 ft. off the ground as well as 10 ft. 1 in. off the ground on an outrigger. The outrigger was locat ed seven feet upwind of the tower. On November 21, 2003 the first set of dow nwind profile tests were conducted with the outrigger located at distan ces of 18 ft. 5 in., 29 ft. 3 in., and 39 ft. 8 in. downwind from the airboat propeller. Wind speeds were recorded at 100 Hz for one minute at motor rpm starting at 1000 rpm and increasi ng to 4500 rpm in increments of 500 rpm. Data collected is summarized in Table 3-2. Before additional test ing, the airboat stern was elevated eight inches. The airboat propelle r was then centered at an elevation equal to 10 ft. 10 in. and the airboat hull sat at a five degree angle from horizontal. Tower anemometers were abandoned in subsequent testing. Wind speeds were measured on December 3, 2003 to correlate wind speed to motor rpm (Table 3-3). Anemometers were loca ted 17 ft. downwind from the propeller and wind speeds were collected at 100 Hz for approximately four minutes at motor rpm starting at 1000 rpm and increasing to 4500 rp m in increments of 500 rpm. Data is summarized in Table 3-3. Two 3-cup anemometers with directi onal sensors (Met One 034B, Campbell Scientific, Inc., Logan, UT, USA) meas ured wind speeds during testing. These

PAGE 44

32 anemometers were mounted in split ring hange rs welded to telesc oping steel conduit so they could be elevated to vertical center of the tree canopy (calcu lated as: canopy height 2 + min. height). One anemometer was locat ed approximately 7 ft. upwind and the other approximately 14 ft. downwind of each tree dur ing testing. Directional sensors were unstable in generated winds so vanes were removed and wind direction was not recorded during testing. On April 19, 2004 wind speed was correlated to motor rpm using Campbell anemometers (Figure 3.15). Anemometers were located at 10 ft. 2 in. and 23 ft. 3 in. downwind from the airboat propeller and at an el evation of 9 ft. 6 in. with respect to the height of the propeller (thi s was equivalent to the height of Young anemometers when they were 17 ft. downwind of the propeller). Data was collected at 0.5 Hz for three minutes at each rpm starting at 1000 rpm and running up to 4000 rpm in increments of 500 rpm. Campbell anemometers recorded hi gher wind speeds than Young anemometers at the higher rpm (Figure 3.15). Because there were discrepancies be tween wind speeds recorded by Young and Campbell anemometers, a road test was conduc ted to check the accuracy of the Campbell anemometers. On April 21, 2004 anemomet ers were held out the front passenger window of a Jeep Grand Cherokee as it was dr iven from 10 mph to 70 mph in increments of 10 mph. Wind speeds recorded at 0.5 Hz for two minutes at each velocity are summarized in Table 3-4. Wind speeds recorded by the West anemometer (its location relative to the tree during testing) were nearly identical to the speedometer reading. The East anemometer was more variable. It was uncertain if greater error was in the

PAGE 45

33 anemometer or the drivers ability to maintain a constant velocity. Measurements from the East anemometer were not used in the analysis. The East anemometer was included as an aside. Coder s uggested that canopy density can be determined by measuring wi nd speeds on windward and leeward sides of the canopy (Coder 2000). Wind speeds recorded by the East anemometer were erratic despite pruning dose so it was not included in the analysis. Trunk Deflection Tree response to wind loading was measur ed as trunk deflection below the canopy. Before testing, a trunk was marked at 18, 30, 42, and 54 in. above the first root and eight cable extension transducers (CET) (Celesco Transducer Products Inc., Chatsworth, CA, USA) were attached to the trunk in pairs at th ose elevations (Figures 3.14 and 3.16). The four lowest CETs were fixed with 10 in cables (PT1A-10-UP-5K-M6-SG), followed a by a pair of 15 in. CETs (PT1A-15-UP-5K-M6-S G), and finally a pair of 25 in. CETs (PT1A-25-UP-5K-M6-SG). The 10 in. CETs were calibrated using a common 12 in. ruler, while 15 in. and 25 in. CETs were calib rated using a 10 ft. tape measure. Linear regression produced R2 values of 1.0 for all CETs (Figures 3.17 and 3.18). Transducers were bolted to two pieces of 3 in. 2 in. angle iron such that each angle iron had two 10 in., one 15 in., and one 25 in. transducer spaced 1 ft. apart along its length. During testing, angle irons were cl amped to 4 in. 4 in. wooden posts located approximately 16 ft. from the tree and spaced approximately 23 ft. 5 in. apart (Figure 319). Elevation of an angle iron on a post wa s determined on day of testing using line levels that hung from nylon string stretche d between lowest transducers and the 18 in. mark on a trunk. Piano wire (0.01 in. diamet er) was used to create extensions between transducer cables and the trees. Piano wire ex tensions were attached to trees with plastic

PAGE 46

34 cable ties. Deflection of the trunk was calculated at each inte rval as described in Figure 3.20. Longitudinal Youngs modulus is the most significant factor affecting bending (Barnett and Bonham 2004) and it is known to vary among speci es and within species. Every effort was made to select and main tain trees so there would be a uniform longitudinal Youngs modulus among trees te sted. Genetic variation among trees was limited by using clonally propagated trees. Tr ees were young enough (5-6 years old) that they were likely composed entirely of j uvenile wood with no heartwood. Because they were nursery grown, they likely had wide, uni form growth rings and little to no reaction wood. Wood in all trees likely had the sa me moisture content because trees were irrigated regularly up until the day they were tested. Still, because variation in the longitudinal Youngs modulus among trees would confound results based on trunk movement, efforts were made to determine green and kiln dried values of longitudinal Youngs modulus. Testing was conducted on coupons (1 in. 1 in. 18 in. sections) cut from trunks at different elevations, and on whole trunks, following standard test methods for small clear specimens of timber (Bailey 2000 a) and static tests of wood poles (Bailey 2000b) respectively. Figure 3. 21 shows testing apparatus us ed in both procedures. Youngs modulus calculations were conducted by an undergraduate as an University Scholars project and more pr ocedural detail is given in that report (Trachet 2005). Experimental Procedure Testing began May 19, 2004. One tree was tested per day. Testing dates are included in Table 3-1. Trees remained irrigated and undisturbed until the day they were tested. One exception was structurally pr uned tree number two, which was excluded from the analysis because it was moved a day early, and withou t irrigation, it was

PAGE 47

35 severely water stressed before testing was complete. On a day of testing, each tree was moved from the field to the testing site and its rootball and the base of its trunk were secured so the tree would remain upright. Figure 3.22 depicts the apparatus used to secure a rootball and the base of a trunk. Eight cable exte nsion transducers (CETs) were connected to the trunk and the tree was tested at all wind speeds before any pruning was executed. Transducers were then disconnect ed and the tree was pruned to the lowest targeted pruning dose (15% foliage removed) After pruning, CETs were reconnected and the process was repeated. Each tree was tested four times once before pruning and once at each of the three targeted pruning doses. Motor rpm to wind speed correlations i ndicated that to achieve desired wind speeds, testing should proceed from ambien t to 1500, to 3000, and finally 4500 rpm. However, when testing the first raised tree, significant defoliation occurred at and above 3000 rpm. Thus the protocol was changed to proceed from ambient to 1250, to 2000, to 2750, back to 1250 rpm, and finally at ambient once more. Data was collected for two minutes at ambient conditions and for four minut es at individual rpm. Changes in wind speed occurred consecutively but data co llection was interrupted as rpm changed. Measurements from CETs, anemometers, and from a thermistor temperature probe (Model 107 Campbell Scientific, Inc.) were taken at 0.5 Hertz. The temperature probe was suspended 1.5 ft above ground on site and was protected from direct sunlight. The data acquisition system (DAQ) consisted of a Campbell Scientific CR10X datalogger used in combination with a Campbell AM 416 multiplexer (Figure 3.23) and a program written in Loggernet 2.1 (Campbell Scientif ic, Inc.). The DAQ system was powered from a standard 120 VAC socket through a 12 VDC converter.

PAGE 48

36 Statistical Analysis Pruning types were compared using an alysis of variance (ANOVA) followed by least squares means separations adjusted using Tukeys method. Pruning dose and wind speed were analyzed in similar fashion and by regressions using a complete two factor quadratic empirical model. Regressions on wind speed and pruning dose were used to develop response surfaces of trunk movement from which orthogonal levels of each parameter were extracted. Extracted va lues were used in a three way ANOVA to compare effects of pruning type, pruning dos e, and wind speed on tree movement. The results and statistical analysis are covered in detail in chapter four. Data was analyzed using the SAS system for windows release 8.02 ( 1999-2001 SAS Institute Inc., Cary, NC, USA).

PAGE 49

37Table 3-1. Measurements used to dete rmine physical similarity of trees tested. Date tested Pruning typea Treeb Mxchc (ft.) Mnchd (ft.) Tche (ft.) Cdf (ft.) Taperg Rdh (in.) Twi (lbs) Hvccj (ft.) Areak (ft.2) Volumel (ft.3) 19May04 RA01 4 20.005.0414.966.65-0.04 31.0066018.28 312.101038.21 24May04 LT01 3 19.304.8814.436.30-0.05 33.5068017.58 283.90898.58 25May04 TH01 36 19.204.5414.665.30-0.03 29.7553017.31 233.19646.24 26May04 RE01 28 22.105.2916.815.70-0.03 29.2561019.66 284.36857.10 28May04 RA02 40 20.505.2715.236.15-0.04 32.7568018.30 287.40904.03 01Jun04 LT02 24 20.404.6515.756.00-0.04 31.0070018.75 286.56890.13 02Jun04 TH02 39 17.404.7712.636.25-0.04 30.5068015.75 254.80774.27 03Jun04 RE02 30 21.404.7116.696.30-0.02 32.2561519.84 318.221039.77 04Jun04 RA03 27 19.204.7514.455.80-0.04 30.5058517.35 256.84762.92 07Jun04 LT03 26 20.704.9215.786.10-0.04 30.7562518.83 292.73921.76 08Jun04 TH03 5 19.704.5615.145.95-0.02 31.5071018.11 274.80841.10 09Jun04 RE03 32 20.105.0815.025.55-0.03 29.7563017.79 251.19725.97 10Jun04 RA04 41 19.005.0413.966.15-0.04 30.7567017.03 268.75828.59 18Jun04 LT04 49 20.505.1915.316.25-0.04 30.7564518.44 294.36938.78 21Jun04 TH04 1 20.604.5816.025.85-0.04 32.7567518.94 281.77860.28 23Jun04 RE04 31 21.005.0016.005.60-0.03 31.0059518.80 267.38787.51 28Jun04 RA05 42 19.904.6015.306.20-0.01 30.7569018.40 291.24922.81 30Jun04 RA06 43 18.204.9213.286.20-0.04 30.7563016.38 261.63801.40 01Jul04 RA07 37 20.804.7116.096.30-0.04 28.7556519.24 309.051002.40 31Aug04 RE05 22 19.805.1314. 686.40-0.03 30.7558517.88 293.30943.40 02Sep04 RE06 19 21.304.9616. 346.45-0.05 31.0064019.57 321.871067.02 26Oct04 LT05 44 20.804.5216. 286.20-0.03 31.2561719.38 305.98982.14 30Oct04 LT06 21 20.905.0015. 905.90-0.02 27.5050518.85 282.95868.68 03Nov04 LT07 45 21.404.6916.716.05-0.03 30.5062219.74 303.56960.09

PAGE 50

38Table 3-1. Continued. Date tested Pruning typea Treeb Mxchc (ft.) Mnchd (ft.) Tche (ft.) Cdf (ft.) Taperg Rdh (in.) Twi (lbs) Hvccj (ft.) Areak (ft.2) Volumel (ft.3) 04Nov04 ST01 29 21.104.9016.205.70-0.03 30.0061019.05 275.87826.29 10Nov04 ST02 11 19.804.6715.136.05-0.03 31.7565018.16 280.34869.37 11Nov04 ST03 9 18.904.7114.196.15-0.03 29.5058017.27 272.15842.44 23Nov04 RE07 38 19.504.7114.796.20-0.02 30.2560017.89 283.75892.40 mean =20.134.8515.286.06-0.03 30.7362818.31 283.22881.92 Std =1.0510.2261. 030.300.01 1.2449.91.02 20.8997.58 aPruning type: Pruning treatment assigned (type and number within type). Pruning types are: LT= lions tailing, RA = raising, R E = reduction, ST = structural, TH = thinning. b Tree: Number of original 50 selected from the field. c Mxch: Maximum canopy height = distance from the top of the rootball to the top of the canopy. d Mnch: Minimum canopy height = distance from the top of the rootball to the origin of the lowest branch. e Tch: Total canopy height. = Mxch Mnch. f Cd: Canopy diameter = widest point of the canopy measured in two perpendicular directions and averaged. The main outline of t he canopy was used to determine diameter single shoots extending beyond the main body of the canopy were neglected g Taper: Trunk taper parameter = -(R-r)/R where R = radius at 6 in above the top-most root, and r = radius at 54 in. above the top-most root. h Rd: Rootball diameter = average width of the root ball measured at the surface in two perpendicular directions. i Tw: Total weight of the tree and its rootball. j Hvcc: Height to vertical center of the canopy = ((0.5 Ch) + Mnch) k Area: Canopy projected surface = 0.5 the ve rtical surface area of a cone (whose dime nsions are height = 0.667 Ch and radiu s = 0.5 Cd) + 0.25 the surface area of an ellipsoid (whose dimensions are radius of the long axis = 0.333 Ch and radius of the short axis = 0.5 Cs d. l Volume: Canopy volume.= the volume of a cone (whose dimensions are height = 0.667 Ch and radius = 0.5 Cd) + 0.5 the volu me of an ellipsoid (whose dimensions are radius of the long axis = 0.333 Ch and radius of the short axis = 0.5 Cd).

PAGE 51

39Table 3-2. Downwind profile of wind speeds generated by an airboat. Wind speeds were recorded for one minute at 100 Hz using gill propeller anemometers (Model 27106R R.M. Young Company, Traverse City, MI, USA) and a proprietary data acquisition program. Wind speed (mph) Test 1 (21Nov2003) Test 2 (21Nov2003) Test 3 (21Nov2003) Distancea 29 ft. 3 in. 36 ft. 3 in. 39 ft. 8 in. 46 ft. 8 in. 39 ft. 8 in. 46 ft. 8 in. Elevationb 10 ft. 1 in. 16 ft. 5 in. 10 ft. 1 in. 16 ft. 5 in. 10 ft. 1 in. 16 ft. 5 in. RPM Mean Max.c Mean Max. Mean Max. Mean Max. Mean Max. Mean Max. 1000 5 7 2 2 3 5 2 3 5 9 2 4 1500 10 15 2 3 9 15 3 5 10 15 4 9 2000 10 20 3 5 12 17 3 5 18 25 5 10 2500 12 32 5 10 20 25 4 9 21 25 5 12 3000 15 28 7 12 25 32 7 15 25 35 6 12 3500 28 28 3 3 26 34 7 10 18 30 3 18 4000 33 33 4 4 25 32 3 5 19 27 6 12 4500 37 37 5 5 25 38 4 7 25 35 10 20 Wind speed (mph) Test 4 (21Nov2003) Distance 18 ft. 5 in. 25 ft. 5 in. Elevation 10 ft. 1 in. 16 ft. 5 in. RPM Mean Max. Mean Max. 1000 15 25 2 4 1500 22 30 3 8 2000 33 38 6 10 2500 35 45 5 9 3000 44 52 8 10 3500 38 48 13 22 4000 33 45 4 6 4500 40 50 8 10 a Distance: Distance from the airb oat propeller to the anemometer. b Elevation: Elevation from th e ground to the anemometer. c Max.: Maximum wind speed recorded.

PAGE 52

40 Table 3-3. Calibration of motor rpm to wi nd speed. Wind speeds were recorded for one minute at 100 Hz using gill propeller anemometers (Model 27106R R.M. Young Company, Traverse City, MI, USA) and a proprietary data acquisition program. Wind speed (mph) Distancea 17 ft. 0 in. Elevationb 10 ft. 1 in. RPM Mean Max.c Std. dev. ambient 2.6 5.7 0.9 1000 9.1 15.0 2.9 1500 17.4 23.2 2.0 2000 24.5 29.4 2.4 2500 32.4 39.9 2.7 3000 38.4 46.8 2.6 3500 44.7 56.5 3.3 4000 50.0 60.4 3.7 4500 58.7 70.9 3.9 a Distance: Distance from the airb oat propeller to the anemometer. b Elevation: Elevation from th e ground to the anemometer. c Max.: Maximum wind speed recorded.

PAGE 53

41 Table 3-4. Wind speeds recorded during a road test. Data was collected for 2 min. at 0.5 Hz using two 3-cup anemometers (Met One 034B, Campbell Scientific, Inc., Logan, UT, USA). Wind speed (mph) Westa (21Apr2004) East (21Apr2004) Speedometer Mean Peak Std. dev. Mean Peak Std. dev. 10 9.7 10.50.79.112.3 1.2 20 20.7 23.01.022.125.7 1.7 30 30.3 32.80.930.633.7 1.7 40 40.2 47.12.442.544.5 1.0 50 50.6 52.51.152.553.4 0.6 60 60.6 63.21.364.867.7 1.2 70 73.5 78.51.575.780.2 2.2 a West and East refer to the location of the anem ometer with respect to the tree during testing.

PAGE 54

42 Figure 3.1. Three examples of lions ta il pruning taken from urban landscapes.

PAGE 55

43 A B C D Figure 3.2. Example of geometrically de fined lions tail pruning (lions tailed tree number 1). Progression of targeted pruni ng doses: A) no pruning; B) 15% pruning; C) 30% pruning ; and D) 45% pruning.

PAGE 56

44 A B C D Figure 3.3. Example of visually defined lio ns tail pruning (lions tailed tree number 7). Progression of targeted pruning doses: A) no pruning; B) 15% pruning; C) 30% pruning; and D) 45% pruning.

PAGE 57

45 Figure 3.4. Two examples of raising pruning type taken from urban landscapes.

PAGE 58

46 A B C D Figure 3.5. Example of geometrically de fined raised pruning (raised tree number 3). Progression of targeted pruning doses: A) no pruning; B) 15% pruning ; C) 30% pruning; a nd D) 45% pruning.

PAGE 59

47 A B C D Figure 3.6. Example of visually define d raised pruning (raised tree number 7). Progression of targeted pruning doses: A) no pruning; B) 15% pruning; C) 30% pruning; and D) 45% pruning.

PAGE 60

48 Figure 3.7. Two examples of reduction pruning type taken from urban landscapes.

PAGE 61

49 A B C D Figure 3.8. Example of geometrically de fined reduction pruning (reduced tree number 1) Progression of targeted pruning doses : A) no pruning; B) 15% prun ing; C) 30% pruning; and D) 45% pruning.

PAGE 62

50 A B C D Figure 3.9. Example of visually define d reduction pruning (reduced tree number 7). Progression of targeted pruning doses: A) no pruning; B) 15% pruning; C) 30% pruning; and D) 45% pruning.

PAGE 63

51 A B C D Figure 3.10. Two examples of structural pruning type taken from urban landscapes before (A and C) and after (B and D) respectively.

PAGE 64

52 A B C D Figure 3.11. Examples of structural pruni ng (structurally pruned tree number 1). Progression of targeted pruning doses: A) no pruning; B) 15% pruning; C) 30% pruning; and D) 45% pruning.

PAGE 65

53 A B Figure 3.12. Example of thinning taken from urban landscapes before (A) and after (B).

PAGE 66

54 A B C D Figure 3.13. Example of thinni ng (thinned tree number 4). Progr ession of targeted pruning doses : A) no pruning; B) 15% prunin g; C) 30% pruning; and D) 45% pruning.

PAGE 67

55 A B Figure 3.14. Airboat used to generate wind: A) side view and B) rear view.

PAGE 68

56 0 10 20 30 40 50 60 70 80 90 100 010002000300040005000 Motor rpmMean wind speed (mph) y=0.02x-2.00 y=0.02x-4.59 y=0.01x-0.64 West East Gurley Figure 3.15. Calibration curv es of the mean wind speeds as measured by Campbell (West and East) and Young (Gurley) anemometers. West anemometer was located 10 ft. 2 in. downwind; East anemometer 23 ft. 3 in. downwind; and Gurley anemometry located 17 ft. downwind.

PAGE 69

57 Figure 3.16. The four South cable extension transducers (CET).

PAGE 70

58 0 5 10 15 20 25 0.00.10.20.30.40.50.60.70.80.91.0 mVin. y = 10.84x 0.03 y = 10.84x + 0.02 y = 16.05x 0.04 y = 26.75x 0.24 s18 s30 s42 s54 Figure 3.17. Southern cable extension tran sducer (CET) calibration curves. Labels indicate the location of the CET with re spect to the tree and propeller and the elevation on the trunk during testing (e.g .: s18 = southern CET elevated to 18 inches). R2 = 1.0 for all regressions. mV = millivolts; in. = inches

PAGE 71

59 0 5 10 15 20 25 0.00.10.20.30.40.50.60.70.80.91.0 mVin. y = 10.86x 0.06 y = 10.82x 0.03 y = 16.05x 0.13 y = 26.73x 0.19 n18 n30 n42 n54 Figure 3.18. Northern cable extension tran sducer (CET) calibration curves. Labels indicate the location of the CET with re spect to the tree and propeller and the elevation on the trunk during testing (e.g .: n18 = northern CET elevated to 18 inches). R2 = 1.0 for all regressions. mV = millivolts; in. = inches

PAGE 72

60 anemometer East (aesp) anemometer West (awsp) airboat South CETs (s18, s30, s42, s54) North CETs (n18, n30, n42, n54) N Primary direction of wind flow. A South CETs (s54) (s42) (s18) (s30) anemometer West anemometer East B Figure 3.19. Schematics (birds-eye (A) a nd profile (B) views of cable extension transducer (CET) and anemometer positions.

PAGE 73

61 ai wire extension + average of n# CET readings at ambient by pruning dose bi wire extension + average of s# C ET readings at ambient by pruning dose c distance between n# and s# CETs measured manually at set-up Ai = Cos-1((bi 2c2ai 2) 2bic) Bi = Cos-1((ai 2c2bi 2) 2aic) yi = (biSinAi + aiSinBi)/2 di = yi TanBi ei = yi TanAi af wire extension + n# CET reading at de signated time, rpm, and pruning dose bf wire extension + s# CET reading at designated time, rpm, and pruning dose c distance between n# and s# CETs measured manually at set-up Af = Cos-1((bf 2c2af 2) 2bfc) Bf = Cos-1((af 2c2bf 2) 2afc) yf = (bfSinAf + afSinBf)/2 df = yf TanBf ef = yf TanAf x# = di-df ei-ef y# = yi-yf m# = sqrt( x2+ y2) = displacement of the trunk Figure 3.20. Cable extension transducer calc ulations (these are for one specific height represented in inches by subscript #). Tree Bf yf y Af af ei di yi Bi Ai c bi ai ef df c bf m n# CET s# CET x

PAGE 74

62 A B Figure 3.21. Apparatus used to determin e longitudinal Youngs modulus: A) Coupon test, and B) trunk test.

PAGE 75

63 A B Figure 3.22. Apparatus used to fix the trunk and rootball of trees duri ng testing. A) Steel plate and basket used to fi x the rootball. B) Tree fi xed as it was during testing .

PAGE 76

64 Figure 3.23. Data acquisition system used in the field: A) Campbell Scientific CR10X datalogger, B) AM 416 multiplexer and C) 12 VDC regulated power supply. A B C

PAGE 77

65 CHAPTER 4 RESULTS Randomization Trees were randomly assigned to a pruning type. Ten physical characteristics used to select trees were used to compare trees assigned to pruning types. There were no differences among pruning types with one excepti on. Elevation to th e vertical center of the canopy was statistically ( P = 0.039) lower (143 in.) for trees assigned to the thinned pruning type than for those assigned to the re duced pruning type (154 in .). Since this was the only difference among types, no adjustments were made in the assignment of trees to pruning types. Pruning Dose All pruning doses were calculated from an average canopy dry mass. Average canopy dry mass was calculated from 13 of the 27 trees tested. Table 4-1 shows foliage, stem, and total canopy dry mass for the 13 trees used to determine average canopy dry mass. Foliage, stem, and total canopy dry mass for all trees were within three standard deviations of the mean for each category. Thus none of the 13 trees used to determine average canopy dry mass were obvious outlier s as defined by the Empirical Rule in statistics (Ott a nd Longnecker 2001). Pruning doses were calculated as percenta ge of dry mass removed from the average tree and from the actual tree wh ere possible (Table 4-2). Ta ble 4-2 also shows targeted pruning dose and foliage, stem, and total dry mass removed at each pruning. As expected, pruning doses calculated from aver age canopy dry mass were different than

PAGE 78

66 those calculated from actual canopy dry mass for the 13 trees used to determine the average. Difference in calculated dose was most apparent in the percent foliage dry mass removed 3.5% on average with the greatest difference 13% for structurally pruned tree number 3 at the 45% targeted dose level. To keep dose consistent in the analysis, dose was calculated using the aver age canopy dry mass. Percent foliage dry mass was chosen for use in the statistical analyses beca use ANSI (2001) defines pruning dose as a percentage of foliage removed. Percen t stem dry mass and percent total dry mass removed from the canopy correlated well to percent foliage dry mass (R2 = 0.76 and 0.85 respectively). Figure 4.1 shows the simila rity between pruning doses calculated from foliage dry mass and those calculated from total dry mass. Dose levels were more variable than expected. This was significant since comparisons among pruning types depended on consistent dose levels among types. Efforts to determine pruning dose using ca nopy dimensions were futile. Figure 4.1A shows that foliage removed from lions tailed trees one through three at all dose levels was considerably less than the targeted dose levels. On the other hand, excess foliage was removed from raised and reduced trees tw o through four at all dose levels. Lions tailed trees four through seven, and raised and reduced trees five through seven were added to better approximate targeted pruning dose levels. Pruning dose levels for those additional ten trees were determined in the field as a visual estimate of the foliage removed. Dose levels for structurally pr uned trees one through th ree and thinned trees one through four were also determined visua lly and they were well distributed about the targeted levels (Figure 4.1A). No additional trees were tested for structural or thinned pruning types.

PAGE 79

67 Wind Speed Vertical wind profile was measured once be fore and twice after testing all 27 trees. Vertical wind profile was calculated as an average of 120 measurements taken per elevation per day on the three days it was eval uated (Figure 4.2A). Generated winds did not disperse vertically. As motor rpm increased, vertical profile became more concentrated (conical). One three cup anemomet er was used to measure all elevations in the vertical wind profile seque ntially. The same three cup anemometer was used to measure wind speeds at the vertical center of each tree canopy during testing. Figure 4.2B shows mean wind speeds re corded for all 27 trees at 2750 rpm on test dates as well as average vertical wind profile measured prior to and post testing. Wind speeds recorded on testing dates did not correspond w ith values in the average vertical wind profile (Figure 4.2B). It should be noted th at because generated wi nds did not disperse vertically, upper and lower por tions of tree canopies were subjected to lower wind speeds than the center of a canopy. An experiment-wise comparison of genera ted wind speeds showed that selected motor rpm produced desired steps in wind speed (Table 4-3). The statistical difference between wind speeds generated at 1250 mo tor rpm before and after higher rpm was attributed to a large number of measurements used in comparisons. This statistical difference had no substantial effect since the two means at 1250 rpm differed by less than three miles per hour, while means between sepa rate rpm differed by nearly 15 miles per hour. Wind speeds were compared among pruning types by motor rpm. Raised trees generally experienced the hi ghest generated wind speeds while lions tailed and structurally pruned trees expe rienced lower wind speeds at every rpm (Table 4-4). Least

PAGE 80

68 squares means adjusted with Tukeys method indicated that at the highest motor rpm, wind speeds were statistically different among all pruning types except among thinned and reduced types (Table 4-4). Average wi nd speed recorded for lions tailed trees differed from that recorded for reduced trees by 11.47 mph. This variation in generated wind speeds was corrected for in the analysis. Wind speeds were compared among trees a nd pruning dose levels by rpm. Time series plots show that recorded wind speeds were not consistent between trees (Figure 4.3A and B). Wind speeds were fairly consis tent within an rpm on some days (Figure 4.3A and B) but showed tremendous variation within an rpm on others (Figure 4.3C and D). Table 4-5 shows wind speeds recorded for each tree at the highest motor rpm. Response Two variables were considered to represen t tree response to wind loading. The first was trunk movement at an elevation 54 inch es above the topmost root (m54 Figure 4.4A). The second was the area below the de flected trunk in the plane of primary wind flow (dya Figure 4.5A). Trunk moveme nt (m54) and deflected area (dya) were perfectly correlated with each other (R2 = 1.0 at highest motor rpm and no foliage removed) and both were comparably correlated with wind speed (R2 = 0.26 and 0.27 respectively at highest motor rpm and no foliage removed). Both measures showed that trunk movement tracked changes in wind speed s relatively we ll within a tree (Figures 4.4B and 4.5B). Both also showed ther e were differences in trunk movement among trees independent of wind speed. Mean wind speed at the highest rpm (Table 4-5) was decidedly greater for lions tailed tree number 2 (LT02) than it was for lions tailed tree number 1 (LT01), but m54 and dya were both lower for LT02 than for LT01. Deflected area appeared more responsive to changes in wind speeds but it was also more variable

PAGE 81

69 than m54 and it provided a measure of trunk movement in only one horizontal direction, that parallel to the primary wind flow. Trunk movement (m54) provided a measure of deflection in both horizontal directions, para llel and perpendicular to the primary wind flow and it was most responsive of all indi vidual measurements. Therefore, m54 was chosen for use in the analysis and dya was considered redundant. Trunk movement (m54) proved to be an effective measure of tree response regardless of quality of the corresponding wind speed profile. Time series plots of m54 (Figure 4.6) showed better re solution of individual rpm than time series plots of wind speed (Figure 4.3). They also showed eff ect of pruning dose as re duced movement with each repeated sequence of motor rpm (Figure 4.6). Tree testing date was expected to influe nce trunk movement. Trees were tested from May 2004 to November 2004 so there wa s ample time for grow th and development of trunks. Trees tested early were blocked in time by type but additional lions tailed (numbers 4-7), raised (numbers 5-7), and redu ced (numbers 5-7) trees and all structurally pruned trees were not blocked in time. Fi gure 4.7 shows that trunk movement (m54), at the highest motor rpm and no foliage removed, appeared to decrease with time from May through November while wind speeds did not (R2 = 0.53). However, comparison of lions tailed trees tested early in the year with those tested later showed no differences ( P = 0.664) in m54 attributable to date tested There was not enough evidence to conclude that tree growth had a significant impact on trunk movement over the dates tested. Longitudinal Youngs modulus was expected to influence trunk movement as well. Results from efforts to determine Youngs modulus of whole trunks and coupons were inconclusive (Trachet 2005), so effort s were abandoned and Youngs modulus was

PAGE 82

70 assumed to be uniform within and among trees tested. Structurally pruned tree number two suffered significant water stress whic h provided reason to believe its Youngs modulus differed from that of ot her trees. It was left out of the analysis and there were no other anomalies that knowingly might have caused variation in Youngs modulus among trees. As a result, the assumption of material homogeneity seems appropriate for the trees tested. Analytical Approach A principal goal of this experimental st udy was to compare effects of pruning type on tree response to wind load. Because pr uning dose and wind speed were not recorded at set levels, orthogonal levels of both were sought to simplify comparisons between pruning types. Trunk movement (m54) was regressed against wind speeds (Wind) and pruning doses (Dose) measured for each tree (Tables 4-6 and 4-7). A complete two factor quadratic empirical model was used fo r all regressions as given in Equation 4.1, where b0 b5 are constants generated as the regression coefficients. Dose Wind b Dose b Wind b Dose b Wind b b m 5 2 4 2 3 2 1 054 (4.1) A quadratic model was chosen because wind speed is accounted for in the standard equation used to calculate drag by the fluid ve locity term (which is squared) and pruning dose results in a reduction in surface area (als o a squared term). Two regressions were conducted: one using all measurements of wind speed and m54 (Table 4-6), and one using measurements of wind speed and m54 averaged within pruning type, tree, pruning dose and rpm (Table 4-7). Figure 4.8 show s a graphical representation of all m54 measurements and average m54 measurements for raised tree number 5. Regressions using averaged m54 had higher R2 values than those using all measurements and they

PAGE 83

71 were simpler since pruning dose coefficients we re almost entirely insignificant. All regressions using averages of wind speed and m54 had R2 values in excess of 0.94. Regression equations were used to predic t trunk movement (pm54) for all trees at orthogonal levels of pruning dose and wind speed. For some trees predicted values were interpolated between large gaps in the data (Figure 4.9A). For others predicted values were extrapolated some distance from the da ta (Figure 4.9B and C). Figure 4.10 gives a graphical summary of the procedure used to generate predicted average m54 (pm54) values for an individual tree. The pm54 values were used to compare pruning type, pruning dose and wind speed in a three-way analysis of variance ( ANOVA). Table 4-8 shows that there was essentially no difference between ANOVA resu lts using pm54 values predicted from equations using all measurements of m54 a nd wind speed, and those using the averages m54 and wind speed. Similarly, there was little difference between ANOVAs using m54 versus those using deflected area (dya). Ta ble 4-8 confirms that average m54 was as good a measure of tree response as m54, deflect ed area (dya), or average dya. Average trunk movement (m54) was used to complete the analysis. The three-way ANOVA using predicted tr unk movement (pm54) (Table 4-9) showed that interaction between pruni ng type, pruning dose, and wind speed, and interaction between pruning type and pruni ng dose were statistically insignificant ( P = 1.00 and P = 0.74 respectively). However, inte ractions between pruning type and wind speed (Figure 4.11) and between pruning dose and wind speed (Figure 4.12) were significant ( P < 0.0001 for both). Throwing out earl y lions tailed, ra ised and reduced

PAGE 84

72 trees that had extreme dose levels (Figure 4. 1A) did not affect the results. Further analysis was conducted to quantify th e significance of the interactions. Predicted trunk movement (pm54) was used to compare pruning types at each wind speed averaged across pruning dose. Table 4-9 shows least squares means, adjusted with Tukeys method, ordered by pruning type at each targeted wind speed. Predicted trunk movement of thinned trees was statistically gr eater than the pm54 of structurally pruned, raised, and lions tailed trees at 45 mph and gr eater than all other trees at 60 mph. There were no statistical differences between pruning types at lower wind speeds. These results are seen clearly in the pruning type wind sp eed interaction profile plot (Figure 4.11). Thinning was the least effectiv e pruning type for reducing wind load and there were no differences between the other four types. Predicted trunk movement (pm54) was then used to compare wind speeds within each pruning type averaged across pruning dose. Table 4-10 shows least squares means, adjusted with Tukeys method, ordered by pruning dose at each ta rgeted wind speed. Predicted trunk movement increased for all prun ing types as wind speeds increased. The increase in movement was statistically significant ( P < 0.05) at all wind speeds for all pruning types except structural pruning. In structurally pruned trees, pm54 was only statistically different between 15 and 60 m ph wind speeds. There was no statistical difference in movement between 15 to 45 mph or between 30 to 60 mph wind speeds Overall, increases in wind speed resulte d in increased trunk movement. Regression models (Table 4-7) also show this as coefficients for wind and wind2 terms were almost always significant and positive.

PAGE 85

73 The pruning dose wind speed interaction (Table 4-11) was analyzed in the same fashion as pruning type wind speed. At 30 mph predicted trunk movement was only reduced with a 60% pruning dose compared to no pruning. At the two highest wind speeds, the 30% dose was similar to both unpr uned an the 60% pruning dose, with only pruning doses of 45% or great er predicting less trunk move ment than no pruning. At both highest wind speeds, 60% pruning dose reduced predicted trunk movement about 50% compared to no pruning. This is repres ented graphically in Figure 4.12. Therefore, pruning dose affected tree res ponse to wind load but the e ffect was not statistically significant at the low wind speeds and it was onl y statistically significant at the highest wind speeds once the 45% targeted pruning dose level was reached. Separations of pm54 among wind speeds at each pruning dose level (averaged across pruning type) showed a similar effect. Least squares means, adjusted with Tukeys method, ordered by wind speed at each targeted pruning dose are shown in Table 4-12. The effect of pruning dose is first seen at the 30% targeted dose level as pm54 at 15 mph was not statistically diffe rent than pm54 at 30 mph. At the highest targeted dose level pm54 at 30 mph was not statistically different than pm54 at 15 or 45 mph indicating that pruning dose increasingly offset the effect of wind speed. Prediction models showed this as well (Table 4-7). Although the dose and dose2 coefficients are mostly insignificant, the dose*wind coefficients are almost all signifi cant and negative. It is significant that pruning dose did not reduc e trunk movement until 30% of the foliage was removed, and that further reduction in pm54 required a doubling of that dose (60% foliage dry mass removed). Both of thes e doses exceed current recommended pruning dose levels (ANSI 2001).

PAGE 86

74Table 4-1. Foliage and stem weight for 13 trees harvested to quantify pruning dose. Pruning typea Tree no.b Remaining foliagec (lbs) Pruned foliaged (lbs) Sum foliagee (lbs) Remaining stemf (lbs) Pruned stemg (lbs) Sum stemh (lbs) Sum totali (lbs) LT 4 2.568 2.1714.73926.7805.14731.92736.666 LT 5 2.789 2.1504.93936.7925.70642.49847.436 LT 6 3.098 1.8574.95535.2555.03540.29045.246 LT 7 1.773 2.1403.91331.0585.27736.33540.248 RA 3 1.202 3.4254.62714.62712.73027.35731.984 RA 4 0.808 4.2335.04114.14016.42630.56635.608 RA 6 2.116 2.8374.95219.38711.56230.94935.901 RA 7 2.207 2.1894.39623.7098.62132.32936.725 RE 5 2.114 2.0114.12529.8443.08632.93037.055 RE 6 2.410 1.8604.27033.0963.52036.61640.885 RE 7 1.564 2.2623.82731.0564.04735.10238.929 ST 1 2.162 2.3564.51930.9008.63739.53644.055 ST 3 0.975 2.6833.65826.6328.93035.56239.220 mean =4.458 mean =34.76939.227 std. err.j =0.131 std. err =1.1991.201 a Pruning types: LT = lions tailing, RA = raising, RE = reduction, ST = structural, TH = thinning. b Tree no.: Number assigned to a tree within a pruni ng type (LT 4 was the fourth tree lions tailed). c Remaining foliage: Foliage dry mass remaining on the tree after the heaviest pruning dose. d Pruned foliage: Foliage dry mass removed with all pruning doses. e Sum foliage: Sum of remaining and pruned foliage. f Remaining stem: Stem dry mass remaining on the tree after the heaviest pruning dose. g Pruned stem: Stem dry mass removed with all pruning doses. h Sum stem: Sum of remaining and pruned stem. i Sum total: Sum foliage plus sum stem. j Std. err.: Standard error of 1 standard deviation.

PAGE 87

75Table 4-2. Percent foliage dry weight, stem dry wei ght, and total dry weight removed with each pruning dose. Foliage dry mass removed Stem dry mass removed Total dry mass removed Pruning typea Tree no.b targeted dosec (%) Sumd (lbm) Average dosee (%) Actual dosef (%) Sum (lbm) Average dose (%) Actual dose (%) Sum (lbm) Average dose (%) Actual dose (%) LT 1 15 0.532 12 1.598 5 2.130 5 LT 1 30 0.759 17 2.496 7 3.255 8 LT 1 45 1.328 30 3.933 11 5.261 13 LT 2 15 0.202 5 0.476 1 0.678 2 LT 2 30 0.817 18 2.465 7 3.282 8 LT 2 45 1.336 30 4.051 12 5.386 14 LT 3 15 0.505 11 1.544 4 2.049 5 LT 3 30 0.895 20 2.624 8 3.519 9 LT 3 45 1.448 32 3.625 10 5.074 13 LT 4 15 0.459 10 10 1.231 4 4 1.690 4 5 LT 4 30 1.073 24 23 2.844 8 9 3.918 10 11 LT 4 45 2.171 49 46 5.147 15 16 7.318 19 20 LT 5 15 0.572 13 12 1.878 5 4 2.450 6 5 LT 5 30 1.320 30 27 4.117 12 10 5.437 14 11 LT 5 45 2.150 48 44 5.706 16 13 7.856 20 17 LT 6 15 0.560 13 11 1.975 6 5 2.536 6 6 LT 6 30 1.215 27 25 3.740 11 9 4.955 13 11 LT 6 45 1.857 42 37 5.035 14 12 6.893 18 15 LT 7 15 0.635 14 16 1.902 5 5 2.537 6 6 LT 7 30 1.273 29 33 3.579 10 10 4.852 12 12 LT 7 45 2.140 48 55 5.277 15 15 7.417 19 18

PAGE 88

76Table 4-2. Continued Foliage dry mass removed Stem dry mass removed Total dry mass removed Pruning typea Tree no.b Targeted dosec (%) Sumd (lbm) Average dosee (%) Actual dosef (%) Sum (lbm) Average dose (%) Actual dose (%) Sum (lbm) Average dose (%) Actual dose (%) RA 1 45 3.720 83 16.800 48 20.520 52 RA 2 15 2.583 58 10.291 30 12.875 33 RA 2 30 3.450 77 13.743 40 17.194 44 RA 2 45 3.786 85 14.724 42 18.510 47 RA 3 15 2.153 48 47 9.615 28 35 11.768 30 37 RA 3 30 2.564 58 55 10.744 31 39 13.309 34 42 RA 3 45 3.425 77 74 12.730 37 47 16.155 41 51 RA 4 15 1.858 42 37 6.632 19 22 8.489 22 24 RA 4 30 3.571 80 71 14.203 41 46 17.774 45 50 RA 4 45 4.233 95 84 16.426 47 54 20.659 53 58 RA 5 15 1.070 24 3.549 10 4.619 12 RA 5 30 1.839 41 6.320 18 8.159 21 RA 5 45 2.810 63 11.940 34 14.751 38 RA 6 15 1.101 25 22 3.782 11 12 4.882 12 14 RA 6 30 2.071 46 42 7.365 21 24 9.436 24 26 RA 6 45 2.837 64 57 11.562 33 37 14.399 37 40 RA 7 15 0.735 16 17 2.459 7 8 3.194 8 9 RA 7 30 1.527 34 35 5.819 17 18 7.346 19 20 RA 7 45 2.189 49 50 8.621 25 27 10.809 28 29

PAGE 89

77Table 4-2. Continued Foliage dry mass removed Stem dry mass removed Total dry mass removed Pruning typea Tree no.b Targeted dosec (%) Sumd (lbm) Average dosee (%) Actual dosef (%) Sum (lbm) Average dose (%) Actual dose (%) Sum (lbm) Average dose (%) Actual dose (%) RE 1 15 1.159 26 0.939 3 2.099 5 RE 1 30 1.757 39 2.225 6 3.983 10 RE 1 45 2.239 50 4.717 14 6.956 18 RE 2 15 1.834 41 1.948 6 3.782 10 RE 2 30 2.992 67 4.830 14 7.822 20 RE 2 45 3.651 82 8.808 25 12.458 32 RE 3 15 1.326 30 1.052 3 2.378 6 RE 3 30 2.472 55 3.034 9 5.506 14 RE 3 45 3.252 73 6.440 19 9.692 25 RE 4 15 1.665 37 1.347 4 3.012 8 RE 4 30 2.808 63 3.514 10 6.322 16 RE 4 45 3.789 85 7.826 23 11.615 30 RE 5 15 0.648 15 16 0.641 2 2 1.289 3 3 RE 5 30 1.333 30 32 1.595 5 5 2.929 7 8 RE 5 45 2.011 45 49 3.086 9 9 5.097 13 14 RE 6 15 0.603 14 14 0.659 2 2 1.261 3 3 RE 6 30 1.410 32 33 2.197 6 6 3.607 9 9 RE 6 45 1.860 42 44 3.520 10 10 5.380 14 13 RE 7 15 0.802 18 21 0.929 3 3 1.731 4 4 RE 7 30 1.531 34 40 2.235 6 6 3.766 10 10 RE 7 45 2.262 51 59 4.047 12 12 6.309 16 16

PAGE 90

78Table 4-2. Continued Foliage dry mass removed Stem dry mass removed Total dry mass removed Pruning typea Tree no.b Targeted dosec (%) Sumd (lbm) Average dosee (%) Actual dosef (%) Sum (lbm) Average dose (%) Actual dose (%) Sum (lbm) Average dose (%) Actual dose (%) ST 1 15 0.864 19 19 2.408 7 6 3.272 8 7 ST 1 30 1.691 38 37 4.931 14 12 6.621 17 15 ST 1 45 2.356 53 52 8.637 25 22 10.993 28 25 ST 2 15 0.873 20 3.564 10 4.437 11 ST 2 30 1.539 35 7.501 22 9.041 23 ST 2 45 1.976 44 9.972 29 11.948 30 ST 3 15 0.869 19 24 2.825 8 8 3.693 9 9 ST 3 30 1.860 42 51 5.706 16 16 7.566 19 19 ST 3 45 2.683 60 73 8.930 26 25 11.612 30 30 TH 1 15 0.907 20 1.365 4 2.272 6 TH 1 30 1.781 40 3.046 9 4.827 12 TH 1 45 2.381 53 4.203 12 6.584 17 TH 2 15 0.786 18 0.969 3 1.754 4 TH 2 30 1.578 35 2.171 6 3.749 10 TH 2 45 2.570 58 4.213 12 6.783 17 TH 3 15 0.735 16 1.504 4 2.238 6 TH 3 30 1.639 37 2.715 8 4.355 11 TH 3 45 2.607 58 4.843 14 7.451 19 TH 4 15 1.007 23 1.796 5 2.803 7 TH 4 30 1.970 44 3.943 11 5.913 15 TH 4 45 2.994 67 7.256 21 10.250 26 a Pruning types: LT = lions tailing, RA = raising, RE = reduction, ST = structural, TH = thinning. b Tree no.: Number assigned to a tree within a pruni ng type (i.e. LT 4 is the fourth tree lions tailed). c Targeted dose: Percentage of foliage dry mass intended to be removed by pruning. d Sum: Total dry mass removed at specified pruning dose (folia ge, stem, and total respectively). Dry mass was summed incrementa lly within a tree (i.e. sum at targeted dose 45 = sum of the dry mass removed at 15, 30 and 45 % levels). e Average dose: Dose calculated from an av erage tree canopy dry mass [foliage (4.458 lbm), stem (34.769 lbm), and total (39.227 lbm) respectively] f Actual dose: A value in this column indicates that dose wa s calculated from the actual tree canopy dry mass (foliage, stem, an d total respectively). A dash indicates the entire tree was not weighed.

PAGE 91

79 Table 4-3. Wind speeds (mph) recorded during testing and least squares means ( P = 0.05) adjusted with Tukeys method. Motor rpma 1 1251 2000 2750 1252 2 n 6480 129561292912960129606463 Meanb 2.70 ac 23.26 b38.05 c52.51 d20.99 e2.93 a Std. err. 0.02 0.06 0.09 0.12 0.06 0.02 Max 10.47 34.62 53.40 75.77 32.83 12.26 a Motor rpm 1 and 2 are ambient conditions and 1251 and 1252 are 1250 rpm before and after the higher rpm respectively. b Mean is an average of all measurements within an rpm across pruning type, tree, and pruning dose. c Means with the same letter within ro ws are not significantly different ( P < 0.05) based on LS mean separations adjusted using Tukeys method Table 4-4. Wind speed (mph) by pruning type and motor rpm. 1a 2750 Pruning typeb Meanc Max.d Pruning type Mean Max LT 2.72 0.03 10.47 LT 47.25 0.27 ae 75.77 RA 2.16 0.04 9.57 RA 58.72 0.19 b 73.08 RE 2.43 0.03 6.89 RE 52.71 0.24 c 72.19 ST 4.03 0.06 9.57 ST 50.37 0.41 d 72.19 TH 2.93 0.05 8.68 TH 53.64 0.19 c 72.19 1251 1252 Pruning type Mean Max Pruning type Mean Max LT 21.27 0.13 34.62 LT 18.54 0.12 32.83 RA 25.93 0.10 33.73 RA 24.55 0.08 31.04 RE 23.80 0.11 33.73 RE 21.45 0.11 31.94 ST 19.10 0.20 32.83 ST 17.44 0.18 31.04 TH 24.91 0.10 33.73 TH 21.81 0.12 31.94 2000 2 Pruning type Mean Max Pruning type Mean Max LT 34.92 0.19 53.40 LT 3.02 0.04 12.26 RA 42.22 0.15 52.51 RA 2.42 0.04 10.47 RE 37.58 0.17 53.40 RE 2.68 0.03 9.57 ST 34.80 0.29 50.72 ST 4.06 0.05 7.79 TH 40.53 0.15 52.51 TH 3.16 0.05 9.57 a Motor rpm 1 and 2 are ambient conditions and 1251 and 1252 are 1250 rpm before and after the higher rpm respectively. b Pruning types: LT = lions tailing, RA = raising, RE = reduction, ST = structural, TH = thinning. c Mean: Mean standard error of one standard deviation with n from 720 to 3360 per pruning type. d Max: Maximum value recorded. e Means with the same letter within columns are not significantly different ( P < 0.05) based on LS mean separations adjusted using Tukeys method

PAGE 92

80 Table 4-5. Wind speed, trunk movement (m54) and deflected area (dya) at 0 pruning dose and 2750 rpm. Wind speed (mph) m54 (in.) dya (in.2) Pruning typea Tree no.b Meanc Std. err.d Maxe Mean Std. err. Max Mean Std. err. Max LT 1 47.50 0.50 60.56 3.05 0.03 4.19 106.73 1.22 145.91 LT 2 66.59 0.54 73.98 2.96 0.04 4.05 102.27 1.26 137.73 LT 3 36.74 0.57 50.72 2.48 0.02 3.15 87.50 0.69 111.86 LT 4 33.63 0.91 54.30 2.76 0.02 3.33 95.81 0.66 118.30 LT 5 28.63 0.76 58.77 1.46 0.02 2.16 51.37 0.54 75.40 LT 6 41.00 0.93 63.24 1.39 0.01 1.75 48.89 0.37 61.93 LT 7 51.50 0.46 60.56 1.87 0.02 2.51 66.01 0.84 90.24 RA 2 62.69 0.32 69.51 4.08 0.03 4.74 150.95 1.10 177.49 RA 3 61.42 0.50 70.40 2.29 0.01 2.67 80.84 0.48 95.89 RA 4 63.68 0.61 72.19 4.57 0.02 5.14 157.42 0.80 177.72 RA 5 65.06 0.25 71.29 3.62 0.02 4.04 127.52 0.62 144.44 RA 6 58.96 0.79 70.40 2.36 0.01 2.76 85.08 0.53 99.33 RA 7 66.00 0.41 71.29 2.81 0.02 3.23 99.81 0.60 116.27 RE 1 52.74 0.91 66.82 2.76 0.03 3.45 97.76 1.13 122.59 RE 2 65.11 0.35 70.40 4.13 0.02 4.54 145.99 0.62 161.65 RE 3 48.62 1.03 70.40 3.56 0.03 4.65 124.42 1.10 162.83 RE 4 61.14 0.32 67.72 2.98 0.01 3.30 106.51 0.52 117.54 RE 5 56.49 0.55 65.93 2.17 0.02 2.69 76.33 0.89 94.85 RE 6 39.01 1.18 61.46 1.91 0.01 2.23 67.90 0.40 81.81 RE 7 63.01 0.28 70.40 2.57 0.02 3.05 91.18 0.69 107.84 ST 1 65.97 0.20 69.51 2.21 0.01 2.50 77.85 0.53 88.33 ST 2 32.59 0.87 56.98 1.21 0.01 1.42 42.44 0.37 50.20 ST 3 42.74 0.93 62.35 2.11 0.04 2.87 72.53 1.27 99.75 TH 1 55.33 0.21 63.24 4.23 0.03 5.28 146.49 1.19 184.15 TH 2 52.13 0.35 60.56 3.21 0.02 3.85 111.81 0.71 137.40 TH 3 40.28 0.31 48.93 3.00 0.02 3.67 104.18 0.75 129.55 TH 4 61.73 0.77 72.19 3.60 0.02 4.09 126.35 0.87 143.72 a Pruning types: LT = lions tailing, RA = raising, RE = reduction, ST = structural, TH = thinning. b Tree no.: Number assigned to a tree within a pruni ng type (i.e. LT 4 is the fourth tree lions tailed). c Mean: Average of 120 measurements. d Std err: Standard error of one standard deviation. e Max: Maximum value recorded.

PAGE 93

81Table 4-6. Regressi on coefficients and R2 values generated using all measurements of trunk movement (m54) and wind speed (Wind) within a pruning type, tree, pruning dose (Dose), and rpm treatment combination in a complete two factor quadratic empirical model. R2 values for regressions of deflected area (dya) are given for comparison. Pruning typea Tree no.b Intercept Wind Dose Wind2 Dose2 Wind*Dose R2 (m54)c R2 (dya)d LT 1 NSe 0.0394638 0.0047193 0.0004759 -0.0002174 -0.0010233 0.925 0.924 LT 2 0.0521557 0.0283154 -0.0159439 0.0001968 0.0005834 -0.0004388 0.911 0.917 LT 3 NS 0.0590989 -0.0065756 -0.0001017 0.0001145 -0.0007349 0.805 0.805 LT 4 0.0678023 0.0590676 -0.0098421 0.0000748 0.0001746 -0.0007114 0.818 0.818 LT 5 -0.1399732 0.0564957 -0.0026469 -0.0003689 0.0000586 -0.0003598 0.761 0.759 LT 6 0.0562677 0.0228512 -0.0085027 0.0001034 0.0001678 -0.0003292 0.883 0.884 LT 7 NS 0.0351883 -0.0065309 -0.0000014 0.0001463 -0.0006956 0.836 0.834 RA 2 NS 0.0478202 NS 0.0002678 0.0000184 -0.0006647 0.952 0.949 RA 3 -0.0642992 0.0296015 0.0017010 0.0000995 0.0000141 -0.0004418 0.955 0.950 RA 4 0.1265204 0.0482748 NS 0.0003028 -0.0000210 -0.0005730 0.961 0.962 RA 5 NS 0.0345920 0.0021254 0.0003331 -0.0000616 -0.0003397 0.975 0.974 RA 6 -0.0916750 0.0293212 0.0049636 0.0001484 -0.0000387 -0.0004692 0.909 0.909 RA 7 NS 0.0311609 0.0024453 0.0001711 -0.0000527 -0.0003323 0.953 0.951 RE 1 0.0473252 0.0432073 0.0051511 0.0001259 -0.0001726 -0.0005428 0.909 0.905 RE 2 -0.0933176 0.0357488 0.0054225 0.0004122 -0.0000259 -0.0004044 0.895 0.894 RE 3 NS 0.0596352 NS 0.0001248 -0.0000479 -0.0004766 0.904 0.901 RE 4 NS 0.0333231 0.0051231 0.0002828 -0.0000796 -0.0003062 0.966 0.962 RE 5 0.0489778 0.0269396 -0.0044869 0.0001827 0.0000754 -0.0002095 0.926 0.924 RE 6 NS 0.0530362 NS -0.0002288 NS -0.0002129 0.747 0.747 RE 7 0.0492041 0.0508709 NS 0.0003363 -0.0001186 -0.0000891 0.958 0.956

PAGE 94

82Table 4-6. Continued. Pruning typea Tree no.b Intercept Wind Dose Wind2 Dose2 Wind*Dose R2 (m54)c R2 (dya)d ST 1 -0.1543650 0.0288068 0.0037946 0.0001217 -0.0000384 -0.0003226 0.959 0.957 ST 2 NS 0.0405986 -0.0057574 -0.0002261 0.0000806 -0.0004180 0.751 0.744 ST 3 -0.0680923 0.0411714 NS 0.0001394 NS -0.0004593 0.870 0.865 TH 1 0.0695847 0.0402283 -0.0091875 0.0005953 0.0001810 -0.0006801 0.935 0.932 TH 2 -0.0586411 0.0366951 NS 0.0004529 0.0000266 -0.0004189 0.956 0.952 TH 3 NS 0.0468575 NS 0.0003885 0.0000353 -0.0005330 0.897 0.897 TH 4 -0.1534129 0.0405974 0.0035574 0.0002915 NS -0.0004013 0.933 0.935 a Pruning types: LT = lions tailing, RA = raising, RE = reduction, ST = structural, TH = thinning. b Tree no.: Number assigned to a tree within a pruni ng type (LT 4 was the fourth tree lions tailed). c R2 (m54): regressions using all wind speed and m54 measurements within pruning type, tree, pruning dose, and rpm. d R2 (dya): regressions using all wind speed and dya measurements within pruning type, tree, pruning dose, and rpm. e NS: Not statistically significant at P = 0.05.

PAGE 95

83Table 4-7. Regressi on coefficients and R2 values generated using averag es of trunk movement (avm54) and wind speed (Wind) within a pruning type, tree, pruning dose (Dose), and rpm treatment combination in a comp lete two factor quadratic empirical model. R2 values for regressions of average deflected area (avdya) are given for comparison. Pruning typea Tree no.b Intercept Wind Dosec Wind2 Dose2 Wind*Dose R2 (avm54)d R2 (avdya)e LT 1 NSf 0.0346330NS 0.0006744NS -0.00125330.9910.991 LT 2 NS 0.0249140NS 0.0002559NS -0.00043170.9860.989 LT 3 NS 0.0496499NS NS NS -0.00117340.9490.950 LT 4 NS 0.0588409NS NS NS -0.00089940.9510.952 LT 5 NS 0.0203197NS 0.0010574NS -0.00042780.9940.994 LT 6 NS 0.0206290NS 0.00025130.0001765-0.00050260.9800.981 LT 7 NS 0.0326324NS NS NS -0.00041300.9830.983 RA 2 NS 0.0460744NS 0.0002787NS -0.00064880.9800.978 RA 3 NS 0.0272470NS 0.0001357NS -0.00043050.9790.976 RA 4 NS 0.0467960NS 0.0003508NS -0.00057820.9760.979 RA 5 NS 0.0300226NS 0.0004070NS -0.00030640.9950.995 RA 6 NS 0.0247701NS 0.0002553NS -0.00047740.9790.979 RA 7 NS 0.0232733NS 0.0003050NS -0.00025410.9940.994 RE 1 NS 0.0441449NS 0.0002051NS -0.00066380.9750.975 RE 2 NS 0.0251932NS 0.0005598NS -0.00028760.9480.949 RE 3 NS 0.0463334NS 0.0004676NS -0.00048370.9900.990 RE 4 NS 0.0326210NS 0.0003099-0.0000629-0.00032460.9890.989 RE 5 NS 0.0200714NS 0.0003093NS NS 0.9830.984 RE 6 NS 0.0363065NS 0.0004618NS -0.00034100.9470.947 RE 7 NS 0.0200764NS 0.0003667NS -0.00013500.9860.986

PAGE 96

84Table 4-7. Continued. Pruning typea Tree no.b Intercept Wind Dosec Wind2 Dose2 Wind*Dose R2 (avm54d) R2 (avdyae) ST 1 NS 0.0266158NS 0.0001453NS -0.00031570.9850.986 ST 2 NS 0.0345746NS NS NS -0.00067890.9680.969 ST 3 NS 0.0275358NS 0.0005033NS -0.00048890.9880.988 TH 1 NS 0.0365198NS 0.0006660NS -0.00065020.9880.989 TH 2 NS 0.0337152NS 0.0005155NS -0.00042850.9930.993 TH 3 NS 0.0313188NS 0.0008477NS -0.00061260.9820.983 TH 4 NS 0.0299773NS 0.0005137NS -0.00041970.9860.986 a Pruning types: LT = lions tailing, RA = raising, RE = reduction, ST = structural, TH = thinning. b Tree no.: Number assigned to a tree within a pruni ng type (LT 4 is the fourth tree lions tailed). c Dose: Percentage of foliage dry mass removed from the tree with pruning. d avm54: Wind speed and m54 measurements averaged within pruning type, tree, pruning dose, and rpm. e avdya: Wind speed and dya measurements averaged within pruning type, tree, pruning dose, and rpm. f NS: Not statistically significant at P = 0.05.

PAGE 97

85Table 4-8. ANOVA of predicted tr unk movement (p_avm54) and pred icted deflected area (p_avdya). p_m54a (in.) p_avm54b (in.) p_dyac (in.2) p_avdyad (in.2) Source of variation F P > F F P > F F P > F F P > F Pruning typee 41.49< 0.00121.94< 0.001 42.16< 0.00121.98< 0.001 Pruning dosef 83.49< 0.00166.18< 0.001 84.22< 0.00167.11< 0.001 Pruning type pruning dose 0.580.8980.750.739 0.620.8700.780.708 Wind speedg 395.29< 0.001399.48< 0.001 396.02< 0.001402.90< 0.001 Pruning type wind speed 7.37< 0.0013.69< 0.001 7.26< 0.0013.54< 0.001 Pruning dose wind speed 5.96< 0.0015.02< 0.001 5.93< 0.0015.05< 0.001 Pruning type pruning dose wind speed 0.071.0000.121.000 0.071.0000.121.000 a p_m54: predicted trunk movement based on all measurements within pruning type, tree, pruning dose and motor rpm. b p_avm54: predicted trunk movement based on measurements averaged within pruning type, tree, pruning dose and motor rpm. c p_dya: predicted trunk deflected area based on all measurements within pruning type, tree, pruning dose and motor rpm. d p_avdya: predicted trunk deflected area base d on measurements averaged within pruning type, tree, pruning dose and motor rpm. e Pruning types: LT = lions tailing, RA = raising, RE = reduction, ST = structural, TH = thinning. f Pruning dose (percentage of foliage dry mass removed) levels at which average trunk movement was predicted from regression mod els. g Wind: wind speed (mph) levels at which average trunk movement was predicted from regression models.

PAGE 98

86Table 4-9. Least squares means of predic ted average trunk movement (p_avm54) due to interaction of pruning type and wind speed type by wind speed. p_avm54a (in.) Typeb 15 (mph)c 30 (mph) 45 (mph) 60 (mph) ST NSd NS 1.31ae 2.05a RA NS NS 1.50a 2.25a LT NS NS 1.51a 2.39a RE NS NS 1.77ab 2.70a TH NS NS 2.07b 3.32b a p_avm54: Average trunk movement predicted from regression models. b Pruning types: LT = lions tailing, RA = raising, RE = reduction, ST = structural, TH = thinning. c Wind speed (mph) levels at which average trunk movement was predicted from regression models. d NS: Not statistically significant at P = 0.05. e Means with the same letter within columns are not significantly different ( P < 0.05) based on LS mean separations adjusted using Tukeys method. Table 4-10. Least squares means of predic ted average trunk movement (p_avm54) due to interaction of pruning type and wind spee d wind speed by type. p_avm54a (in.) Windb Lions tailing Raising Re duction Structural Thinning 15 0.34ac 0.40a 0.41a 0.28ab 0.41a 30 0.82b 0.89b 1.00b 0.72abc 1.10b 45 1.51c 1.50c 1.77c 1.31bcd 2.07c 60 2.39d 2.25d 2.70d 2.05cd 3.32d a p_avm54: Average trunk movement predicted from regression models. b Wind speed (mph) levels at which average trunk movement was predicted from regression models. c Means with the same letter within columns are not significantly different ( P < 0.05) based on LS mean separations adjusted using Tukeys method.

PAGE 99

87Table 4-11. Least squares means of pred icted average trunk movement (p_avm54) due to interaction of pruning dose and wind spee d dose by wind speed. p_avm54a (in.) Doseb 15 (mph)c 30 (mph) 45 (mph) 60 (mph) 60 NSd 0.51abe 1.01ab 1.71ab 45 NS 0.70abc 1.31abc 2.11abc 30 NS 0.90abcd 1.62bcd 2.53bcd 15 NS 1.11bcd 1.94cde 2.96cde 0 NS 1.32cd 2.27de 3.40de a p_avm54: Average trunk movement predicted from regression models. b Dose: Pruning dose (percentage of foliage dry mass removed) levels at which average trunk movement was predicted from regressi on models. c Wind speed levels at which average trunk move ment was predicted from regression models. d NS: Not statistically significant at P = 0.05. e Means with the same letter within columns are not significantly different ( P < 0.05) based on LS mean separations adjusted using Tukeys method. Table 4-12. Least squares means of pred icted average trunk movement (p_avm54) due to interaction of pruning dose and wind spee d wind speed by dose. p_avm54a (in.) Windb No foliage dry mass removedc 15 % foliage dry mass removed 30 % foliage dry mass removed 45 % foliage dry mass removed 60 % foliage dry mass removed 15 0.56ad 0.46a 0.36a 0.27a 0.19a 30 1.32b 1.11b 0.90a 0.70a 0.51ab 45 2.27c 1.94c 1.62b 1.31b 1.01b 60 3.40d 2.96d 2.53c 2.11c 1.71c a p_avm54: Average trunk movement predicted from regression models. b Wind speed (mph) levels at which average trunk movement was predicted from regression models. c Pruning dose (percentage of foliage dry mass removed) levels at which average trunk movement was predicted from regression mode ls. d Means with the same letter within columns are not significantly different ( P < 0.05) based on LS mean separations adjusted using Tukeys method.

PAGE 100

88 0 10 20 30 40 50 60 70 80 90 100L T0 1 L T0 2 L T 03 LT04 LT05 LT0 6 LT07 R A0 2 RA03 R A0 4 R A 05 RA06 R A 07 R E0 1 R E 02 R E0 3 R E0 4 RE05 R E0 6 R E 07 S T0 1 S T0 2 S T0 3 TH0 1 TH0 2 TH03 TH04TreeFoliage dry mass removed (%) A Figure 4.1. Pruning dose represented as perc entage of foliage dry mass removed (A) a nd percentage of total dry mass removed (B ) with respect to an average tree canopy mass. Black bar = 15%, gray bar = 30%, and wh ite bar = 45% targeted pruning dose levels. LT = lions tailing, RA = ra ising, RE = reduction, ST = structural TH = thinning pruning types. Number represents a tree assigned to the sp ecific pruning type (27 trees total).

PAGE 101

89 0 10 20 30 40 50 60 70 80 90 100L T0 1 L T0 2 L T 03 LT04 LT05 LT0 6 LT07 R A0 2 RA03 R A0 4 R A 05 RA06 R A 07 R E0 1 R E 02 R E0 3 R E0 4 RE05 R E0 6 R E 07 S T0 1 S T0 2 S T0 3 TH0 1 TH0 2 TH03 TH04TreeTotal dry mass removed (%) B Figure 4.1. Continued.

PAGE 102

90 0 24 48 72 96 120 144 168 0.020.040.060.0 Wind speed (mph)Elevation (in.) A 96 120 144 168 0.020.040.060.0 Wind speed (mph)Elevation (in.) B Figure 4.2. Vertical profile of average ge nerated wind speeds. A) Measured before and after testing on three different days (27May2004, 9Mar2005, and 15Mar2005). Profiles represent ambient (solid line), 1250 (dash line), 2000 (dash dot line), and 2750 (dot line) moto r rpm. Wind speeds were recorded at 0.5 Hz for 4 minutes at each elevation and averaged across days within an elevation (solid circles). B) Measured during testing 27 trees at 2750 rpm. Circles and solid line represent av erage wind speeds for 27 trees while triangles and dash line re present the equivalent portion of the 2750 motor rpm profile measured before and after testing.

PAGE 103

91 A B Figure 4.3. Four profiles of wind speed s generated by the airboat during testing (measured at 0.5 Hz). A) A desirable profile (structural pruned tree number 1), B) an undesirable profile (lions ta iled tree number 5), C) a profile that progresses from good to poor (raised tr ee number 5), and D) a profile that progresses from poor to good (lions tailed tree number 6). Each profile was recorded on a different day of the year Peaks represent a sequence of rpm repeated four times.

PAGE 104

92 C D Figure 4.3. Continued.

PAGE 105

93 A B Figure 4.4. Trunk movement measured at an elevation 54 inches above topmost root (m54 thinned tree number 3 before any pruning). A) Birds eye, plan view of m54 in winds generated at all motor rpm (black 1, grey 2, blue 1251, purple 1252, green 2000, red 2750 rpm). Trunk position starts at the origin and was recorded at 0.5 Hz. Prim ary direction of wind flow is from top to bottom of the page. B) Time series of wind speed (black) and m54 (red) at 2000 rpm.

PAGE 106

94 A B Figure 4.5. Trunk movement measured as an area of deflection (dya thinned tree number 3 before pruning). A) Transver se view of dya (x-axis is exaggerated for illustration). Cable extension transducers were located at 18, 30, 42, and 54 inches along the trunk (circles). Pr imary direction of wind flow was from left to right across the page. B) Time series of wind speed (black) and dya (red) at 2000 rpm recorded at 0.5 Hz.

PAGE 107

95 A B Figure 4.6. Four time series profiles of trunk movement at an elevation 54 inches above topmost root (m54 recorded at 0.5 Hz). A) A desirable wind profile (structural pruned tree number 1), B) an undesirable wind profile (lions tailed tree number 5), C) a wind profile that progresses from good to poor (raised tree number 5), and D) a wind profile th at progresses from poor to good (lion tailed tree number 6). Peaks represent a sequence of motor rpm used to generate wind speeds that was repeated for each pruning dose. Pruning dose increases in severity from left to right across the page.

PAGE 108

96 C D Table 4.6. Continued.

PAGE 109

97 A B Figure 4.7. Wind speed (A) trunk movement (m54) (B) at all pruning doses by date tested. Solid line is no foliage dry mass removed. Dash line is 15%, dot line is 30%, and dash-dot line is 45% targ eted dose level. Tree label is a combination of pruning type and tree within the pruning type. Pruning types are: lt = lions tailing, ra = raising, re = reduction, st = st ructural, and th = thinning (lt01 represents the fi rst tree that was lions tailed).

PAGE 110

98 A B Figure 4.8. Three dimensional scatterplots of trunk movement at an elevation 54 inches above topmost root: A) all measuremen ts within a pruning type and tree raised tree number 5, B) trunk moveme nt and wind speed averaged within a motor rpm. Plots show one of the be tter distributions of wind speed and pruning dose. Pyramids represent orthogonal levels of wind speed and pruning dose targeted to simplify statistical analysis.

PAGE 111

99 A B Figure 4.9. Three less desira ble distributions of wind speed and pruning dose (% foliage dry mass removed): A) A good distribut ion of wind speed but nothing in the low end of pruning dose raised tree number 2, B) a good distribution of wind speed but nothing in the high end of pruning dose lion tail tree number 2, C) a poor distributi on of both wind speed and pruning dose lion tail tree number 5. Pyramids represent orthog onal levels of wind speed and pruning dose targeted to simplify statistical analysis.

PAGE 112

100 C Figure 4.9. Continued.

PAGE 113

101 A Figure 4.10. A graphical summa ry of the procedure used to generate predicted response values (thinned tree number 4): A) Trunk movement measured 54 inches above topmost root (m54 averaged w ithin motor rpm) plotted against wind speed (averaged within motor rpm) a nd pruning dose (percent of foliage dry mass removed). Pyramids represent orthogonal levels of pruning dose and wind speed targeted to simplify statistical analysis. B) Response surface representing predicted m54 (m54 = 0.02998 wind + 0.0005 wind2 0.0004 wind dose, R2 = 0.986). C) Predicted levels of m54 (pm54) at orthogonal levels of pruning dose and wind speed. Predicted levels in C were used in the statistical analysis.

PAGE 114

102 B C Figure 4.10. Continued.

PAGE 115

103 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 01530456075 Wind speed (mph)pm54 (in.) Figure 4.11. Interaction between pruning t ype and wind speed (p<0 .0001) resulting from predicted average trunk move ment at an elevation 54 inches above topmost root (pm54). Vertical axis is the l east square means of pm54 adjusted using Tukeys method. Lines represent pruni ng types. Circles and solid line = thinned, triangles and dash line = redu ced, squares and dash dot line = raised, diamonds and dot line = lion tailed, and Xs and dash dot dot line = structural.

PAGE 116

104 0 0.5 1 1.5 2 2.5 3 3.5 01530456075 Wind speed (mph)pm54 (in.) Figure 4.12. Interaction between pruning dos e and wind speed (p<0. 0001) resulting from predicted average trunk move ment at an elevation 54 inches above topmost root (pm54). Vertical axis is the l east square means of pm54 adjusted using Tukeys method. Lines represent pruning dose levels. Circles and solid line = 0% foliage dry mass removed, triangles and dash line = 15%, squares and dash dot line = 30%, diamonds and dot line = 45%, and Xs and dash dot dot line = 60%.

PAGE 117

105 CHAPTER 5 DISCUSSION AND CONCLUSIONS Live oak trees were an appropriate speci men for aeromechanical investigation. Live oak leaves are stiff with acute bases a nd short petioles. Vogel (1989) demonstrated that leaves with similar construction had hi gh individual leaf drag coefficients that increased with increasing wind speeds. Hi ghrise oak leaves presumably have high individual drag coefficients, exemplified by the defoliation seen on raised tree number one in winds generated above 3000 motor rpm (> 50 mph). Whether results of pruning type and pruning dose would be more or le ss pronounced among trees where foliage had a low individual leaf drag coefficient; or wher e foliage reconfigured to reduce leaf drag is uncertain. However, effects of pruning type and pruning dose were only apparent at high wind speeds. Vogel (1989) demonstrated that at such wind speeds, drag coefficients declined for all leaves (except those that we re stiff with acute bases and short petioles), and especially for leaves that reconfigured, since leaves that rec onfigured did so more effectively at high wind speeds. Drag coefficients for clusters of leaves decreased for all leaf types, but less for stiff leaves with acute bases and short petioles than for other types. Thus, the effects of pruning type and pruni ng dose reported here would likely be less pronounced in other species based on the type of foliage alone. In addition to leaf characteristics, live oak foliage was well distributed throughout the canopy. Hoag et al. (1971) showed that distribution of foliage is a good measure of the location of the resultant drag force (locati on of the load if it were applied as a point load) acting on a limb. Pruning types differed in how they affected foliage distribution

PAGE 118

106 throughout the canopy. Therefore differences in trunk movement between pruning types might be attributed to differences in the locati on of the resultant drag force. Elevation of the resultant drag force was very different be tween raised and reduced trees, particularly at the highest pruning dose, yet trunk movement was statistically similar for both types at all wind speeds and all pruning doses. Rais ing and lions tailing likely elevated the location of the resultant drag force while re duction and structural likely lowered it. Thinning should have had the least effect on lo cation of the resultant drag force since the goal of thinning was to create an even distribution of foliage throughout the canopy without changing canopy dimensions. Howeve r, thinning was the only pruning type that was statistically different; trunk movement was greatest at the highest predicted wind speeds for thinned trees. This was likely b ecause canopy dimensions did not change for thinned trees while they di d for other pruning types. It was fortunate that thinning was the one pruning type th at differed from all others. The only significant difference in physical characteristics among unpruned trees assigned to pruning types (difference in elevation to vertical center of canopy) was between thinned and reduced trees. Elevation to ver tical center of canopy defined the length of the lever arm over which the force of wind acted. Thus it also affected location of the resultant drag force and could have been confounded with pruning type. Elevation to vertical center of canopy was lower for unprune d thinned trees than for unpruned reduced trees. Despite the shorter lever arm, tr unk movement was greater for unpruned thinned trees than for unpruned reduced trees. Ther efore, the only significant difference in physical characteristics among unpruned trees li kely did not alter the effect of pruning type, assuming trunk elasticity was uniform am ong pruning types. Additionally, because

PAGE 119

107 the interaction between pruning type and pruning dose was not statistically significant, differences between unpruned thinned and unpruned reduced trees did not likely alter effects of pruning dose either. Pruning dose primarily affected surface area projected into the wind and was applied as a measured decrease in percent fo liage dry weight. Dose was calculated as a percent of foliage dry weight because folia ge is the crux of ANSI (2001) definitions. However, others have reported that foliage is the most appropriate estimate of area in calculations involving aerodynamic dr ag of trees. Hoag et al (1971) went so far as to say: When there are leaves on the branches, the leaf area is the only significant area. Vollsinger et al. (2005) reporte d that leaves had more drag per unit frontal area than stems and trunk. Moore and Maguire (2004) and Roodbaraky et al. (1994) also agreed that aerodynamic drag of foliage was the most significant factor affecting external damping of tree sway. Foliage dry weight is highly correlated to foliage surface area (Ohara and Valappil 1995; Meadows and H odges 2002) so percent foliage dry weight was a good measure of projected surface area. Also, percent foliage dry weight was an effective way to measure the effect of pruning dose. Pruning dose levels were extremely variable among trees. For many trees, the highest predicted pruning dose levels were ex trapolated from actual data. For example, no lions tailed tree had 60% of its foliage removed during pruning. As a result, caution should be used when viewing comparisons at the higher pruning dos e levels. However, all but the lowest predicted pruning dose le vels are beyond biologically acceptable limits, so these extrapolations are not dangerously precarious. The same is not true for wind speed.

PAGE 120

108 Wind speeds generated by the airboat we re extremely variable among trees, especially for winds generate d at the highest motor rpm. For many trees, the highest predicted wind speeds were also extrapolated from actual data, but that was not the case for every tree in any comparison. While thes e results are credible, caution should be used when viewing comparisons at high wind speeds. We have no reasonable explanations for th e variability in generated wind speeds. Differences in elevation of the anemometer are the most likely explanation. Vertical wind profiles were not laminar and the anemomet er was not located at the same elevation for all trees. While some differences betw een wind speeds were seemingly explained by elevation of the anemometer, others were not. Elevation of the anemometer was not clearly responsible for variation in wind speeds. Other possible explanations for varia tion in wind speeds proved similarly inconclusive. There were no recognizable pa tterns within pruning type or day of year tested. There were also no no ticeable patterns attributable to time of day or a motor condition (e.g. a need to warm the motor up or a temporary motor fatigue). Correlations between generated wind speeds and temperature, time of day, and day of year were low (R2 = 0.06, 0.03, and 0.08 respectively for the highest rpm). Barometric pressure and relative humidity are known to affect moto r performance but neither was measured on site. Inconsistencies in wind speeds could have been due to operator error. This would likely be minimal because there were large increments between rpm used, and the motor was controlled primarily by one operator. The tachometer could have malfunctioned since it was connected to the motor via a 75 foot cable. But this was unlikely since tachometer and wiring were all new at the start of the expe rimental study and there were

PAGE 121

109 no noticeable inconsistencies in gauge output or in sound of the motor. Inconsistencies in wind speeds could also have been due to va riation in performance of an old motor. Wind fields generated by the airboat were smaller than tree canopies because winds did not dissipate much radially. The airboat was elevated so the entire width and lower half of a canopy were in the primary wind flow. However, the upper half of a canopy was likely outside the primary wind flow and the effects this had on results can not be known with certainty. Effects of a small wind field should have been most conspicuous on raised trees, but trunk movement of rais ed trees was not signifi cantly different than movement of lions tailed, reduced, or stru cturally pruned trees at any wind speed. Therefore, limited size of the wind field like ly did not affect comparisons among pruning types or doses. Researchers using wind tunnels and trucks to generate winds have met similar complications. Mayhead (1973) noted that lead ing shoots of several trees he tested were outside the primary wind flow. Vollsinger et al. (2005) and Rudnicki et al. (2004) trimmed foliage so trees fit in their wind t unnel, and only the upper portion of the main leader was harvested for testing. Trees ha ve been harvested for all wind tunnel studies using real plant tissue, and by Hoag et al. (1971) when a limb was mounted to a truck. An advantage of using the airboat is that trees did not have to be ha rvested for testing. Harvesting affects water relations in the tr ee; moisture content affects wood material properties (Ellis and Steiner 2002); therefore moisture content affects trunk movement. Comparisons among pruning type, pruning dose, and wind speed were based on trunk movement measured by displacement tr ansducers. Milne (1991) and Roodbaraky et al. (1994) used displaceme nt transducers to analyze natural sway frequency and

PAGE 122

110 damping of trees. Gardiner (1995) used di splacement transducers to correlate tree movement and wind loading. Gardiner pr eferred strain gaug es to displacement transducers because strain gauges had great er resolution of small displacements. Resolution of our transducers decreased propor tional to trunk movement so separations among pruning types and pruning doses at low wind speeds might have emerged if the displacement transducers had a higher resoluti on. Milne and Gardiner both recorded trunk movement at 10 Hz, and Roodbaraky el al. recorded at 4 Hz while our data acquisition system was only capable of reco rding at 0.5 Hz, which explains the sharp lines in our birds eye plot compared to t hose in Gardiners plots (see Gardiner 1995 Fig 2.1). James (personal communication) f ound that 40 Hz is optimal for recording measurements of tree movement using strain gauges. Sampling rate likely did not affect these results since wind speed and trunk move ment was averaged over a four minute time interval. Trunk movement provided a good representati on of the effects of pruning and wind speed on these trees. According to elastic beam theory, trunk movement is directly proportional to aerodynamic drag times the cube of the height to the po int of loading, and inversely proportional to three times the sec ond moment of area tim es the longitudinal Youngs modulus. Height to the point of lo ading was the same for almost all unpruned trees. Though there were statistical differences between elevations to vertical center of the canopy for unpruned reduced and unpruned thi nned trees, means only varied by 14.95 inches, 8.2 % of total canopy height. The s econd moment of area was assumed to be circular and constant for a ll trees even though trunk caliper was not measured on the day of testing. Longitudinal Youngs modulus wa s also assumed to be constant among trees

PAGE 123

111 tested. Efforts were made during tree sele ction to account for factors known to cause differences in longitudinal Youngs modulus, and trunks were assumed to possess similar material properties based on selection criteria. Unfortunately, efforts to determine longitudinal Youngs modulus using coupons an d whole trunks were inconclusive. By combining terms, trunk movement was ther efore directly proportional to aerodynamic drag times a constant. If pruning type, pr uning dose, or wind speed affected aerodynamic drag, the effect would be manifest in trunk movement. Of the affects tested, wind speed ha d the most significant effect on trunk movement, as was evident in individual tree regressions. This could have been anticipated since aerodynamic drag does not sc ale linearly with wind speed, as it does with surface area or drag coefficient, in the classical drag equation. Wind speed pruning type and wind speed pruning dose intera ctions were also significant. This too could have been anticipated since pruni ng affects both surface area and the drag coefficient, and past research has indicated that both surface area a nd the drag coefficient scale proportionally to some power of wi nd speed. Mayhead (1973) reported drag coefficients that decreased with increasing wi nd speed. Hoag et al. (1971) suggested that decreases in drag coefficients could be attr ibuted instead to decreases in surface area. Vollsinger et al. (2005) and R udnicki et al. (2004) separated changes in surface area from changes in drag coefficients by using wind sp eed specific frontal areas to calculate drag coefficients. Drag coefficients still decreased inverse to wind speed but not as sharply since decreases in frontal area were acc ounted for. Remaining decreases in drag coefficients might be accounted for by change s in skin friction but wind speed specific skin friction would be very di fficult to measure on a flexib le porous structure. Both

PAGE 124

112 studies reported that drag pe r unit frontal area increased pr oportionally to wind speed for all species tested. They also presented regr essions that illustrated a squared wind speed term gave better coefficients of determination than a linear wind speed term when using wind speed specific frontal area. Thus, even though changes in surface area and the drag coefficient affected aerodynamic drag on thei r trees, changes in wind speed had a greater affect than either of the two. It seems reasonable then that pr uning had limited effect on tree response relative to increasing wind speed. Interaction of pruning type and wind speed was seen in two ways. First, trunk movement increased significantly at each wind speed for all pruning types, averaged across pruning dose, except for structurally pr uned trees. With this exception, no pruning type effectively redu ced predicted aerodynamic drag, even at wind speeds up to 60 mph. Thus changes in tree aerodynamic drag do not justify use of one pruning type over another. We have no explanation for the response of structurally pruned trees except possibly small sample size. Two structurally pruned trees were included in the analysis. Additional trees would have increased the sample size resulting in more powerful statistical tests. Interaction of pruning type and wind speed was also seen among wind speeds. At high wind speeds trunk movement of thinned tree s was greater than that of trees assigned to other pruning types. This might be e xplained by the affect of pruning type on crown dimensions. The geometric dimensions of th inned trees did not change with pruning as they did with all other pruning types. Sin ce wind does not pass through a tree canopy in significant measure (Zhu, Matsuzaki et al 2000), thinned trees likely presented the largest surface area to wind flow. Another explanation could be a difference in whole

PAGE 125

113 canopy skin friction. Thinning increases ca nopy porosity and since wind does not pass through a tree canopy, a more porous ca nopy might translate into a rougher canopy surface increasing whole canopy skin friction. Bo th explanations are plausible and either would increase tree aerodynamic drag. Dimensions of lions tailed tree canopies al so changed very little with pruning. However, trunk movement of lions tailed trees was not different th an trunk movement of raised, reduced or structural trees. The combination of a small wind profile centered on a canopy where pruning removed all but the larges t branches may have allowed more wind through the canopy of lions tailed trees. Another expl anation is that branches may have become more flexible with lions tailing, increasing canopy streamlining which translated into less trunk movement. The former explan ation is more plausibl e since only the acute effects of pruning were evaluated. Th ere was not enough time between pruning and testing for additional growth to alter branch taper or branch wood ma terial properties. Effects of structural pruning may not have b een expressly manifest here since all trees were structurally pruned for years prior to te sting. Other studies that evaluated pruning have not compared pruning types only pr uning doses (Rudnicki, Mi tchell et al. 2004; Moore and Maguire 2005; Vollsin ger, Mitchell et al. 2005). Interaction of pruning dose and wind speed was manifest only after 30% of foliage was removed from a canopy. Vollsinger et al (2005) indicated that pruning experiments supported their proposal that a strong relationship exists be tween branch mass and drag, but they could not evaluate the effect of dose on streamlining because branches on the trees they used acted independently. Rudni cki et al. (2004) repor ted that effect of pruning dose was species specific. For red ce dar, drag per unit branch mass decreased

PAGE 126

114 with increasing pruning dose and wind speed. With hemlock drag per unit branch mass increased with increasing pruning dose up to wind speeds close to 25 mph then it decreased with increasing wind speeds. For lodgepole pine, drag per unit branch mass increased with increasing pruning dose and wind speed. In both studies, trees were raised by removing first 30% then 60% of total branch mass. Changes in drag per unit branch mass were only significant for the lodgepol e pine at 60% branch mass removed and 40 mph wind speeds. Moore and Maguire (2005) reported that raisi ng did not influence natural frequency of trees until 80% of bran ch mass was removed. In all three reports and in this study, pruning dose was not eff ective until pruning dose exceeded biologically acceptable limits. Biologically acceptabl e limits are defined by pruning effects on biological functions like light harvesting and carbon assimilation, water transport, and reception and transduction of environmental signals. Removing excess foliage affects partitioning of tree growth among foliage, fruit, stems, and roots which can have negative affects on tree health as well as tree structure (Zeng 2003). American National Standards Institute (Accredited Standards Committ ee A300 2001) recommends no more than 25% of foliage be removed with any one pruning. Additional results from this study were pr ocedural in nature. Dose was most efficiently and effectively determined as a visual estimate of foliage removed. Using anemometry to estimate density of the canopy was ineffective. A single measurement of trunk movement at the base of the canopy was as effective as multiple measurements along the clear trunk. Small differences in th e height of the lever arm (elevation to the vertical center of the canopy) were insignificant. Wind speeds were not reproduced

PAGE 127

115 consistently by calibrating wind speed to moto r rpm. While none of this was the primary goal of this research, all of it is useful fo r future work on similar sized landscape trees. An important caveat is that this study provides no information on the aerodynamic drag experienced by individual branches. It w ould be foolish to extr apolate these results to larger trees using the trees tested here to represent br anches or parts of a larger structure. Further testing is required to determine effects of pruning on the aeromechanical behavior of individual branch es when they are coupled as a continuous dynamic structure. It should also be noted that all studi es to date have evaluated relatively straight line winds appl ied perpendicular to tree growth. Here we used whole canopy aerodynamic drag, measured as the magnitude of trunk deflection, to make comparisons between pr uning type, pruning dose, and wind speed. Wind speed was the most significant factor influencing trunk movement at all wind speeds. Pruning reduced trunk movement at all wind speeds and the effect of pruning increased with increasing wind speed. Howeve r, pruning did not have a significant effect on trunk movement until the amount of foliage removed exceeded biologically acceptable limits. Thinning did not reduce trunk movement at high wind speeds ( 45 mph) as well as the other pruning types evaluated, while rais ing, reduction, lions tailing, and structural pruning all reduced trunk movement equally well at all wind speeds.

PAGE 128

116 CHAPTER 6 FUTURE WORK It is impossible to study the way a tree s mechanical structure responds to wind without wind. This experimental study showed that an airboat was a viable option for generating wind outdoors. Since inception of this pr oject, colleagues of Dr. Kurt Gurley at Florida International University have built a portable wind machine that generates a larger wind field and higher wind speeds. The machine is composed of two modular subunits that could be repeated to create a larg er wall of wind. Both subunits are mounted on a single trailer with a standard hitch so the wind can be taken to a tree in the landscape, rather than taking a tree to the wind by pl acing it in a wind tunnel or on a truck. Inconsistencies seen in wind generated by the airboat are minimized by controlling motor rpm with feedback from anem ometry. Opportunities created by such a machine are vast and it would be inappropriate to attempt their enumeration here. However, one opportunity is to repeat the re search described herein on sim ilar sized or larger trees. With an improved wind field it would be possi ble to determine what if any effect the narrow wind field had on these results. There are multiple ways to approach the relationship between trees and wind. The approach evaluated here was to reduce ca nopy drag by reducing surface area projected into the wind through pruning. While no pr uning type or pruning dose manifest an immediate benefit here, pruning was only evaluated for its acute effects on canopy drag. Pruning clearly effects tree growth in a predic table manner, therefore the chronic effects of pruning need to be considered as well. Apart from its short term effect on canopy

PAGE 129

117 drag, the enduring effects of pruning on tree structure need to be evaluated. Some information has been reported (Zeng 2003). Ho wever, more research should be designed to evaluate long term effects pruning type s and pruning doses have on tree form and construction. Tree construction is defined by the com position of individual tree parts and how those parts are joined together. Altering tree construction, strengtheni ng the structure, is another approach to the relationship between tr ees and wind. It has been shown that trees respond to external mechanical stimuli by producing reaction wood (K won, Bedgar et al. 2001) and wound wood (Kane and Ryan 2003) which are both stronger than normal wood. However, while anatomical features of the stronger wood are clearly defined (Barnett and Bonham 2004), signal reception an d transduction pathways that control its growth are not (Blancaflor 2002) Further research in this area will be facilitated by sequencing of the poplar genome (Brunner, Bus ov et al. 2004). There is less information defining how individual tree parts are joined together, although there is some (Gilman 2003). More work will be done in this area as individuals attempt to define how energy is dissipated through a tree (James 2003). Still, more in formation is need ed to develop cultural practices that promote stronger tree structure. This study was designed to pr ovide practical information to tree care professionals. Apart from efforts to understand trees and wi nd, more work needs to be conducted to evaluate current cultu ral practices and hypothetical solu tions to reduce wind damage in trees. There is far too much speculation by practitioners and not enough information to muffle opinion.

PAGE 130

118 Tree biomechanics is still a budding field. Current research on relationships between wind and trees is spread among various disciplines and it is becoming increasingly complex. Hopefully, future rese arch will take a structured course so that great strides will be made in a very short time.

PAGE 131

119 LIST OF REFERENCES Accredited Standards Committee A300 (2001). American National Standard for Tree Care Operations Tree, Shrub, and Ot her Woody Plant Maintenance Standard Practices (Pruning) Manchester, NH, National Arborist Association, Inc. Allen, J. R. L. (1992). "Trees and Their Re sponse to Wind Mid Flandrian Strong Winds, Severn Estuary and Inner Bristol Channe l, Southwest Britain." Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 338 (1286): 335-364. ANSI (2001). ANSI A300 (P art 1) 2001 Pruning Manchester, NH, National Arborist Association, Inc. Bailey, S. J., Ed. (2000a). Standard Test Me thods for Small Clear Specimens of Timber. ASTM D143-94 (Reapproved 2000). Annual Book of ASTM Standards. West Conshohocken, PA, USA, ASTM Internati onal (formerly American Society for Testing and Materials). Bailey, S. J., Ed. (2000b). Standard Test Met hods of Static Tests of Wood Poles. ASTM D1036-99. Annual Book of ASTM Standards. West Conshohocken, PA, USA, ASTM International (formerly American Society for Testing and Materials). Barnett, J. R. and V. A. Bonham (2004). "Ce llulose microfibril angl e in the cell wall of wood fibres." Bi ological Reviews 79 (2): 461-472. Baskin, T. I. (2001). "On the alignment of cellu lose microfibrils by co rtical microtubules: a review and a model." Protoplasma 215: 150-171. Beismann, H., F. Schweingruber, et al. ( 2002). "Mechanical properties of spruce and beech wood grown in elevated CO2." Trees-Structure and Function 16 (8): 511518. Bertram, J. E. A. (1989). "Size-dependent diffe rential scaling in bran ches: the mechanical design of trees revisited." Trees-Structure and Function 4: 241-253. Blackburn, P., J. A. Petty, et al. (1988). "An Assessment of the Static and Dynamic Factors Involved in Windthrow." 61 (1): 29-43. Blancaflor, E. B. (2002). "The cytoskeleton and gravitropism in higher plants." Journal of Plant Growth Regulation 21 (2): 120-136.

PAGE 132

120 Brickell, C. and D. Joyce (1996). The Am erican Horticultural Society Pruning and Training: A Fully Illustrated Plant-by-Plant Manual London, Dorling Kindersley Limited. British Forestry Commission (2005). Tree stability and climate, http://www.forestry.gov.uk/website/fo restresearch.nsf/byunique/INFD-639A92 last accessed November 2005 Brown, G. E. (1995). The Pruning of Trees, Shrubs, and Conifers Portland, OR, U.S.A., Timber Press, Inc. Brchert, F., G. Becker, et al. (2000). "The mechanics of Norway spruce Picea abies (L.) Karst : mechanical properties of standi ng trees from different thinning regimes." Forest Ecology and Management 135 (1-3): 45-62. Brunner, A. M., V. B. Busov, et al. (2004). "Poplar genome sequence: functional genomics in an ecologically dominant plant species." Trends in Plant Science 9 (1): 49-56. Cannell, M. G. R. and J. Morgan (1987). "Y oung's modulus of sections of living branches and tree trunks." Tree Physiology 3: 355-364. Coder, K. D. (2000). Estimating wind forces on tree crowns, http://www.forestry.uga.e du/warnell/service/library last accessed June 2003 Coutts, M. P. (1983). "Root Architecture and Tree Stability." Plant and Soil 71 (1-3): 171-188. Coutts, M. P. (1986). "Components of Tree Stab ility in Sitka Spruce on Peaty Gley Soil." Forestry 59 (2). Coutts, M. P. and J. Grace, Eds. (1995). Wind and Trees Edinburgh, Scotland, Cambridge University Press. Cremer, K. W., C. J. Borough, et al. ( 1982). "Effects of Stocking and Thinning on Wind Damage in Plantations." New Zeal and Journal of Forestry Science 12 (2): 244268. Denny, M. W. (1994). "Extreme Drag For ces and the Survival of Wind-Swept and Water-Swept Organisms." Jour nal of Experimental Biology 194: 97-115. Duryea, M. L., G. M. Blakeslee, et al. (1996). "Wind and Trees: A Survey of Homeowners After Hurricane Andr ew." Journal of Arboriculture 22 (1): 44-50. Ellis, S. and P. Steiner (2002). "The Behavi our of Five Wood Sp ecies in Commpression." Iawa Journal 23 (2): 201-211.

PAGE 133

121 Ennos, A. R. (1999). "The aerodynamics and hydrodynamics of plants." 202 (23): 32813284. Evans, R. and J. Ilic (2001). "Rapid predic tion of wood stiffness fr om microfibril angle and density." Forest Products Journal 51: 53-57. Ezquerra, F. J. and L. A. Gil (2001). "Wood an atomy and stress distribution in the stem of Pinus pinaster Ait." Investigacion Agraria, Sistemas y Recursos Forestales 10 (1): 165-177. Finnigan, J. J. and Y. Brunet (1995). Turbulent airflow in forests on flat and hilly terrain. Wind and Trees M. P. Coutts and J. Grace. Cambridge, Cambridge University Press: 3-40. Fourcaud, T. and P. Lac (2003). "Numerical modelling of shape regulation and growth stresses in trees I. An incremental static finite element formulation." TreesStructure and Function 17 (1): 23-30. Fraser, A. I. (1962). "The Soil and Roots as Factors in Tree Stability." Forestry 35: 117127. Fraser, A. I. (1964). "Wind Tunnel and Othe r Related Studies on Coniferous Trees and Tree Crops." Scottish Forestry 18: 84-92. Gardiner, B. A. (1995). The interactions of wind and tree movement in forest canopies. Wind and Trees M. P. Coutts and J. Grace. Cambridge, Cambridge University Press: 41-59. Gardiner, B. A. and C. P. Quine (2000). "M anagement of forests to reduce the risk of abiotic damage a review with particular reference to the effects of strong winds." Forest Ecology and Management 135 (1-3): 261-277. Gilman, E. F. (2002). An Illustrated Guide to Pruning Albany, NY, U.S.A., Delmar, a division of Thomson Learning, Inc. Gilman, E. F. (2003). "Branch-to-stem diameter ratio affects strength of attachment." Journal of Arboriculture 29 (5): 291-193. Gilman, E. F. and S. J. Lilly (2002). Best Management Practices: Tree Pruning Champaign, IL, US, International Society of Arboriculture. Gordon, J. E. (1981). Structures : or Why Things Don't Fall Down London, Da Capo Press, Inc. Grace, J. (1977). Plant Response to Wind London, Academic Press. Graham, A. W. (1990). "England Before and After the Hurricane of 1987." Journal of Arboriculture 16 (10): 269-274.

PAGE 134

122 Green, D. W., J. E. Winandy, et al. (1 999). Mechanical Properties of Wood. Wood handbook Wood as an engineering mate rial. Gen. Tech. Rep. FPL-GTR-113 Madison, WI, USA, U.S. Department of Agriculture, Forest Service, Forest Products Laboratory: 465. Ham, D. L. and R. C. Rowe (1990). "Tree Devastation by Hurricane Hugo." Journal of Arboriculture 16 (1): ix-x. Harris, R. W., J. R. Clark, et al. (2004). Arboriculture: Integrated Management of Landscape Trees, Shrubs, and Vines Upper Saddle River, New Jersey, Prentice Hall. Hedden, R. L., T. S. Fredericksen, et al. (1995). "Modeling the Eff ect of Crown Shedding and Streamlining on the Survival of L oblolly-Pine Exposed to Acute Wind." Canadian Journal of Forest Research-Re vue Canadienne De Recherche Forestiere 25 (5): 704-712. Hoag, D. L., R. B. Fridley, et al. (1971) "Experimental Measurement of Internal and External Damping Properties of Tree Li mbs." Transactions of the American Society of Agricultural Engineers 14: 20-28. James, K. (2003). "Dynamic Loading of Trees." Journal of Arboriculture 29 (3): 165-171. Kane, B. C. P. and I. Ryan, H. Dennis P. (2003). "Examining formulas that assess strenght loss due to decay in trees : woundwood toughness improvement in red maple ( Acer rubrum )." Journal of Arboriculture 29 (4): 209-217. Kaufmann, M. R. and C. A. Troendle ( 1981). "The Relationship of Leaf-Area and Foliage Biomass to Sapwood Conducting Area in 4 Subalpine Forest Tree Species." Forest Science 27 (3): 477-482. Kemper, J. D. (1968). "Large Deflections of Tapered Cantilever Beams." International Journal of Mechanical Sciences 10: 469-478. Kerzenmacher, T. and B. A. Gardiner (1998) "A mathematical model to describe the dynamic response of a spruce tree to th e wind." Trees-Structure and Function 12 (6): 385-394. Kwon, M., D. L. Bedgar, et al. (2001). "I nduced compression wood formation in Douglas fir (Pseudotsuga menziesii) in microgravity." Phytochemistry 57 (6): 847-857. Lang, S. and S. B. Editors (1998). Pruning Menlo Park, CA, US, Sunset Books Inc. Lee, X. (2000). "Air motion within and above forest vegetation in non-ideal conditions." Forest Ecology and Management 135 (1-3): 3-18.

PAGE 135

123 Leiser, A. T. and J. D. Kemper (1973). "Ana lysis of Stress Distri bution in the Sapling Tree Trunk." Journal of the American Society of Horticultural Science 98 (2): 164-170. Lilly, S. J., J. R. Clark, et al. ( 1993). Arborists' Certification Study Guide Champaign, IL, US, International Society of Arboriculture. Luley, C. J., S. Sisinni, et al. (2002). "The Effect of Pruni ng on Service Requests, Branch Failures, and Priority Maintenance in the City of Rochester, New York, U.S." Journal of Arboriculture 28 (3): 137-143. Matheny, N. P. and J. R. Clark (1994). A Photographic Guide to the Evaluation of Hazard Trees in Urban Areas Urbana, Illinois USA, In ternational Society of Arboriculture. Mattheck, C., K. Bethge, et al (1993). "Safety Factors in Tr ees." Journal of Theoretical Biology 165: 185-189. Mattheck, C. and H. Breloer (1994). The Body Language of Trees: a Handbook for Failure Analysis London, Department of the E nvironment, Transport and the Regions. Mattheck, C. and H. Kubler (1997). W ood the Internal Optimization of Trees Berlin, Springer-Verlag. Mattheck, C. and U. Vorberg (1991). "The Bi omechanics of Tree Fork Design." Botanica Acta 104 (5): 399-404. Mayhead, G. J. (1973). "Some Drag Coefficien ts for British Forest Trees Derived from Wind Tunnel Studies." Agricultural Meteorology 12: 123-130. McDowell, N., H. Barnard, et al. (2002). "T he relationship between tree height and leaf area: sapwood area ratio." Oecologia 132 (1): 12-20. McMahon, T. A. (1973). "The Mechan ical Design of Trees." Science 233: 92-102. Meadows, J. S. and J. D. Hodges (2002). "Sap wood area as an estimator of leaf area and foliar weight in cherrybark oak and green ash." Forest Science 48 (1): 69-76. Medhurst, J. L. and C. L. Beadle (2002). "Sapwood hydraulic conductiv ity and leaf area sapwood area relationships following thinni ng of a Eucalyptus nitens plantation." Plant Cell and Environment 25 (8): 1011-1019. Milne, R. (1991). "Dynamics of swaying of Picea sitchensis ." Tree Physiology 9: 383399.

PAGE 136

124 Mitchell, S. J. (2000). "Stem growth respons es in Douglas-fir a nd Sitka spruce following thinning: implications for assessi ng wind-firmness." Forest Ecology and Management 135 (1-3): 105-114. Moore, J. R. (2000). "Differences in ma ximum resistive bending moments of Pinus radiata trees grown on a range of soil types." Forest Ecology and Management 135 (1-3): 63-71. Moore, J. R. and B. A. Gardiner (2001). "Relative windfirmness of New Zealand-grown Pinus radiata and Douglas-fir: a preliminary i nvestigation." New Zealand Journal of Forestry Science 31 (2): 208-223. Moore, J. R. and D. A. Maguire (2002). The Mechanics of Trees Under Wind Loading Tree Structure and Mechanics Conference Proceedings: How Trees Stand Up and Fall Down, Savannah, Georgia, U.S., The International Society of Arboriculture and The Southern Chapter, Intern ational Society of Arboriculture. Moore, J. R. and D. A. Maguire (2004). "Nat ural sway frequencies and damping ratios of trees: concepts, review and synthesis of previous studies." Trees-Structure and Function 18 (2): 195-203. Moore, J. R. and D. A. Maguire (2005). "Nat ural sway frequencies and damping ratios of trees: influence of crown structur e." Trees-Structure and Function 19 (4): 363373. Moore, J. R. and C. P. Quine (2000). "A comp arison of the relative risk of wind damage to planted forests in Border Forest Pa rk, Great Britain, and the Central North Island, New Zealand." Forest Ecology and Management 135 (1-3): 345-353. Morgan, J. and M. G. R. Cannell (1987). "Structural Analysis of tree trunks and branches: tapered cantilever beams subj ect to large deflec tions under complex loading." Tree Physiology 3: 365-374. Morgan, J. and M. G. R. Cannell (1994). "Sha pe of tree stems a re-examination of the uniform stress hypothesis." Tree Physiology 14: 49-62. Niklas, K. J. (1992). Plant Biomechanics: an engineering approach to plant form and function Chicago, University of Chicago Press. Niklas, K. J. (1994). "Interspecific Allometr ies of Critical Buckling Height and Actual Plant Height." American Journal of Botany 81 (10): 1275-1279. Niklas, K. J. (1997a). "Mechanical propertie s of black locust (R obinia pseudoacacia L) wood. Size and age-dependent variations in sapand heartwood." Annals of Botany 79 (3): 265-272.

PAGE 137

125 Niklas, K. J. (1997b). "Mechanical propertie s of black locust (Robinia pseudoacacia) wood: Correlations among elastic and r upture moduli, proportional limit, and tissue density and specific gravity." Annals of Botany 79 (5): 479-485. Niklas, K. J. (1997c). "Sizeand age-depende nt variation in the pr operties of sapand heartwood in black locust (Robinia ps eudoacacia L)." Annals of Botany 79 (5): 473-478. Niklas, K. J. (1998). "The influence of grav ity and wind on land plant evolution." Review of Palaeobotany and Palynology 102 (1-2): 1-14. Niklas, K. J. (1999). "Changes in the factor of safety within the superstructu re of a dicot tree." American Journal of Botany 86 (5): 688-696. Niklas, K. J. (2000a). "Computing factors of safety against wind-induced tree stem damage." Journal of Experimental Botany 51 (345): 797-806. Niklas, K. J. (2000b). "The evolution of plan t body plans A biomech anical perspective." Annals of Botany 85 (4): 411-438. Niklas, K. J. and H. C. Spatz (2000). "Windinduced stresses in cherry trees: evidence against the hypothesis of c onstant stress levels." Tr ees-Structure and Function 14 (4): 230-237. Ohara, K. L. and N. I. Valappil (1995). "Sapwood Leaf-Area Pr ediction Equations for Multi-Aged Ponderosa Pine Stands in Western Montana and Central Oregon." Canadian Journal of Forest Research-Re vue Canadienne De Recherche Forestiere 25 (9): 1553-1557. Oliver, H. R. and G. J. Mayhead (1974). "Win d Measurements in a Pine Forest During a Destructive Gale." Forestry 47 (2): 185-195. Ott, R. L. and M. Longnecker (2001). An In troduction to Statistical Methods and Data Analysis, Fifth Edition Pacific Grove, CA, USA, Duxbury. Papesch, A. J. G., J. R. Moore, et al. (1997). "Mechanical Stability of Pinus radiata Trees at Eyrewell Forest Investigated Using Static Tests." New Zealand Journal of Forestry Science 27 (2): 188-204. Peltola, H., B. A. Gardiner, et al. (2000). "Wind and other abiotic risks to forests Introduction." Forest Ecology and Management 135: 1-2. Peltola, H., S. Kellomaki, et al. (2000). "M echanical stability of Scots pine, Norway spruce and birch: an analysis of treepulling experiments in Finland." Forest Ecology and Management 135: 143-153.

PAGE 138

126 Peltola, H., M. L. Nykanen, et al. (1 997). "Model computations on the critical combination of snow loading and windspeed for snow damage of Scots pine, Norway spruce and Birch sp. at stand edge." Forest Ecology and Management 95 (3): 229-241. Petty, J. A. and C. Swain (1985). "Factors Influencing Stem Breakage of Conifers in High Winds." Forestry 58 (1): 75-84. Petty, J. A. and R. Worrell (1981). "Stability of Coniferous Tree Stems in Relation to Damage by Snow." Forestry 54 (2): 115-128. Rebertus, A. J. and A. J. Meier (2001). "B lowdown dynamics in oak-hickory forests of the Missouri Ozarks." Journal of the Torrey Botanical Society 128 (4): 362-369. Robbins, K. (1986). How to Recognize and Reduce Tree Hazards in Recreation Sites (NA-FR-31). Northeastern Area, United States Department of Agriculture, Forest Service. Rogers, R. and T. M. Hinckley (1979). "F oliar Weight and Area Related to Current Sapwood Area in Oak." Forest Science 25 (2): 298-303. Roodbaraky, H. J., C. J. Baker, et al. (1994). "Experimental-Observations of the Aerodynamic Characteristics of Uban Trees." 52 (1-3): 171-184. Roots Plus Growers Associ ation of Florida (2005). http://rootsplusgrowers.org last accessed December 2005 Ruck, B., C. Kottmeier, et al., Eds. (2003). Proceedings of the In ternational Conference Wind Effects on Trees Karlsruhe, Germany, University of Karlsruhe. Rudnicki, M., S. J. Mitchell, et al. (2 004). "Wind tunnel measurements of crown streamlining and drag relationships for three conifer species." 34 (3): 666-676. Ryan, M. G. and J. Yoder (1997). "Hydraulic Limits to Tree Height and Tree Growth." BioScience 47 (4): 235-242. Salvadori, M. (1980). The Strength of Architecture: Why Buildings Stand Up New York, W. W. Norton & Company, Inc. Saunderson, S. E. T., A. H. England, et al (1999). "A dynamic model of the behaviour of sitka spruce in high winds." J ournal of Theoretical Biology 200 (3): 249-259. Savidge, R. A. (1996). "Xylogenesis, genetic and environmental regulation A review." Iawa Journal 17 (3): 269-310. Schuler, R. T. and H. D. Bruhn (1973). "Str ucturally Damped Timoshenko Beam Theory Applied to Vibrating Tree Limbs." Transa ctions of the American Society of Agricultural Engineers : 886-889.

PAGE 139

127 Smiley, E. T. and J. C. Bones (2000). NAA Pocket Guide: Identifying Hazard Trees Manchester, NH, National Arborist Association. Smiley, E. T. and K. D. Coder, Eds. ( 2002). Tree Structure and Mechanics Conference Proceedings: How Trees Stand Up and Fall Down Champaign, IL, USA, International Society of Arboriculture. Sommerville, A. (1979). "Root An chorage and Root Morphology of Pinus radiata on a Range of Ripping Treatments." New Z ealand Journal of Forestry Science 9 (3): 294-315. Spatz, H. C. and F. Bruechert (2000). "B asic biomechanics of self-supporting plants: wind loads and gravitationa l loads on a Norway spruce tree." Forest Ecology and Management 135 (1-3): 33-44. Spatz, H. C., L. Kohler, et al. (1999). "Mechanical behaviour of plant tissues: Composite materials or structures?" J ournal of Experimental Biology 202 (23): 3269-3272. Talkkari, A., H. Peltola, et al. (2000). "Integratio n of component models from the tree, stand and regional levels to assess the ri sk of wind damage at forest margins." Forest Ecology and Management 135: 303-313. Ter-Mikaelian, M. T. and M. D. Korzukhin (1997). "Biomass equations for sixty-five North American tree species." Forest Ecology and Management 97 (1): 1-24. Trachet, A. (2005). "Examining the Variation in Modulus of Elasticity of Live Oak ( Quercus virginiana Highrise) Tree Trunks." Univ ersity of Florida Journal of Undergraduate Research 7 (1). Vogel, S. (1989). "Drag and Reconfigurat ion of Broad Leaves in High Winds." 40 (217): 941-948. Vogel, S. (1994). Life in Moving Fl uids, The Physical Biology of Flow Princeton, N.J., USA, Princeton University Press. Vogel, S. (1995). "Twist-to-Bend Ratios of Wo ody Structures." Jour nal of Experimental Botany 46 (289): 981-985. Vollsinger, S., S. J. Mitchell, et al. (2005). "Wind tunnel measurements of crown streamlining and drag relationships fo r several hardwood species." Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere 35 (5): 1238-1249. Waring, R. H., P. E. Schroede r, et al. (1982). "Application of the Pipe Model-Theory to Predict Canopy Leaf-Area." Canadian J ournal of Forest Research-Revue Canadienne De Recherche Forestiere 12 (3): 556-560.

PAGE 140

128 Westphal, L. M. (2003). "Urban Greening a nd Social Benefits: A Study of Empowerment Outcomes." Journal of Arboriculture 29 (3): 137-147. White, D. A. (1993). "Relationships between Foliar Number and the Cross-Sectional Areas of Sapwood and Annual Rings in Red Oak (Quercus-Rubra) Crowns." Canadian Journal of Forest Research-Re vue Canadienne De Recherche Forestiere 23 (7): 1245-1251. Whittier, J., D. Rue, et al. (1995). "Urban tree residues: results of the first national inventory." Journal of Arboriculture 21 (2): 57-62. Wood, C. J. (1995). Understanding wind forces on trees. Wind and Trees M. P. Coutts and J. Grace. Cambridge, Cambridge University Press: 133-164. Zeng, B. (2003). "Aboveground biomass partiti oning and leaf development of Chinese subtropical trees following pruning. Forest Ecology and Management 173: 135144. Zhu, J., T. Matsuzaki, et al. (2000). "Wi nd speeds within a single crown of Japanese black pine ( Pinus thunbergii Parl.)." Forest Ecology and Management 135: 19-31.

PAGE 141

129 BIOGRAPHICAL SKETCH Scott Alan Jones is the second in a family of eleven children. He was born May 10, 1972, to Phillip R. Jones and Gwen Stephenson J ones. Scott spent most of his formative years in southern Idaho working on farms a nd playing outdoors, which led him to pursue a career in horticulture. An undergraduate advi sor at the University of Idaho, Dr. Robert R. Tripepi, and the director of the University of Idaho Arboretum, Dr Richard J. Naskali, were jointly responsible for Scotts interest in arboriculture. After graduating from the University of Idaho in May 2000 with a Bachelor of Science in plant science, Scott went to work for Ryan Lawn and Tree (RLT) in Ov erland Park, KS. With RLT he was trained as a climber, crew chief, and department coordinator; and became acquainted with many of the day to day issues arbor ists face. Scott met Dr. Edwa rd F. Gilman at the annual conference of the Kansas Arborist Associ ation in 2002 and afte r exchanging a few emails, he accepted a position as a graduate resear ch assistant at the University of Florida in the Department of Environmental Horticultu re. After graduation, Scott will work as an arborist representative for The F. A. Bartlett Tree Expert Company in Cincinnati, OH.


Permanent Link: http://ufdc.ufl.edu/UFE0013321/00001

Material Information

Title: Effect of Pruning Type, Pruning Dose, and Wind Speed on Tree Response to Wind Load
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013321:00001

Permanent Link: http://ufdc.ufl.edu/UFE0013321/00001

Material Information

Title: Effect of Pruning Type, Pruning Dose, and Wind Speed on Tree Response to Wind Load
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013321:00001


This item has the following downloads:


Full Text












EFFECT OF PRUNING TYPE, PRUNING DOSE, AND
WIND SPEED ON TREE RESPONSE
TO WIND LOAD















By

SCOTT ALAN JONES


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2005








































Copyright 2005

by

Scott Alan Jones




























To tree care professionals in all their varieties.















ACKNOWLEDGMENTS

This project was beyond my individual capacity. I would be ungrateful if I did not

acknowledge the many creative minds and helping hands that contributed to its

completion. I want to first thank Dr. Ed Gilman, my committee chair, who provided the

sure foundation for the work that was done. The project clearly would not have been

possible without his interest and support. He gave me liberty to manage the project as I

saw fit and was invariably patient with my personality and demands. He was a fine

mentor and will remain an excellent example and friend.

My graduate committee was better than I could have imagined. Dr. Perry Green

provided a desperately needed engineering dimension to the committee. I want to thank

him for his patient tutoring as I spent many hours at his desk or in his lab designing and

testing various parts of the experiment, the lion's share of which are not even represented

in this publication. I want to thank Dr. Richard Beeson for his limitless generosity and

invaluable experience. The data acquisition system would not have been built or

functional without him. And I want to thank Dr. Jason Grabosky for his loyalty and

enthusiastic moral support. I always found his door open despite the geographic distance.

Every member of my committee was exceptional and I will forever remember their

instruction and care.

I should next thank the benefactors that funded the project. I want to thank Dr.

Gilman again here. He and the Department of Environmental Horticulture at the

University of Florida provided a research assistantship that allowed me to work on a









project of my choice. The Tree Research and Education Endowment Fund awarded the

project a Hyland R Johns Grant with which we purchased much of the needed equipment

and supplies. Marshall Tree Farm donated the trees that were used and they were kind

enough to let me visit their farm and hand-pick the trees I wanted. Rinker Materials

donated 5 yards of concrete, with delivery, from which the pier walls were built. Sundry

odds and ends were donated from Dr. Green's and Dr. Beeson's other research interests.

A few other people deserve special recognition here. Chris Harchick, the manager

of the Environmental Horticulture Teaching Unit, was my right arm throughout much of

the project. The dirty work would not have been done without him. Alison Trachet, an

undergraduate student in civil and coastal engineering, collaborated on the efforts to

determine the Young's Modulus of the trunks and assisted in almost all the data

collection. Chuck Broward, laboratory manager in civil and coastal engineering,

facilitated the Young's Modulus work and was always generous with the resources at his

disposal. Dr. Kurt Gurley, an associate professor in civil and coastal engineering,

measured the downwind wind speed profile generated by the airboat and spent several

hours teaching me about wind and how to manage my data. Dr. Kenneth Portier, an

associate professor in the Institute of Food and Agricultural Sciences Statistical

Consulting Unit, and Marinela Campanu, one of Dr. Portier's graduate assistants,

provided needed assistance with the statistical analysis. Justin Sklaroff, Patricia Gomez,

Maria (Pili) Paz, Amanda Bisson, Ryan Yadav, and Jon Martin all provided extra hands

for some of the mundane manual labor. I thank them all.

Finally, I want to thank my family and friends for their love, attention, and support

despite my preoccupation with AMELIA.
















TABLE OF CONTENTS



A C K N O W L E D G M E N T S ................................................................................................. iv

LIST OF TABLES ............. ... ........ .................. .............. viii

LIST OF FIGURES ......... ....... .................... .......... ....... ............ ix

ABSTRACT .............. .................. .......... .............. xi

CHAPTER

1 IN TR O D U C TIO N ......................................................................... .... .. ........

2 L ITER A TU R E R E V IEW .................................................................... ....................3

S ig n ifican ce ....................................................... 3
H history ............................................................................ . 4
E engineering P rinciples................................................................... ................... 5
Scientific A approach ....................................................... 6
W in d F o rc e s .................................. ..................................................... .... 7
D rag coefficients ....................................................... 7
W ind speed ................................................................ ........... .. 9
T ree R resistance ................................................................................. ......... ....11
B ending and w ind snap ........................................... ........................... 12
W ind throw ...............................................................................................15
T ree D y n am ics ................................................................................ 16
P ru n in g ................................................................................ 17
D o se ................... ...................1...................9..........
C o n c lu sio n s ............................................................................................................ 2 1

3 M ATERIALS AND M ETHODS ....................................................... 23

T ree Selection ................................................................................................... ........ 23
E xperim mental D design .............................................................25
P running T ype .................................................................. 25
L ion 's T ailing ................................................................. ............ 26
R raising ...................................................................................................... .......27
R e d u ctio n .....................................................................................2 8
S tru ctu ral ....................................................... 2 9









T inning ................................................................... 29
P running D ose ................................................................... 30
W ind Speed............................................. 30
T runk D election .......... ..... .............................................................. .......... ....... 33
E xperim ental Procedure........................................... ................... ............... 34
Statistical A n aly sis............. .... ........................................................ .. ... ... .......36

4 R E S U L T S .............................................................................6 5

Randomization .......... .. ........... ..... ...............65
P running D ose ................................................................... 65
W ind Speed............................................. 67
R response .......... .. .............................................................................................68
Analytical Approach .................................. ..... ............... 70

5 DISCUSSION AND CONCLUSIONS ................................105

6 FUTURE W ORK ............... ............................ ......... 116

LIST O F R EFEREN CE S ................................119............................

BIOGRAPHICAL SKETCH ................................. ........................ ...............129
















LIST OF TABLES


Table page

3-1. Measurements used to determine physical similarity of trees tested........................37

3-2. Downwind profile of wind speeds generated by an airboat. ................. ................39

3-3. Calibration of motor rpm to wind speed .... ........... .......................................40

3-4. Wind speeds recorded during a road test............................. ................41

4-1. Foliage and stem weight for 13 trees harvested to quantify pruning dose ...............74

4-2. Percent foliage dry weight, stem dry weight, and total dry weight removed with
each pruning dose .................................. ....... .. ................. 75

4-3. W ind speeds (mph) recorded during testing .......... ................................................79

4-4. Wind speed (mph) by pruning type and motor rpm. ............................................79

4-5. Wind speed, trunk movement (m54), and deflected area (dya) at 0 pruning dose
and 2750 rpm .................................................... ................. 80

4-6. Regression coefficients and R2 values generated using all measurements ..............81

4-7. Regression coefficients and R2 values generated using averages............................83

4-8. ANOVA of predicted trunk movement (p_avm54) and predicted deflected area
(p av d y a).............................. ......................................................... ............... 8 5

4-9. Least squares means of predicted average trunk movement (p_avm54) due to
interaction of pruning type and wind speed type by wind speed ........................86

4-10. Least squares means of predicted average trunk movement (p_avm54) due to
interaction of pruning type and wind speed wind speed by type ........................86

4-11. Least squares means of predicted average trunk movement (p_avm54) due to
interaction of pruning dose and wind speed dose by wind speed .......................87

4-12. Least squares means of predicted average trunk movement (p_avm54) due to
interaction of pruning dose and wind speed wind speed by dose. ........................87
















LIST OF FIGURES


Figure page

2.1. Wind forces affecting trees (modified from Grace (1977)).......................................22

3.1. Three examples of lion's tail pruning taken from urban landscapes .......................42

3.2. Example of geometrically defined lion's tail pruning. ............................................43

3.3. Example of visually defined lion's tail pruning......................................................44

3.4. Two examples of raising pruning type taken from urban landscapes......................45

3.5. Example of geometrically defined raised pruning...............................................46

3.6. Example of visually defined raised pruning. .................................. .................47

3.7. Two examples of reduction pruning type taken from urban landscapes. .................48

3.8. Example of geometrically defined reduction pruning. ............................................49

3.9. Example of visually defined reduction pruning ......................................................50

3.10. Two examples of structural pruning type taken from urban landscapes before
(A and C) and after (B and D) respectively. ................................. ..................51

3.11. Exam ples of structural pruning ........................................ ........................... 52

3.12. Example of thinning taken from urban landscapes before (A) and after (B)........53

3.13. E xam ple of thinning ........ .............................................................. .. .... ........54

3.14. Airboat used to generate wind: A) side view and B) rear view ............................. 55

3.15. Calibration curves of the mean wind speeds as measured by Campbell (West
and East) and Young (Gurley) anemometers. .................................. ...............56

3.16. The four South cable extension transducers (CET) ..............................................57

3.17. Southern cable extension transducer (CET) calibration curves............................. 58

3.18. Northern cable extension transducer (CET) calibration curves............................. 59









3.19. Schematics (bird's-eye (A) and profile (B) views of cable extension transducer
(CET) and anemometer positions.................................................. ......... ...... 60

3.20. Cable extension transducer calculations (these are for one specific height
represented in inches by subscript #). ........................................... ............... 61

3.21. Apparatus used to determine longitudinal Young's modulus..............................62

3.22. Apparatus used to fix the trunk and football of trees during testing.....................63

3.23. Data acquisition system used in the field .... ........... ..................................... 64

4.1. Pruning dose represented as percentage of foliage dry mass removed ......................88

4.2. Vertical profile of average generated wind speeds ............... ..................................90

4.3. Four profiles of wind speeds generated by the airboat during testing.....................91

4.4. Trunk movement measured at an elevation 54 inches above topmost root ..............93

4.5. Trunk movement measured as an area of deflection. ..............................................94

4.6. Four time series profiles of trunk movement at an elevation 54 inches above
to p m o st ro o t ...................................... ............................... ................ 9 5

4.7. Wind speed (A) trunk movement (m54) (B) at all pruning doses by date tested... ...97

4.8. Three dimensional scatterplots of runk movement at an elevation 54 inches above
to p m o st ro o t ...................................... ............................... ................ 9 8

4.9. Three less desirable distributions of wind speed and pruning dose...........................99

4.10. A graphical summary of the procedure used to generate predicted response
v alu e s ...................................... .................................................. 10 1

4.11. Interaction between pruning type and wind speed ...............................................103

4.12. Interaction between pruning dose and wind speed ............................................. 104















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

EFFECT OF PRUNING TYPE, PRUNING DOSE, AND
WIND SPEED ON TREE RESPONSE
TO WIND LOAD

By

Scott Alan Jones

December 2005

Chair: Edward F. Gilman
Major Department: Environmental Horticulture

Three to four inch caliper, clonally propagated live oak trees (Quercus virginiana

'QVTIA' PP 11219, Highrise) were used to test the effect of five pruning types and

four pruning doses on trunk movement at four wind speeds. Pruning types evaluated

were lion's tailing, raising, reduction, structural, and thinning. Pruning doses were 15,

30, 45, and 60 percent foliage dry mass removed. Wind speeds were 15, 30, 45, and 60

mph. Of the affects tested, wind speed had the greatest impact on trunk movement.

However, interactions of wind speed with pruning type and wind speed with pruning dose

were also significant. At high wind speeds, thinning did not reduce trunk movement as

effectively as other pruning types. Trunk movement increased with each increase in wind

speed for all pruning types except structural; for structural, differences in trunk

movement were only significant between 15 mph and 60 mph wind speeds. Trunk

movement also increased with each increase in wind speed when 15% or less of the









foliage was removed. Removal of 30% to 45% foliage prevented increases in trunk

movement until wind speeds reached 30 mph. Forty-five mph wind speeds were required

to increase trunk movement when 60% foliage was removed. At low wind speeds, 15

mph or less, trunk movement was similar among all pruning types averaged across

pruning doses, and across all pruning doses averaged across pruning types. Results

indicate no pruning type effectively minimizes wind loads at currently recommended

pruning doses.














CHAPTER 1
INTRODUCTION

Large landscape trees contribute an irreplaceable dimension to urban landscapes,

but they also present a significant, yet oddly acceptable hazard. Large trees, weighing

several thousand pounds, are often found hanging precariously over homes, roads,

recreational areas and other frequently populated sites. Yet relatively little is known

about their construction and less about their response to external loading. Some of the

most damaging external loads trees confront include static loads generated by snow and

ice accumulation and dynamic loads generated by strong winds. This investigation

focused on tree response to loads generated by strong winds.

Wind storms often break and topple trees resulting in damaged property,

interrupted utility service, and personal injury. Pruning is regularly recommended as a

method of abating wind damage in trees. Data supporting that recommendation is scarce

- largely because interactions between wind and trees are elaborate. Large trees are

complex, dynamically built structures and wind is extremely variable in time and space.

Small trees have therefore been used in generated wind fields to simplify the relationship

for investigation.

This study investigated the use of pruning to reduce wind loads on young (5 to 6

year old), Quercus virginiana, live oak trees. Five common pruning practices were

evaluated. Effects of pruning dose and wind speed were also included. Wind velocities

were generated by an airboat elevated so the propeller was at canopy height. Results






2


from this experimental study can be used by tree care professionals to better manage

individual trees in our urban forests.














CHAPTER 2
LITERATURE REVIEW

Significance

Trees break apart and fall over in the wind (Allen 1992). Matheny and Clark

(1994) and many others note there are factors (internal decay, poor architecture, age,

human encroachment, etc.) which cause defects that predispose trees to failure even in

normally tolerable wind events. Many of the defects that lead to mechanical failure are

visible, and efforts have been made to teach tree care professionals to recognize and

proactively address structural defects (Robbins 1986; Mattheck and Breloer 1994; Smiley

and Bones 2000). However, some defects are not visible and some wind events are

severe enough to damage and destroy even healthy, structurally sound trees (Duryea,

Blakeslee et al. 1996).

Wind damage to trees causes tremendous loss. Economic loss is clearly visible in

forest systems where wind damage results in lost materials that might have been

harvested for lumber or paper production. "Wind damage is believed to cost countries in

the European Union approximately 15 million Euros per year, and on occasions

substantially more" (British Forestry Commission 2005). Economic loss can also be

attributed to clean-up and restoration of damaged property in urban environments (Ham

and Rowe 1990). Costs are associated with debris removal (Whittier, Rue et al. 1995),

insurance claims, and restoration of utilities. Repair and replacement of the urban forest

are also viewed as necessary in maintaining a functioning society with an acceptable









standard of living (Westphal 2003). Worst of all, injury and loss of life are often

associated with wind damage in both forests and urban settings (Graham 1990).

History

In 1977 J. Grace published a monograph entitled Plant Response to Wind. One of

his main objectives was to review the literature so that botanists, foresters, agronomists,

etc. could evaluate the state of information about how plants respond to wind

physiologically, anatomically, ecologically, and mechanically. The interaction between

plants and wind is a complex matter. In July 1993, the International Union of Forestry

Research Organizations (IUFRO) brought together scientists from many disciplines for a

first conference on "Wind and Wind-Related Damage to Trees." Proceedings of that

conference were published in 1995 in a volume titled Wind and Trees (Coutts and Grace

1995). In 1998, IUFRO held its second conference on "Wind and other Abiotic Risks to

Forests". Selected papers from that conference were published in Forest Ecology and

Management issue 135 (Peltola, Gardiner et al. 2000). Most recently, a third

international conference, "Wind Effects on Trees," was held in September 2003 and its

proceedings were published in text as were the first (Ruck, Kottmeier et al. 2003) but

public copies are scant (in 2005 the University of Florida's Marston Science Library was

unable to purchase a copy or acquire one through interlibrary loan). All three

conferences aimed at understanding the effects of wind on forest systems and most of the

work presented was done on coniferous species. Roodbaraky et al. (1994) remarked that

by 1990, there was already a body of literature dealing with "the effect of wind on

woodland conifers" but very little existed for broad-leaved species. In the decade

following his remarks, more research was conducted using broad leaf species (Vogel

1995; Niklas and Spatz 2000; Rebertus and Meier 2001; Vollsinger, Mitchell et al. 2005),









but information is still scarce. Arborists in America, interested in the interaction between

wind and trees in urban settings, organized another conference on tree biomechanics held

in March of 2001. Proceedings of that conference were published by the International

Society of Arboriculture (Smiley and Coder 2002).

A few other texts merit attention here for their influence on the primary literature.

Steven Vogel's Life in Moving Fluids (1994) is an excellent resource for biologists and

engineers interested in the interface between biology and fluid mechanics. Pertinent

subjects discussed include: principles of fluid flow, drag, drag coefficients, biological

strategies to reduce drag, and complexities of fluid flow like unsteady flows, velocity

gradients, and boundary layers. Structures: or Why Things Don't Fall Down (Gordon

1981) is both an instructive and enjoyable read introducing pertinent material properties

like stress, strain, shear, torsion, and fracture as well as pertinent design considerations

like safety and efficiency but always in man-made structures. Many of those same

material and design principles are discussed in relation to plants in Plant Biomechanics

(Niklas 1992). Finally, Wood the Internal Optimization of Trees (Mattheck and Kubler

1997) presents an engineering approach to developmental biology of trees including a

discussion of stress transfer through live wood, tree response to wounding, and the

proposed "axiom of uniform stress."

Engineering Principles

Wind damage to trees (as it is considered here) is a structural rather than biological

issue. Biological functions, like photosynthesis, nutrient assimilation, hydraulics,

growth, reproduction, etc, are all determinants of tree growth and development (Ryan and

Yoder 1997). Growth and development are determined even more significantly by a

tree's genetics and evolutionary fitness with the latter not only influenced by the









organism but by its species (Niklas 1998; 2000b). Still, trees are subject to the same

physical laws of nature as any other engineered structure (Niklas 1992; Savidge 1996).

Engineering principles should therefore apply to tree structures just as they apply to any

man-made structure, and indeed they do (Schuler and Bruhn 1973; Mattheck and Vorberg

1991; Spatz, Kohler et al. 1999; Niklas 2000a; Fourcaud and Lac 2003). Constraints

governing man-made construction differ from those governing tree growth and

development. As a result, trees are able to employ strategies that human engineering

avoids such as bending, twisting, and reconfiguration (Vogel 1995). In order to analyze a

tree's mechanical design, engineering principles have to be expanded to deal with

complications like "large deflections" (Kemper 1968), "complex loading" (Morgan and

Cannell 1987), and composite material properties (Spatz, Kohler et al. 1999).

Conversely, tree structure is restrained by efficiency in the allocation of resources.

Therefore, factors of safety (load capacity / self weight) in trees are much smaller than

those found in man-made designs (Niklas 1999). It should be clear that an engineering

approach to tree biomechanics is useful only with an appropriate consideration for

biological elegance.

Scientific Approach

Mechanically, wind damage in trees has been classified as wind tilt, wind throw,

wind prune, and wind snap, (Allen 1992). Wind tilt and wind throw are defined as the

inability of the roots and soil to resist uprooting either partially or completely when

lateral forces acting on a tree reach critical limits. Wind prune and wind snap are the

inability of branches and trunks respectively to resist breaking under those same

conditions. Catastrophic mechanical failure has been classified in similar fashion as soil

failure, root failure, or trunk failure (Sommerville 1979; Moore 2000). Soil failure is









distinguished as wind tilt or wind throw accompanied by extraction of a characteristic

root plate. Root failure is wind tilt or wind throw with minor soil heaving at the base of

the trunk, but no noticeable root plate. Trunk failure is wind snap. An analytical

approach to studying and predicting wind damage (especially catastrophic mechanical

failure) was adopted in the 1970's and consisted of comparing forces imposed by wind to

forces required to break or topple trees (Mayhead 1973; Grace 1977).

Wind Forces

Wind associated forces acting on a tree are shown in Figure 2.1. Detailed reviews

of the mechanics associated with interactions between wind and trees are available

(Blackburn, Petty et al. 1988; Wood 1995; Spatz and Bruechert 2000). The primary wind

associated force acting on a tree is the drag force as given by Equation 2.1 (f, in Figure

2.1).

1
D = C (z)U 2 (z)A(z) (2.1)
2

Drag force (D) is equal to 12 density of the fluid (p in this case air) multiplied by the

drag coefficient (CD), velocity of the fluid squared (U2) and surface area projected into

the fluid flow (A) at a given elevation (z). Drag force is compounded by the height of the

stem over which it acts as described by Equation 2.2.

M, = (D)1, (2.2)

In Equation 2.2, M1 is the drag force moment, D is the drag force as in Equation 2.1, and

1I is the height on the stem over which the drag force acts.

Drag coefficients

Drag is the sum of bluff body pressure and skin friction components that vary in

their magnitude depending on the shape and roughness of an object (Niklas 1992). The









drag coefficient is a dimensionless constant that accounts for variation in shape,

roughness, and all other "oddities in the behavior of drag" which are described more

thoroughly by Vogel (1994). Drag coefficients for forest trees were reported as early as

1962 (Mayhead 1973). Wind tunnels were used to generate known wind speeds and

resultant drag was measured on both individual trees and model forests (Fraser 1964).

Mayhead (1973) reports drag coefficients for eight coniferous species at wind speeds

"likely to cause wind throw (i.e. [68.0 mph], 30.5 m/sec)" and proposes their use in

predicting critical heights of trees (the height at which the given wind speed causes wind

throw). He notes however that "they are a dangerous extrapolation" and that "good

predictive work will require more accurate values for the drag coefficient". Yet his

values have been used in risk assessment studies as late as 2001 (Moore and Gardiner

2001).

Rudnicki et al. (2004) and Vollsinger et al. (2005) followed up the work of

Mayhead with improvements in the determination of drag coefficients by accounting for

streamlining, the speed specific reorientation of branches and leaves in the wind. Using

digital video to capture wind speed specific frontal area (Ad), they showed that at the

highest wind speeds tested (20 m/s) Ad decreased by as much as 54% in the conifers and

37% in the hardwoods. Still, drag coefficients for hardwood species were "less than half

the values typically reported for needled conifers at equivalent speeds" (Vollsinger,

Mitchell et al. 2005). Also, drag coefficients for both conifers and hardwoods were

greater and didn't decrease as sharply with increasing wind speed when calculated using

Ad, as when using the still air frontal area. Both papers confirmed Mayhead's findings

that drag coefficients vary among species, and they discouraged extrapolation to other









species. Both groups also revisited some of Mayhead's unpublished work and reported a

linear relationship between drag and the product of wind speed (U) and canopy mass (Mc)

in all conifer and hardwood species tested. Therefore, they proposed a simplified drag

equation for risk assessment, Equation 2.3 that eliminates the need for calculating frontal

area and errors associated with using inaccurate drag coefficients.

D = MU (2.3)

With this they recognized that the trees used were small (saplings 3-5 m in height) and

that their individual branches behaved independently. Further work is needed to define

drag relationships in older and larger trees.

In a commentary A. R. Ennos (1999) argued "there is little unequivocal evidence of

drag reduction in large trees as a result of reconfiguration". He cites McMahon (1973)

and Bertram (1989) while reminding readers that mature trees have thicker stems to cope

with larger gravitational loads so they are less flexible. Vogel (1989) showed that leaves

and clusters of leaves will often streamline when oriented appropriately in a straight line

wind and thus reduce their drag, but stiffer leaves did not follow suit. Instead of scaling

proportional to the first power of wind speed, drag on mature trees may scale like Vogel's

(1989) white oak leaf, at a power even larger than that seen in the classical drag equation

(Equation 2.1). Attempts have been made to measure drag coefficients in field grown

trees but they are fraught with uncertainty and are not reliable (Ennos 1999).

Wind speed

Definition of the vertical wind profile and spectra that cause damage to trees is an

almost esoteric subject that is beyond the scope of this work. However, Lee (2000) and

Finnigan and Brunet (1995) reviewed the literature on this subject for the 1998 and 1993









IUFRO conferences, respectively. Wind profiles are reported as mean predicted wind

speeds. Calculations of drag coefficients in wind tunnels were conducted with straight

line wind profiles (Mayhead 1973; Rudnicki, Mitchell et al. 2004; Vollsinger, Mitchell et

al. 2005). Wind profiles used in theoretical modeling (Hedden, Fredericksen et al. 1995;

Peltola, Nykanen et al. 1997; Kerzenmacher and Gardiner 1998) typically follow the

theoretical profile presented by Oliver and Mayhead (1974), which is an exponential

profile within the canopy and a logarithmic profile above it. Niklas (2000a) measured the

wind profile used when calculating safety factors and Nilas and Spatz (2000) measured

the wind profile when calculating wind induced stem stresses in an open-grown cherry

tree. These profiles were best described by a third order polynomial equation. Stem

stresses were then recalculated using logarithmic, constant speed (straight line), square

root, and square (exponential) profiles to compare among different vertical wind profiles

commonly used or seen in nature (Spatz and Bruechert 2000). They reported that "stress

levels generated were insensitive to the 'shape' of the wind speed profile" (Niklas and

Spatz 2000) compared to other factors. None the less, variations in wind spectra are still

thought to explain the random nature of wind damage in trees (Luley, Sisinni et al. 2002).

Lee (2000) noted there was still a dearth of information on wind flow over undulating

terrain, in extreme wind events, and in inhomogeneous canopies with irregular, more

realistic edge transitions and forest clearings; all of which are common in urban forests.

There is also very little information about wind speeds within and around individual

trees. Zhu et al. (2000) reported that vertical and horizontal wind profiles within the

canopy of a single Japanese black pine followed exponential functions described by

elevation and crown thickness, respectively. They noted that average wind speeds within









the crown were only about half what they were outside it. They also proposed equations

for calculating interior wind speeds at any elevation in a crown based on a single

measurement outside a crown, and anywhere within a horizontal plane in a crown based

on a single measurement outside a crown at the same elevation.

Tree Resistance

The second part of the analytical approach introduced above is a determination of

the force required to break or topple a tree. Before any additional force is considered, a

tree must first cope with the load of gravity or its self weight as described by Equation

2.4 and illustrated in Figure 2.1.

f2 = m(1)g (2.4)

In Equation 2.4f2 is the force of gravity; m (/i) is the mass of the canopy at an elevation

along the trunk lI; and g is gravitational acceleration. Calculations based on scaling of

trunk diameter with tree height predicted a safety factor against elastic buckling under

self-weight near four (McMahon 1973). Niklas (1994) and Mattheck et al. (1993)

independently confirmed earlier calculations through experimentation. However, Niklas

(1997a) later argued that estimates of wood density used in previous calculations were

unrealistically low and safety factors against elastic buckling are probably closer to two.

He also reported (Niklas 1997b; c) that ontogenetic changes in size, shape, and wood

properties occurred of necessity, or imposed stresses would reach critical levels as trees

grew in size. Ontogenetic development allows for trunk and proximal branches to be

rigid in support of larger gravitational loads resulting from increased mass. At the same

time distal branches remain flexible so the canopy maintains an ability to reconfigure and









reduce its drag in the wind. As a matter of perspective, most trees are capable of

successfully resisting gravitational loads.

Bending and wind snap

Wind drag increases gravitational load on a tree by generating a moment (Equation

2.5) as the trunk and stems bend (Grace 1977; Spatz and Bruechert 2000).

M2 = f22 (2.5)

In Equation 2.5, M2 is the gravitational moment;f2 is as before; and 12 is the magnitude of

deflection. The magnitude of deflection is described by engineering beam theory and

depends not only on the drag force exerted on the crown, but on the geometry and

material properties of the stem as given in Equations 2.6 and 2.7 (Grace 1977; Gordon

1981).

(D)13
21- ((2.6)
3EI

In Equation 2.6, 12 and D are as before, I is the second moment of area of a cross section,

and E is the modulus of elasticity (Young's modulus).

mZ4
64

In Equation 2.7, is as before, dis the diameter of the stem, and ;r is the area of the unit

circle. It should be noted that there are many equations for I and the one chosen depends

on the specific cross sectional area Equation 2.7 is specific for a circular cross section.

Introductory information about mechanics of materials and beam theory is provided by

Gordon (1981) and Salvadori (1980).

Trunk taper accounts for the change in I over the length of the stem and influences

stress distribution within the stem. Petty and Swain (1985) have suggested that "taper is









probably the most important factor affecting susceptibility to stem breakage". Niklas and

Spatz (2000) found that because of changes in taper and canopy shape, safety factors

against wind induced stress likely decrease as trees mature. They also submit that wind

induced stresses and related safety factors are not uniform throughout the canopy of a tree

(Niklas 2000a). Niklas and Spatz (2000) found that material yield stresses decreased

with increasing stem diameter for all trees examined, while stress levels were lowest at

the most distal and basal portions of the tree for all but the oldest tree. For the oldest tree,

stress levels were highest at the trunk base. Leiser and Kemper (1973) had earlier

demonstrated that assuming homogeneity of material properties, taper determines the

location of maximal stress in a sapling tree trunk. Severely tapered trunks had the

greatest maximal stresses of all trunks analyzed with those stresses located high on the

trunk near the point of loading. Untapered trunks also had large maximum stresses but

they were located at the trunk base. Both of these findings support ecological adaptations

that are proposed in the literature. Distribution of maximal bending stresses to the distal

portions of the tree may allow for preferential branch shedding (a form of "self pruning")

to reduce drag in high winds (Hedden, Fredericksen et al. 1995; Mattheck and Kubler

1997; Niklas 2000a; Beismann, Schweingruber et al. 2002). It may also be an

evolutionary strategy facilitating asexual reproduction (Blackburn, Petty et al. 1988).

From a purely mechanical perspective, the best designs distribute stress uniformly

along an entire structure. With that in mind, Mattheck and Kubler (1997) presented the

"axiom of uniform stress" based on their work using the Finite Element Method to

analyze notch stresses and allometry of trunks, branches, and unions. Uniform stress was

originally proposed by Metzger, who discovered that taper of spruce trees (height ~









diameter3) assured a uniform distribution of bending stresses along the trunk. Other

analyses of shapes of stems (Morgan and Cannell 1994) and scaling factors of trees

(McMahon 1973) provide further evidence of the "axiom of uniform stress". However,

in all these studies, wood mechanical properties were assumed to be uniform or to change

negligibly throughout the specimen. The assumption of uniform stress has been made

repeatedly (Fraser 1962; Petty and Worrell 1981; Milne 1991; Peltola, Nykanen et al.

1997; Saunderson, England et al. 1999) and appears to be valid at least in young trees

(Leiser and Kemper 1973; Petty and Worrell 1981). Still, Wood (1995) admits there are

questions as to the accuracy of the assumption of uniform stress. And those who include

measures of wood anatomical features along with taper in their analysis of stress

distribution report variations in the longitudinal stress distribution within the stem

(Ezquerra and Gil 2001).

From Equation 2.6 it is clear that the modulus of elasticity or Young's modulus is

another important factor determining bending stresses in stems. Young's modulus is a

measure of the "stiffness" of a material (Salvadori 1980; Gordon 1981). Tabulated

values of average Young's moduli for both kiln dried and green wood are readily

available on a number of commercially important species (Green, Winandy et al. 1999).

It should be noted however, that tabulated values are averages taken from small clear

specimens of wood (Green, Winandy et al. 1999; Bailey 2000a) and it has been shown

that there is tremendous variability in the Young's modulus within a tree (Niklas 1997a;

b; Niklas and Spatz 2000). Cannell and Morgan (1987) reported that Young's moduli

measured in intact branches and trunks are lower than tabulated green wood values and

suggested that a structural modulus of elasticity (the modulus of the intact living stem) be









used to more accurately represent material properties of live wood. Accordingly, efforts

have been made to measure the modulus of elasticity for living trees but with the

assumption that it is uniform throughout the trunk (Milne 1991; Peltola, Kellomaki et al.

2000).

Foresters have long known that material properties of wood are not isotropic

(Green, Winandy et al. 1999).Wood scientists have shown that 86% of the variability in

the longitudinal Young's modulus is accounted for by orientation of cellulose

microfibrils (microfibril angle (MFA)) in cell walls (Evans and Ilic 2001). Barnett and

Bonham (2004) recently reviewed the literature on MFA. High MFA's are associated

with low Young's moduli and are common in "juvenile wood" and "compression wood"

while low MFA's convey a high Young's modulus and are associated with "adult wood"

and "tension wood". Factors that control the orientation of cellulose microfibrils in cell

walls are still unknown (Baskin 2001). Work on MFA refutes the assumption of material

homogeneity of stems but does not disprove the possibility of uniform stress.

Wind throw

Sommerville (1979) suggested that widespread wind snap poses a greater economic

threat to forest systems than wind throw. However, widespread wind throw is more

prevalent (Blackburn, Petty et al. 1988; Papesch, Moore et al. 1997; Moore 2000) and as

such it has received far more attention in the literature. In an effort to elucidate factors

involved in wind throw, Fraser (1962) introduced a pull test using a winch and cable and

demonstrated that Fomes annosus root rot caused lower resistance to lodging while

improved drainage had an opposite affect in Douglas fir (Pseudotsuga menziesii). Others

used similar pull tests to demonstrate that lodging is dependent on taper (Petty and Swain

1985), rooting depth and root morphology (Sommerville 1979), soil type and moisture









content (Moore 2000), and species (Peltola, Kellomaki et al. 2000; Moore and Gardiner

2001). Coutts (1986) used tree pulling tests to demonstrate that applied wind loads were

much more significant in causing wind throw than associated gravitational loads a

finding that was later confirmed by Papesch et al. (1997). Coutts also annotated the

sequence of wind throw in shallow rooted Sitka Spruce (Coutts 1983) and defined

components of root anchorage (Coutts 1986). Collectively, pull tests have been used to

determine maximum bending moments of trees in order to make comparisons with

maximum applied loads. Those comparisons are then used to create wind throw

prediction models (Peltola, Nykanen et al. 1997; Kerzenmacher and Gardiner 1998;

Saunderson, England et al. 1999; Moore and Quine 2000; Talkkari, Peltola et al. 2000)

and to develop improved cultural practices (Cremer, Borough et al. 1982; Brichert,

Becker et al. 2000; Gardiner and Quine 2000; Mitchell 2000). However, pull tests have

mainly been used to evaluate static loads (Fraser 1962; Blackburn, Petty et al. 1988) and

Oliver and Mayhead (1974) revealed that wind damage occurs at mean wind speeds well

below those predicted by tree pulling tests. Static tests are a crude approximation of

actual loads experienced by trees in wind events.

Tree Dynamics

Wind is a dynamic force that generates complex loads in stems as trees sway.

Moore and Maguire (2004) recently reviewed the literature investigating tree dynamics.

Most of the research has been conducted in forest systems where trees are in close

proximity and exhibit strongly excurrent growth habits. Milne (1991) used pull tests to

study natural frequency and damping of Sitka spruce (Picea sitchensis) in a forest setting.

He reported that the greatest component damping a tree's sway was interaction between

crowns and branches of neighboring trees. Moore and Maguire (2005) followed up on









Milne's work and found that interactions between trees were insignificant. They reported

that aerodynamic damping of foliage had the greatest effect on tree sway. They also

reported that it is inappropriate to represent tree crowns as a series of lumped masses

when calculating wind and gravitational loads, something that Mayhead had discovered

earlier but never published (Moore and Maguire 2005). Still, Mayhead (1973), Rudnicki

et al. (2004), and Vollsinger et al. (2005) have all shown strong relationships between

canopy mass and drag. A model describing a tree crown as a series of independent mass

dampers was presented by Kerzenmacher and Gardiner (1998) and again by James

(2003). Mass damping seems very likely and research is underway to evaluate the effect

of branches as coupled mass dampers (James 2003; Moore and Maguire 2005).

As a final note on tree dynamics and wind loading in general, Denny (1994)

warned that the probability of wind damage calculated from extreme measurements is

much greater than that calculated from an average as is typically done. Therefore, if

models that are generated to predict catastrophic failure in trees are to be used in urban

settings, consideration needs to be given to the severity of a single catastrophic failure.

Since trees do not construct to man-made factors of safety, models should compensate to

some degree.

Pruning

Mechanical pruning is commonly thought to reduce wind damage from strong

winds because it reduces the surface area of the tree canopy (Mattheck and Breloer

1994). Duryea et al. (1996) noted that pruned trees withstood wind damage from

Hurricane Andrew better than their unpruned counterparts. Ham and Rowe (1990) felt

that despite losing over 4800 street trees, damage to the city of Charlotte, NC from

Hurricane Hugo was "lessened by their program of routine [tree] maintenance."









However, a survey of street trees in Rochester, NY showed no difference in the frequency

of storm damaged trees in pruned versus unpruned areas of the city (Luley, Sisinni et al.

2002). Pruning is one of the most prominent tree maintenance practices (Accredited

Standards Committee A300 2001), and it is widely recommended as a means of reducing

wind damage to trees (Matheny and Clark 1994; Brown 1995; Gilman 2002; Harris,

Clark et al. 2004). Never the less, research supporting recommendations for pruning

trees to reduce wind damage is almost nonexistent in the primary literature. Moore and

Maguire (2002) mention that the natural frequency of Douglas fir trees does not appear to

be significantly affected by pruning until the topmost third of the canopy is removed.

They suggest this is due to the paucity of foliage in the lower canopy and higher wind

velocities at higher elevations. Rudnicki et al. (2004) and Vollsinger et al. (2005)

evaluated whole branch removal in their calculation of drag coefficients and reported that

pruning did not influence streamlining in coniferous or hardwood species tested.

However, they noted that their test specimens were small with branches that behaved

independently, so effects of streamlining were inconclusive. Even with these recent

reports, the primary literature appears void of evidence for pruning as a means to mitigate

wind damage. There is even less information about how pruning should be

accomplished.

The American National Standards Institute (ANSI) A-300 (2001) pruning standard

and other pruning references (Lilly, Clark et al. 1993; Brown 1995; Brickell and Joyce

1996; Lang and Editors 1998; Gilman 2002; Harris, Clark et al. 2004) list four primary

pruning practices: cleaning, thinning, raising (also called lifting or skirting), and

reduction. Other colloquial terms defining specific pruning practices include utility,









structural, risk reduction, balancing, vista, restoration, topping, tipping, or lion's tailing

but each of those are a type or combination of the primary four. There are also specialty

pruning practices such as coppicing, pollarding, pleaching, topiary, espalier, bonsai, and

fruit tree pruning (Accredited Standards Committee A300 2001) which have historically

been used for specific effects, but those are not commonly used in landscapes. For

definitions of the four primary pruning practices refer to the ANSI A-300 (2001) pruning

standard section 5.6 or Gilman (2002).

Dose

Pruning dose is "the amount of live tissue removed at one pruning" (Accredited

Standards Committee A300 2001). ANSI recommends limits to pruning as defined by a

"percent of the foliage removed [from a tree] within an annual growing season" (Gilman

2002; Gilman and Lilly 2002; Harris, Clark et al. 2004). However, ANSI provides no

information about how to quantify the percent of foliage removed. Other resources for

arborists refer back to the ANSI pruning standard on questions of dose (Gilman 2002;

Gilman and Lilly 2002). They also advise practitioners to quantify pruning dose based

on the desired objective (Brown 1995; Harris, Clark et al. 2004), or appearance of the tree

following the pruning (Waring, Schroeder et al. 1982). Thus measurement of pruning

dose by practitioners is largely qualitative and subjective.

The most reliable way to quantify pruning dose is destructive. Following pruning,

the foliage and stems removed and the remaining canopy are cut into small pieces and

dried to a constant weight. Pruning dose is then calculated as a percentage of total dry

mass of foliage removed. Alternative, nondestructive methods for determining pruning

dose could be considered using biomass prediction equations calculated from sapwood

area or from trunk diameter at breast height (DBH). However, while those alternatives









may prove useful to investigators, they would need some modifications before finding

their way to the practitioner.

Sapwood cross sectional area has been used to generate prediction equations for

leaf area based on the pipe model theory (Kaufmann and Troendle 1981; Waring,

Schroeder et al. 1982; Meadows and Hodges 2002). Prediction equations are species

specific (Rogers and Hinckley 1979), but have been generated for a number of species

including Quercus velutina and Q. alba (White 1993), Q. rubra (Meadows and Hodges

2002), Q. falcata var. pagodifolia and Fraxinuspennsylvanica (Kaufmann and Troendle

1981), Picea engelmannii, Pinus contorta, Abies lasiocarpa, and Populus tremuloides

(Ohara and Valappil 1995; Meadows and Hodges 2002; Medhurst and Beadle 2002).

Leaf area is first correlated to leaf biomass, and then to sapwood area. Within a species,

correlation is independent of age, site, strata, crown class, crown condition, and stand

density (Rogers and Hinckley 1979). However, there is debate as to whether leaf area

should be correlated to current sapwood area (Meadows and Hodges 2002) or total

sapwood area (McDowell, Barnard et al. 2002). There are also complications to the

models because leaf area: sapwood area generally decreases as total height of the tree

increases (Meadows and Hodges 2002), and leaf area: unit weight is larger in the lower,

more shaded, parts of the canopy than in the intermediate and upper parts (Ter-Mikaelian

and Korzukhin 1997). Prediction equations need to be determined for more species

before this method becomes useful to a large degree. Prediction equations using DBH

are more commonly used.

There is a sizeable amount of information relating various biomass components to

DBH. Ter-Mikaelian and Korzukhin (1997) provide an extensive review of the literature









for 65 North American tree species. Prediction equations reviewed are all of the same

form but there are often multiple prediction equations for the same biomass component.

As a result, the authors present all the different prediction equations together, as well as

various components used to generate these equations such as the range, sample size, and

standard error of the estimate; as well as geographic region of the sample, and a reference

to the paper where it was cited. The additional information is included as an aid in

determining which prediction equation is most appropriate for individual circumstances.

There are a considerable number of non-coniferous species listed, but species included

are limited to those important in forest systems. Additional work is needed to generate

similar equations for commonly used landscape species. Another impasse to using the

biomass equations based on DBH is that they predict the oven-dry weight of the

individual component. This is fine for a researcher in predicting pruning dose but

completely impractical for a practitioner.

Conclusions

This review presents a summary of work published in the primary literature

associated with wind damage and pruning of amenity trees. Tremendous strides are

being made towards the understanding of wind forces on trees and trees associated

mechanical response. Additional information is needed in many areas including: wind

profiles and wind spectra for both forests and individual trees, drag coefficients of large,

open grown, broad-leaved species, root and soil interactions affecting anchorage, and

wood material properties. There is especially a dearth of practical information that can

be used by tree care providers in urban settings. The subsequent study is aimed at

providing useful information regarding pruning as a means for reducing wind damage in

amenity trees.































Figure 2.1. Wind forces affecting trees (modified from Grace (1977)).

Wind force: (f,)= parCd ( )U2 ( )A(1,).
2

Wind moment: M1 = (f,)/,.

Gravitational force: f2 = m(l1)g.

Gravitational moment: M2 = f212.


Magnitude of deflection: 12 =- )
3EI


Second moment of area: I =
64

Constants are as follows: pair = density of air, Cd =drag coefficient, 1 = elevation
to the point of loading, U = wind speed, A = projected surface area of the canopy,
m =mass of the canopy at l1 g = acceleration of gravity, I2 magnitude of
deflection, E = modulus of elasticity, d = diameter of cross section, and r = area
of the unit circle.














CHAPTER 3
MATERIALS AND METHODS

Tree Selection

Clonally propagated trees were chosen as subjects for testing to improve similarity

among trees assigned to different experimental treatments. Live oak trees (Quercus

virginiana 'QVTIA' PP 11219, Highrise) were used because they are a commonly

planted species in the southeastern United States and were readily available from a local

nursery. Trees were selected for physical similarity from a population of Highrise live

oaks grown at Marshall Tree Farm (MTF; Morriston, FL, U.S.A.). Marshall Tree Farm

transplanted the trees from five gallon containers into native soils (Orlando fine sand or

Sparr fine sand) in June 2000. In May 2003, clear trunk, total canopy height and

diameter, trunk taper parameter, football diameter, height to vertical center of a canopy,

projected area of a canopy and canopy volume were measured on 50 trees and used as

characteristics for selection (Table 3-1). English units were chosen for use throughout

this thesis in order to follow American engineering convention. Clear trunk was the

distance between the top-most root and first main branch. A minimum clear trunk of 54

in. was required for selection. Total canopy height (Tch) was determined by measuring

maximum (Mxch) height (distance from top of the football to top of the canopy),

minimum (Mnch) height (distance from top of the football to origin of the lowest

branch), and subtracting Mnch from Mxch Canopy diameter (Cd) was the average width

of a canopy measured at its widest point in two perpendicular directions. A visual

estimate of the canopy outline was used to determine Cd single shoots extending









beyond the canopy outline were neglected. Trunk taper parameter (Taper) was calculated

as -(R-r)/R (Leiser and Kemper 1973) where R = radius at 6 in. above the top-most root,

and r = radius at 54 in. above the top-most root. Rootball diameter (Rd) was average

width measured at the soil surface in two perpendicular directions. Height to vertical

center of a canopy (Hvcc) was one half Tch plus Mnch. Projected area (Area) was

calculated as 0.5 x the vertical surface area of a cone (whose dimensions were: height =

0.667 x Tch and radius = 0.5 x Cd) plus 0.25 x the surface area of an ellipsoid (whose

dimensions were: radius of the long axis = 0.333 x Ch and radius of the short axis = 0.5 x

Cd). Canopy volume (Volume) was calculated as the volume of a cone (calculated as

before) plus one half the volume of an ellipsoid (calculated as before). Measurement of

any characteristic that was greater than three standard deviations from its sample mean

was cause for rejection. This eliminated obvious outliers according to the Empirical Rule

in statistics (Ott and Longnecker 2001).

On November 6, 2003 forty-four trees were moved from MTF to the

Environmental Horticulture Teaching Unit (Tree Unit, Gainesville, Florida, U.S.A.).

Trees met Roots Plus GrowersTM standards (Roots Plus Growers Association of Florida

2005); meaning they were dug and completely hardened-off at the nursery with visible

roots on the outside of the root ball prior to shipment. Upon arrival, trees were weighed

footballl included) using a dynamometer (Model WT-1-1000H John Chatillon & Sons,

New York, NY) then healed into pre-dug holes. Holes were the same dimensions as the

root balls and trees were healed in without removing burlap, wire, or nylon bag that

secured the football. Trees were irrigated three times per day with approximately four

gallons of water per irrigation. After healing in, tree movement was minimized prior to









testing. None of the trees suffered significant defoliation resulting from transport to the

Tree Unit.

Experimental Design

Three effects were evaluated: 1) pruning type, 2) pruning dose, and 3) wind speed.

From those three effects, 100 treatment combinations were constructed. Trees were

randomly assigned to a pruning type following a completely randomized design. Five

pruning types were included: 1) lion's tailing, 2) raising, 3) reduction, 4) structural, and

5) thinning. Each tree, within a type, was pruned three times to approximate targeted,

orthogonally spaced, pruning dose levels. Targeted pruning dose levels were: 1) 15%

foliage removed, 2) 30% foliage removed, and 3) 45% foliage removed. Within a

pruning dose, each tree was subjected to a sequence of wind speeds. Four, equally

spaced wind speed levels were targeted: 1) 15 mph, 2) 30 mph, 3) 45 mph, and 4) 60

mph. Data was collected on every tree, before it was pruned and at all targeted pruning

dose levels, at ambient wind speeds and at all targeted wind speeds. Therefore, data was

collected 20 different times on the same tree within a pruning type. The first four lion's

tailed, raised, reduced, and thinned trees were blocked in time by tree within type

forming a complete block design. The last three lion's tailed, raised, and reduced trees

and all structurally pruned trees were added in no particular order.

Pruning Type

Pruning types are defined in the American National Standard for Tree Care

Operations (ANSI 2001) and in An Illustrated Guide to Pruning Second Edition

(Gilman 2002). Five types evaluated here are common in practice or are recommended

as a means to reduce wind damage to trees (Matheny and Clark 1994; Brown 1995;

Gilman 2002; Harris, Clark et al. 2004). One person was chosen to execute all pruning to









maintain treatment consistency. Prior to testing, trees similar to those selected for

experimentation were pruned by a team of individuals and pruning types and doses

discussed. During experimentation, parts of a canopy removed by pruning were collected

and stored in paper bags for gravimetric analysis of actual pruning dose.

Lion's Tailing

Four trees were lion's tailed as part of the complete block design. Pruning

consisted of removing primary and higher order branches smaller than 0.5 inches in

diameter at their point of attachment to the trunk or parent stem. Branch diameter was

determined using a digital caliper as average width measured in two perpendicular

directions. Pruning dose levels were determined in the field from geometric dimensions

of the canopy; all tissue removed was dried and weighed to calculate actual dose. The

15% pruning dose consisted of pruning within the lowest 15% of canopy height and the

most interior 15% of canopy radius. The 30% and 45% pruning doses were applied in

similar fashion. Canopy height and diameter measurements were taken on the day of

testing following the procedure used for tree selection. Pruned volume was determined

as follows. The canopy's main leader was marked at 15%, 30%, and 45% of total canopy

height calculated as: canopy height x (0.15, 0.30, and 0.45 respectively) + min. height.

Fifteen, thirty, and forty-five percent of canopy radius was calculated by multiplying 0.5

x canopy diameter by 0.15, 0.30 and 0.45 respectively. During pruning, canopy radius

was measured with a common tape measure.

Foliage removed from the first four lion's tailed trees was inadequate at all dose

levels. Therefore, three additional trees were included to better approximate targeted

pruning dose levels. The three additional lion's tailed trees were blocked in time with

each other but not with other pruning types. Pruning, canopy appearance, and canopy









structure of the three additional lion's tailed trees was similar to the first four, but pruning

dose in the field was determined as a visual estimate of live foliage removed. All tissue

removed was dried and weighed to calculate actual dose. Examples of lion's tailing from

an urban landscape and pruning of test trees at each dose are provided in Figures 3.1 to

3.3. Seven lion's tailed trees were included in the statistical analysis.

Raising

The first four raised trees were blocked in time with other pruning types. The first

raised tree was the first of all trees tested. It was not included in the analysis because

experimental procedure changed for all subsequent trees. Pruning dose levels for the first

four raised trees were determined in the field from geometric dimensions of the canopy as

follows. On the day of testing, canopy height was measured and the main leader was

marked at 15%, 30%, and 45% of total canopy height following the procedure used for

lion's tailing. The 15% pruning dose was applied by pruning from the base of the canopy

up the main leader to the 15% mark. The 30% and 45% dose levels were applied in

similar fashion. All tissue removed was dried and weighed to calculate actual dose.

Pruning was conducted by removing all branches from the main leader at their point of

attachment. Branches originating high in the canopy but drooping below the elevation of

the mark on the main leader were also removed to that elevation so the entire canopy

width was lifted. Foliage removed at each pruning dose for the first four raised trees

exceeded targeted levels so three additional trees were included to better approximate

targeted pruning doses.

The three additional raised trees were blocked in time with each other but not with

other pruning types. Pruning dose for the additional raised trees was determined in the

field as a visual estimate of live foliage removed; all tissue removed was dried and









weighed to calculate actual dose. Pruning was carried out as before, but if removal of a

large limb caused an excessive dose, it was treated as a second leader and raised as per

the main leader. Examples of raising from an urban landscape and pruning of test trees

are provided (Figures 3.4 to 3.6). Six of the seven raised trees were included in the

statistical analysis.

Reduction

Reduction pruning involved making heading cuts (shearing) to reduce the

geometric size of a canopy. Pruning did not involve 'drop-crotch pruning' or using

reduction cuts as is commonly recommended for reducing the size of a tree or part of a

tree in a landscape.

The first four reduced trees were blocked in time with other pruning types. Pruning

dose for the first four reduced trees was determined in the field by geometric dimensions

of a canopy. On the day of testing, canopy height and average canopy diameter were

determined as before. The main leader was marked at 85%, 70%, and 55% of total

canopy height calculated as: max height canopy height x (0.85, 0.70, and 0.55

respectively). Pruning was accomplished by first removing the main leader at the

designated mark followed by pruning the exterior of the remaining canopy to re-establish

each trees original three dimensional shape, but in a smaller version. No foliage was

removed from interior parts of a canopy. All tissue removed was dried and weighed to

calculate actual dose. Foliage removed from all but the first dimensionally reduced tree

exceeded targeted pruning doses at all levels. Therefore, three additional trees were

reduced to better approximate targeted dose levels.

The three additional reduced trees were not blocked in time with each other or with

other pruning types. Pruning was carried out in as per other reduced trees but dose was









determined in the field as a visual estimate of live foliage removed trunks were not

marked prior to pruning. All tissue removed was dried and weighed to calculate actual

dose. Examples of reduction from an urban landscape and pruning of test trees at each

dose are provided (Figures 3.7 to 3.9). Seven reduced trees were included in the

statistical analysis.

Structural

Three trees were structurally pruned. Structurally pruned trees were blocked in

time with each other but were not blocked with other pruning types. They were all

evaluated at the end of the data collection period. Structural pruning involved making

reduction and removal cuts to shorten and slow growth of stems competing with the main

trunk, and to develop scaffold branches. Little thought was given to canopy size, shape,

or density. Pruning dose was determined in the field as a visual estimate of live foliage

removed; all tissue removed was dried and weighed to calculate actual dose. Examples

of structural pruning from an urban landscape and pruning of test trees at each dose are

provided (Figures 3.10 and 3.11). Structurally pruned tree number 2 was left out of the

statistical analysis because of an inadvertent change in procedure. Two structurally

pruned trees were included in the statistical analysis.

Thinning

Four trees were thinned and all four were blocked in time with other pruning types.

Thinning was conducted by making reduction and removal cuts throughout the entire

canopy, especially at the outer edge of a canopy. Pruning dose was determined in the

field as a visual estimate of live foliage removed. All tissue removed was dried and

weighed to calculate actual dose. Thinning produced a uniformly dense canopy without

changing the canopy's initial geometric dimensions. Examples of thinning from an urban









landscape and pruning of test trees at each dose are provided (Figures 3.12 and 3.13).

Four thinned trees were included in the statistical analysis.

Pruning Dose

Pruning dose is defined in Sections 5.5.3 and 5.5.4 of A300 (ANSI 2001) as a

percentage of foliage removed. Pruning dose used in the statistical analyses was

percentage of foliage dry weight removed. All parts of a canopy collected and stored

during pruning were dried at 700C until they reached a constant weight. Foliage was then

separated from stems and both foliage and stem weight were recorded separately for each

treatment combination. The remaining canopy (clear trunk excluded) of 13 of the 27

trees tested was also cut into small sections, dried, and measured as per pruned cuttings to

calculate an average canopy dry weight. Pruning dose was calculated as dry weight of an

individual pruning dose (summed incrementally)/average canopy dry weight x 100.

Pruning dose was also calculated from actual tree canopy dry weight for the 13 trees used

to calculate the average canopy dry weight.

Wind Speed

High winds are not regularly experienced at the University of Florida Tree Unit.

Winds were generated using an airboat (Figure 3.14) driven by a 1988 Chevy 350 engine,

a 2-1-power reduction unit, and a 2-blade, Sensenich wide blade, 78-inch, left hand

rotation, composite propeller (Sensenich Wood Propeller Co., Inc., Plant City, FL, USA).

Airboat rudders were locked in an orientation perpendicular to the long axis of the

propeller and the boat was set on two concrete piers. Piers were engineered to elevate the

propeller's midpoint to the estimated center of pressure on an average unpruned,

undisturbed canopy. This corresponded to one third average total canopy height, or an

elevation of 10 ft. 2 in. The airboat was lifted into place with a crane and was fixed to the









piers with 2 in. angle iron that ran across the hull and 0.75 in. diameter threaded rod

secured into the concrete with epoxy.

A downwind profile of generated wind speed was used to determine the location

for placement of trees during testing. Gill propeller anemometers (Model 27106R R.M.

Young Company, Traverse City, MI, USA) and a proprietary data acquisition program

were used to measure and record wind speeds. Orthogonal anemometers were mounted

on a steel tower 16 ft. 5 in. and 33 ft. off the ground as well as 10 ft. 1 in. off the ground

on an outrigger. The outrigger was located seven feet upwind of the tower.

On November 21, 2003 the first set of downwind profile tests were conducted with

the outrigger located at distances of 18 ft. 5 in., 29 ft. 3 in., and 39 ft. 8 in. downwind

from the airboat propeller. Wind speeds were recorded at 100 Hz for one minute at

motor rpm starting at 1000 rpm and increasing to 4500 rpm in increments of 500 rpm.

Data collected is summarized in Table 3-2. Before additional testing, the airboat stem

was elevated eight inches. The airboat propeller was then centered at an elevation equal

to 10 ft. 10 in. and the airboat hull sat at a five degree angle from horizontal. Tower

anemometers were abandoned in subsequent testing.

Wind speeds were measured on December 3, 2003 to correlate wind speed to motor

rpm (Table 3-3). Anemometers were located 17 ft. downwind from the propeller and

wind speeds were collected at 100 Hz for approximately four minutes at motor rpm

starting at 1000 rpm and increasing to 4500 rpm in increments of 500 rpm. Data is

summarized in Table 3-3.

Two 3-cup anemometers with directional sensors (Met One 034B, Campbell

Scientific, Inc., Logan, UT, USA) measured wind speeds during testing. These









anemometers were mounted in split ring hangers welded to telescoping steel conduit so

they could be elevated to vertical center of the tree canopy (calculated as: canopy height/

2 + min. height). One anemometer was located approximately 7 ft. upwind and the other

approximately 14 ft. downwind of each tree during testing. Directional sensors were

unstable in generated winds so vanes were removed and wind direction was not recorded

during testing.

On April 19, 2004 wind speed was correlated to motor rpm using Campbell

anemometers (Figure 3.15). Anemometers were located at 10 ft. 2 in. and 23 ft. 3 in.

downwind from the airboat propeller and at an elevation of 9 ft. 6 in. with respect to the

height of the propeller (this was equivalent to the height of Young anemometers when

they were 17 ft. downwind of the propeller). Data was collected at 0.5 Hz for three

minutes at each rpm starting at 1000 rpm and running up to 4000 rpm in increments of

500 rpm. Campbell anemometers recorded higher wind speeds than Young anemometers

at the higher rpm (Figure 3.15).

Because there were discrepancies between wind speeds recorded by Young and

Campbell anemometers, a road test was conducted to check the accuracy of the Campbell

anemometers. On April 21, 2004 anemometers were held out the front passenger

window of a Jeep Grand Cherokee as it was driven from 10 mph to 70 mph in increments

of 10 mph. Wind speeds recorded at 0.5 Hz for two minutes at each velocity are

summarized in Table 3-4. Wind speeds recorded by the West anemometer (its location

relative to the tree during testing) were nearly identical to the speedometer reading. The

East anemometer was more variable. It was uncertain if greater error was in the









anemometer or the driver's ability to maintain a constant velocity. Measurements from

the East anemometer were not used in the analysis.

The East anemometer was included as an aside. Coder suggested that canopy

density can be determined by measuring wind speeds on windward and leeward sides of

the canopy (Coder 2000). Wind speeds recorded by the East anemometer were erratic

despite pruning dose so it was not included in the analysis.

Trunk Deflection

Tree response to wind loading was measured as trunk deflection below the canopy.

Before testing, a trunk was marked at 18, 30, 42, and 54 in. above the first root and eight

cable extension transducers (CET) (Celesco Transducer Products Inc., Chatsworth, CA,

USA) were attached to the trunk in pairs at those elevations (Figures 3.14 and 3.16). The

four lowest CETs were fixed with 10 in. cables (PT1A-10-UP-5K-M6-SG), followed a by

a pair of 15 in. CETs (PT1A-15-UP-5K-M6-SG), and finally a pair of 25 in. CETs

(PT1A-25-UP-5K-M6-SG). The 10 in. CETs were calibrated using a common 12 in.

ruler, while 15 in. and 25 in. CETs were calibrated using a 10 ft. tape measure. Linear

regression produced R2 values of 1.0 for all CETs (Figures 3.17 and 3.18).

Transducers were bolted to two pieces of 3 in. x 2 in. angle iron such that each

angle iron had two 10 in., one 15 in., and one 25 in. transducer spaced 1 ft. apart along its

length. During testing, angle irons were clamped to 4 in. x 4 in. wooden posts located

approximately 16 ft. from the tree and spaced approximately 23 ft. 5 in. apart (Figure 3-

19). Elevation of an angle iron on a post was determined on day of testing using line

levels that hung from nylon string stretched between lowest transducers and the 18 in.

mark on a trunk. Piano wire (0.01 in. diameter) was used to create extensions between

transducer cables and the trees. Piano wire extensions were attached to trees with plastic









cable ties. Deflection of the trunk was calculated at each interval as described in Figure

3.20.

Longitudinal Young's modulus is the most significant factor affecting bending

(Barnett and Bonham 2004) and it is known to vary among species and within species.

Every effort was made to select and maintain trees so there would be a uniform

longitudinal Young's modulus among trees tested. Genetic variation among trees was

limited by using clonally propagated trees. Trees were young enough (5-6 years old) that

they were likely composed entirely of juvenile wood with no heartwood. Because they

were nursery grown, they likely had wide, uniform growth rings and little to no reaction

wood. Wood in all trees likely had the same moisture content because trees were

irrigated regularly up until the day they were tested. Still, because variation in the

longitudinal Young's modulus among trees would confound results based on trunk

movement, efforts were made to determine green and kiln dried values of longitudinal

Young's modulus. Testing was conducted on coupons (1 in. x 1 in. x 18 in. sections) cut

from trunks at different elevations, and on whole trunks, following standard test methods

for small clear specimens of timber (Bailey 2000a) and static tests of wood poles (Bailey

2000b) respectively. Figure 3.21 shows testing apparatus used in both procedures.

Young's modulus calculations were conducted by an undergraduate as an University

Scholars' project and more procedural detail is given in that report (Trachet 2005).

Experimental Procedure

Testing began May 19, 2004. One tree was tested per day. Testing dates are

included in Table 3-1. Trees remained irrigated and undisturbed until the day they were

tested. One exception was structurally pruned tree number two, which was excluded

from the analysis because it was moved a day early, and without irrigation, it was









severely water stressed before testing was complete. On a day of testing, each tree was

moved from the field to the testing site and its football and the base of its trunk were

secured so the tree would remain upright. Figure 3.22 depicts the apparatus used to

secure a football and the base of a trunk. Eight cable extension transducers (CETs) were

connected to the trunk and the tree was tested at all wind speeds before any pruning was

executed. Transducers were then disconnected and the tree was pruned to the lowest

targeted pruning dose (15% foliage removed). After pruning, CETs were reconnected

and the process was repeated. Each tree was tested four times once before pruning and

once at each of the three targeted pruning doses.

Motor rpm to wind speed correlations indicated that to achieve desired wind

speeds, testing should proceed from ambient to 1500, to 3000, and finally 4500 rpm.

However, when testing the first raised tree, significant defoliation occurred at and above

3000 rpm. Thus the protocol was changed to proceed from ambient to 1250, to 2000, to

2750, back to 1250 rpm, and finally at ambient once more. Data was collected for two

minutes at ambient conditions and for four minutes at individual rpm. Changes in wind

speed occurred consecutively but data collection was interrupted as rpm changed.

Measurements from CETs, anemometers, and from a thermistor temperature probe

(Model 107 Campbell Scientific, Inc.) were taken at 0.5 Hertz. The temperature probe

was suspended 1.5 ft above ground on site and was protected from direct sunlight. The

data acquisition system (DAQ) consisted of a Campbell Scientific CR10X datalogger

used in combination with a Campbell AM 416 multiplexer (Figure 3.23) and a program

written in Loggernet 2.1 (Campbell Scientific, Inc.). The DAQ system was powered

from a standard 120 VAC socket through a 12 VDC converter.









Statistical Analysis

Pruning types were compared using analysis of variance (ANOVA) followed by

least squares means separations adjusted using Tukey's method. Pruning dose and wind

speed were analyzed in similar fashion and by regressions using a complete two factor

quadratic empirical model. Regressions on wind speed and pruning dose were used to

develop response surfaces of trunk movement from which orthogonal levels of each

parameter were extracted. Extracted values were used in a three way ANOVA to

compare effects of pruning type, pruning dose, and wind speed on tree movement. The

results and statistical analysis are covered in detail in chapter four. Data was analyzed

using the SAS system for windows release 8.02 ( 1999-2001 SAS Institute Inc., Cary,

NC, USA).












Table 3-1. Measurements used to determine physical similarity of trees tested.
Date Pruning Treeb MxchC Mnchd Tche Cd' Taper" Rdh Tw' Hvcc Areak Volume'
tested typea (ft.) (ft.) (ft.) (ft.) (in.) (lbs) (ft.) (ft.2) (ft.3)
19May04 RAO1 4 20.00 5.04 14.96 6.65 -0.04 31.00 660 18.28 312.10 1038.21
24May04 LTO1 3 19.30 4.88 14.43 6.30 -0.05 33.50 680 17.58 283.90 898.58
25May04 TH01 36 19.20 4.54 14.66 5.30 -0.03 29.75 530 17.31 233.19 646.24
26May04 RE01 28 22.10 5.29 16.81 5.70 -0.03 29.25 610 19.66 284.36 857.10
28May04 RA02 40 20.50 5.27 15.23 6.15 -0.04 32.75 680 18.30 287.40 904.03
01Jun04 LT02 24 20.40 4.65 15.75 6.00 -0.04 31.00 700 18.75 286.56 890.13
02Jun04 TH02 39 17.40 4.77 12.63 6.25 -0.04 30.50 680 15.75 254.80 774.27
03Jun04 RE02 30 21.40 4.71 16.69 6.30 -0.02 32.25 615 19.84 318.22 1039.77
04Jun04 RA03 27 19.20 4.75 14.45 5.80 -0.04 30.50 585 17.35 256.84 762.92
07Jun04 LT03 26 20.70 4.92 15.78 6.10 -0.04 30.75 625 18.83 292.73 921.76
08Jun04 TH03 5 19.70 4.56 15.14 5.95 -0.02 31.50 710 18.11 274.80 841.10
09Jun04 RE03 32 20.10 5.08 15.02 5.55 -0.03 29.75 630 17.79 251.19 725.97
10Jun04 RA04 41 19.00 5.04 13.96 6.15 -0.04 30.75 670 17.03 268.75 828.59
18Jun04 LT04 49 20.50 5.19 15.31 6.25 -0.04 30.75 645 18.44 294.36 938.78
21Jun04 TH04 1 20.60 4.58 16.02 5.85 -0.04 32.75 675 18.94 281.77 860.28
23Jun04 RE04 31 21.00 5.00 16.00 5.60 -0.03 31.00 595 18.80 267.38 787.51
28Jun04 RA05 42 19.90 4.60 15.30 6.20 -0.01 30.75 690 18.40 291.24 922.81
30Jun04 RA06 43 18.20 4.92 13.28 6.20 -0.04 30.75 630 16.38 261.63 801.40
01Jul04 RA07 37 20.80 4.71 16.09 6.30 -0.04 28.75 565 19.24 309.05 1002.40
31Aug04 RE05 22 19.80 5.13 14.68 6.40 -0.03 30.75 585 17.88 293.30 943.40
02Sep04 RE06 19 21.30 4.96 16.34 6.45 -0.05 31.00 640 19.57 321.87 1067.02
260ct04 LT05 44 20.80 4.52 16.28 6.20 -0.03 31.25 617 19.38 305.98 982.14
300ct04 LT06 21 20.90 5.00 15.90 5.90 -0.02 27.50 505 18.85 282.95 868.68
03Nov04 LT07 45 21.40 4.69 16.71 6.05 -0.03 30.50 622 19.74 303.56 960.09












Table 3-1. Continued.
Date Pruning


tested
04Nov04
10Nov04
11Nov04
23Nov04


type
ST01
ST02
ST03
RE07


Treeb MxchC Mnchd Tche Cdt Taperg Rdh Tw'
(ft.) (ft.) (ft.) (ft.) (in.) (lbs)
29 21.10 4.90 16.20 5.70 -0.03 30.00 610
11 19.80 4.67 15.13 6.05 -0.03 31.75 650
9 18.90 4.71 14.19 6.15 -0.03 29.50 580
38 19.50 4.71 14.79 6.20 -0.02 30.25 600
mean = 20.13 4.85 15.28 6.06 -0.03 30.73 628
Std = 1.051 0.226 1.03 0.30 0.01 1.24 49.9


Hvcc'
(ft.)
19.05
18.16
17.27
17.89
18.31
1.02


Areak
(ft.2)
275.87
280.34
272.15
283.75
283.22
20.89


Volume'
(ft.3)
826.29
869.37
842.44
892.40
881.92
97.58


aPruning type: Pruning treatment assigned (type and number within type). Pruning types are: LT= lion's tailing, RA = raising, RE = reduction, ST = structural,
TH = thinning.
b Tree: Number of original 50 selected from the field.
Mxch: Maximum canopy height = distance from the top of the football to the top of the canopy.
d Mnch: Minimum canopy height = distance from the top of the football to the origin of the lowest branch.
e Tch: Total canopy height. = Mxch Mnch.
f Cd: Canopy diameter = widest point of the canopy measured in two perpendicular directions and averaged. The main outline of the canopy was used to
determine diameter single shoots extending beyond the main body of the canopy were neglected
g Taper: Trunk taper parameter = -(R-r)/R where R = radius at 6 in. above the top-most root, and r = radius at 54 in. above the top-most root.
h Rd: Rootball diameter = average width of the root ball measured at the surface in two perpendicular directions.
1 Tw: Total weight of the tree and its football.
SHvcc: Height to vertical center of the canopy = ((0.5 Ch) + Mnch)
k Area: Canopy projected surface = 0.5 x the vertical surface area of a cone (whose dimensions are height = 0.667 x Ch and radius = 0.5 x Cd) + 0.25 x the
surface area of an ellipsoid (whose dimensions are radius of the long axis = 0.333 x Ch and radius of the short axis = 0.5 x Csd.
Volume: Canopy volume.= the volume of a cone (whose dimensions are height = 0.667 x Ch and radius = 0.5 x Cd) + 0.5 x the volume of an ellipsoid (whose
dimensions are radius of the long axis = 0.333 x Ch and radius of the short axis = 0.5 x Cd).













Table 3-2. Downwind profile of wind speeds generated by an airboat. Wind speeds were recorded for one minute at 100 Hz using gill
propeller anemometers (Model 27106R R.M. Young Company, Traverse City, MI, USA) and a proprietary data acquisition
program.


Test 1 (21Nov2003)
29 ft. 3 in. 36 ft. 3 in.
10ft. l in. 16 ft. 5 in.
Mean Max.c Mean Max.
5 7 2 2
10 15 2 3
10 20 3 5
12 32 5 10
15 28 7 12
28 28 3 3
33 33 4 4
37 37 5 5


Wind speed (mph)
Test 2 (21Nov2003)
39 ft. 8 in. 46 ft. 8 in.
10ft. 1 in. 16 ft. 5 in.
Mean Max. Mean Max.
3 5 2 3
9 15 3 5
12 17 3 5
20 25 4 9
25 32 7 15
26 34 7 10
25 32 3 5
25 38 4 7


Test 3 (21Nov2003)
39 ft. 8 in. 46 ft. 8 in.
10 ft. 1 in. 16 ft. 5 in.
Mean Max. Mean Max.
5 9 2 4
10 15 4 9
18 25 5 10
21 25 5 12
25 35 6 12
18 30 3 18
19 27 6 12
25 35 10 20


Wind speed (mph)
Test 4 (21Nov2003)
18 ft. 5 in. 25 ft. 5 in.
10ft. 1 in. 16 ft. 5 in.
Mean Max. Mean Max.
15 25 2 4
22 30 3 8
33 38 6 10
35 45 5 9
44 52 8 10
38 48 13 22
33 45 4 6
40 50 8 10


a Distance: Distance from the airboat propeller to the anemometer.
b Elevation: Elevation from the ground to the anemometer.
' Max.: Maximum wind speed recorded.


Distancea
Elevationb
RPM
1000
1500
2000
2500
3000
3500
4000
4500


Distance
Elevation
RPM
1000
1500
2000
2500
3000
3500
4000
4500










Table 3-3. Calibration of motor rpm to wind speed. Wind speeds were recorded for one
minute at 100 Hz using gill propeller anemometers (Model 27106R R.M.
Young Company, Traverse City, MI, USA) and a proprietary data acquisition
program.
Wind speed (mph)
Distance 17 ft. 0 in.
Elevationb 10 ft. 1 in.
RPM Mean Max.c Std. dev.
ambient 2.6 5.7 0.9
1000 9.1 15.0 2.9
1500 17.4 23.2 2.0
2000 24.5 29.4 2.4
2500 32.4 39.9 2.7
3000 38.4 46.8 2.6
3500 44.7 56.5 3.3
4000 50.0 60.4 3.7
4500 58.7 70.9 3.9
a Distance: Distance from the airboat propeller to the anemometer.
b Elevation: Elevation from the ground to the anemometer.
' Max.: Maximum wind speed recorded.









Table 3-4. Wind speeds recorded during a road test. Data was collected for 2 min. at 0.5
Hz using two 3-cup anemometers (Met One 034B, Campbell Scientific, Inc.,
Logan, UT, USA).
Wind speed (mph)
Westa (21Apr2004) East (21Apr2004)
Speedometer Mean Peak Std. dev. Mean Peak Std. dev.
10 9.7 10.5 0.7 9.1 12.3 1.2
20 20.7 23.0 1.0 22.1 25.7 1.7
30 30.3 32.8 0.9 30.6 33.7 1.7
40 40.2 47.1 2.4 42.5 44.5 1.0
50 50.6 52.5 1.1 52.5 53.4 0.6
60 60.6 63.2 1.3 64.8 67.7 1.2
70 73.5 78.5 1.5 75.7 80.2 2.2
a West and East refer to the location of the anemometer with respect to the tree during testing.














































igure 3.1. Three examples of lion's tail pruning taken from urban landscapes.







































Figure 3.2. Example of geometrically defined lion's tail pruning (lion's tailed tree number 1). Progression of targeted pruning doses:
A) no pruning; B) 15% pruning; C) 30% pruning; and D) 45% pruning.




























A


Figure 3.3. Example of visually defined lion's tail pruning (lion's tailed tree number 7).
Progression of targeted pruning doses: A) no pruning; B) 15% pruning; C)
30% pruning; and D) 45% pruning.















i1

J 1t t


gure 3.4. Two examples of raising pruning type taken from urban landscapes.







































C D
Figure 3.5. Example of geometrically defined raised pruning (raised tree number 3). Progression of targeted pruning doses: A) no
pruning; B) 15% pruning; C) 30% pruning; and D) 45% pruning.
















































C D
Figure 3.6. Example of visually defined raised pruning (raised tree number 7).
Progression of targeted pruning doses: A) no pruning; B) 15% pruning; C)
30% pruning; and D) 45% pruning.










K1
f~ ii


Figure 3.7. Two examples of reduction pruning type taken from urban landscapes.


.... ... .........


~---






































C D
Figure 3.8. Example of geometrically defined reduction pruning (reduced tree number 1). Progression of targeted pruning doses: A)
no pruning; B) 15% pruning; C) 30% pruning; and D) 45% pruning.

















































C D
Figure 3.9. Example of visually defined reduction pruning (reduced tree number 7).
Progression of targeted pruning doses: A) no pruning; B) 15% pruning; C)
30% pruning; and D) 45% pruning.
















































C
Figure 3.10. Two examples of structural pruning type taken from urban landscapes
before (A and C) and after (B and D) respectively.
















































SC D
Figure 3.11. Examples of structural pruning (structurally pruned tree number 1).
Progression of targeted pruning doses: A) no pruning; B) 15% pruning; C)
30% pruning; and D) 45% pruning.

















































M. -B
Figure 3.12. Example of thinning taken from urban landscapes before (A) and after
(B).








































C D
Figure 3.13. Example of thinning (thinned tree number 4). Progression of targeted pruning doses: A) no pruning; B) 15% pruning; C)
30% pruning; and D) 45% pruning.

















































iN.
"'
win-


ilk




r e 3 .. s.d > -

Figure 3.14. Airboat used to generate wind: A) side view and B) rear view.






56


100
90 + y=0.02x-2.00 West
80 x y=0.02x-4.59
X y=0.01x-0.64 East
E 70
60 +
un 50 ..- Gurley


30 -+ ..o'
20 /

10 ^


0 1000 2000 3000 4000 5000
Motor rpm

Figure 3.15. Calibration curves of the mean wind speeds as measured by Campbell
(West and East) and Young (Gurley) anemometers. West anemometer was
located 10 ft. 2 in. downwind; East anemometer 23 ft. 3 in. downwind; and
Gurley anemometry located 17 ft. downwind.















.. i '
* i ... .: ."


Figure 3.16. The four South cable extension transducers (CET).


)g













oy = 10.84x- 0.03
- y = 10.84x + 0.02
Sy = 16.05x- 0.04
Sy = 26.75x- 0.24


.0
.B


El'" ,8

0 A, --


s54 E",




s42









s30
s18


0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

mV

Figure 3.17. Southern cable extension transducer (CET) calibration curves. Labels
indicate the location of the CET with respect to the tree and propeller and the
elevation on the trunk during testing (e.g.: sl8 = southern CET elevated to 18
inches). R2 = 1.0 for all regressions. mV = millivolts; in. = inches










25 r
Sy= 10.86x 0.06 n54
-y= 10.82x- 0.03
20 1 Ay= 16.05x- 0.13 .'
0y = 26.73x- 0.19 '
n42
15 -


10 .Er




\ n30
Er & '---" n18
0 I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

mV

Figure 3.18. Northern cable extension transducer (CET) calibration curves. Labels
indicate the location of the CET with respect to the tree and propeller and the
elevation on the trunk during testing (e.g.: n18 = northern CET elevated to 18
inches). R2 = 1.0 for all regressions. mV = millivolts; in. = inches















Sailboat
/
anemomet
West (awsl







Primary direction
of wind flow.


North CETs (n18, n30, n42, n54)




anemometer
East (aesp),










South CETs (s18, s30, s42, s54)


anemometer


anemometer
West





South CETs


IB
Figure 3.19. Schematics (bird's-eye (A) and profile (B) views of cable extension
transducer (CET) and anemometer positions.


anemometer
East


























df er
<. ____ _____


n# CET c s# CET C

ai wire extension + average of n# CET readings at ambient by pruning dose
bi wire extension + average of s# CET readings at ambient by pruning dose
c distance between n# and s# CETs measured manually at set-up
Ai = Cos-l((bi2+c2-ai2)/2bic)
Bi = Cos-l((ai2+c2-bi2)/2aic)
yi = (biSinAi + aiSinBi)/2
di = yi/TanBi
ei = yi/TanAi

af wire extension + n# CET reading at designated time, rpm, and pruning dose
bf wire extension + s# CET reading at designated time, rpm, and pruning dose
c distance between n# and s# CETs measured manually at set-up
Af Cos-l((b2+c2-a2)/2bfc)
Bf = Cos-l((af+c2- b)/2afc)
yf = (bfSinAf + afSinBf)/2
df = yf/TanBf
ef = yf/TanAf

Ax# = di-df = ei-ef
Ay# = Yi-yf
m# = sqrt(Ax2+Ay2) = displacement of the trunk

Figure 3.20. Cable extension transducer calculations (these are for one specific height
represented in inches by subscript #).


di ei




















































B
Figure 3.21. Apparatus used to determine longitudinal Young's modulus: A) Coupon
test, and B) trunk test.



















































Figure 3.22. Apparatus used to fix the trunk and football of trees during testing. A) Steel
plate and basket used to fix the football. B) Tree fixed as it was during testing








mI A















Figure 3.23. Data acquisition system used in the field: A) Campbell Scientific CR10X
datalogger, B) AM 416 multiplexer and C) 12 VDC regulated power supply.














CHAPTER 4
RESULTS

Randomization

Trees were randomly assigned to a pruning type. Ten physical characteristics used

to select trees were used to compare trees assigned to pruning types. There were no

differences among pruning types with one exception. Elevation to the vertical center of

the canopy was statistically (P = 0.039) lower (143 in.) for trees assigned to the thinned

pruning type than for those assigned to the reduced pruning type (154 in.). Since this was

the only difference among types, no adjustments were made in the assignment of trees to

pruning types.

Pruning Dose

All pruning doses were calculated from an average canopy dry mass. Average

canopy dry mass was calculated from 13 of the 27 trees tested. Table 4-1 shows foliage,

stem, and total canopy dry mass for the 13 trees used to determine average canopy dry

mass. Foliage, stem, and total canopy dry mass for all trees were within three standard

deviations of the mean for each category. Thus none of the 13 trees used to determine

average canopy dry mass were obvious outliers as defined by the Empirical Rule in

statistics (Ott and Longnecker 2001).

Pruning doses were calculated as percentage of dry mass removed from the average

tree and from the actual tree where possible (Table 4-2). Table 4-2 also shows targeted

pruning dose and foliage, stem, and total dry mass removed at each pruning. As

expected, pruning doses calculated from average canopy dry mass were different than









those calculated from actual canopy dry mass for the 13 trees used to determine the

average. Difference in calculated dose was most apparent in the percent foliage dry mass

removed 3.5% on average with the greatest difference 13% for structurally pruned tree

number 3 at the 45% targeted dose level. To keep dose consistent in the analysis, dose

was calculated using the average canopy dry mass. Percent foliage dry mass was chosen

for use in the statistical analyses because ANSI (2001) defines pruning dose as a

percentage of foliage removed. Percent stem dry mass and percent total dry mass

removed from the canopy correlated well to percent foliage dry mass (R2 = 0.76 and 0.85

respectively). Figure 4.1 shows the similarity between pruning doses calculated from

foliage dry mass and those calculated from total dry mass.

Dose levels were more variable than expected. This was significant since

comparisons among pruning types depended on consistent dose levels among types.

Efforts to determine pruning dose using canopy dimensions were futile. Figure 4.1A

shows that foliage removed from lion's tailed trees one through three at all dose levels

was considerably less than the targeted dose levels. On the other hand, excess foliage

was removed from raised and reduced trees two through four at all dose levels. Lion's

tailed trees four through seven, and raised and reduced trees five through seven were

added to better approximate targeted pruning dose levels. Pruning dose levels for those

additional ten trees were determined in the field as a visual estimate of the foliage

removed. Dose levels for structurally pruned trees one through three and thinned trees

one through four were also determined visually and they were well distributed about the

targeted levels (Figure 4.1A). No additional trees were tested for structural or thinned

pruning types.









Wind Speed

Vertical wind profile was measured once before and twice after testing all 27 trees.

Vertical wind profile was calculated as an average of 120 measurements taken per

elevation per day on the three days it was evaluated (Figure 4.2A). Generated winds did

not disperse vertically. As motor rpm increased, vertical profile became more

concentrated (conical). One three cup anemometer was used to measure all elevations in

the vertical wind profile sequentially. The same three cup anemometer was used to

measure wind speeds at the vertical center of each tree canopy during testing. Figure

4.2B shows mean wind speeds recorded for all 27 trees at 2750 rpm on test dates as well

as average vertical wind profile measured prior to and post testing. Wind speeds

recorded on testing dates did not correspond with values in the average vertical wind

profile (Figure 4.2B). It should be noted that because generated winds did not disperse

vertically, upper and lower portions of tree canopies were subjected to lower wind speeds

than the center of a canopy.

An experiment-wise comparison of generated wind speeds showed that selected

motor rpm produced desired steps in wind speed (Table 4-3). The statistical difference

between wind speeds generated at 1250 motor rpm before and after higher rpm was

attributed to a large number of measurements used in comparisons. This statistical

difference had no substantial effect since the two means at 1250 rpm differed by less than

three miles per hour, while means between separate rpm differed by nearly 15 miles per

hour.

Wind speeds were compared among pruning types by motor rpm. Raised trees

generally experienced the highest generated wind speeds while lion's tailed and

structurally pruned trees experienced lower wind speeds at every rpm (Table 4-4). Least









squares means adjusted with Tukey's method indicated that at the highest motor rpm,

wind speeds were statistically different among all pruning types except among thinned

and reduced types (Table 4-4). Average wind speed recorded for lion's tailed trees

differed from that recorded for reduced trees by 11.47 mph. This variation in generated

wind speeds was corrected for in the analysis.

Wind speeds were compared among trees and pruning dose levels by rpm. Time

series plots show that recorded wind speeds were not consistent between trees (Figure

4.3A and B). Wind speeds were fairly consistent within an rpm on some days (Figure

4.3A and B) but showed tremendous variation within an rpm on others (Figure 4.3C and

D). Table 4-5 shows wind speeds recorded for each tree at the highest motor rpm.

Response

Two variables were considered to represent tree response to wind loading. The first

was trunk movement at an elevation 54 inches above the topmost root (m54 Figure

4.4A). The second was the area below the deflected trunk in the plane of primary wind

flow (dya Figure 4.5A). Trunk movement (m54) and deflected area (dya) were

perfectly correlated with each other (R2 = 1.0 at highest motor rpm and no foliage

removed) and both were comparably correlated with wind speed (R2 = 0.26 and 0.27

respectively at highest motor rpm and no foliage removed). Both measures showed that

trunk movement tracked changes in wind speeds relatively well within a tree (Figures

4.4B and 4.5B). Both also showed there were differences in trunk movement among

trees independent of wind speed. Mean wind speed at the highest rpm (Table 4-5) was

decidedly greater for lion's tailed tree number 2 (LT02) than it was for lion's tailed tree

number 1 (LT01), but m54 and dya were both lower for LT02 than for LT01. Deflected

area appeared more responsive to changes in wind speeds but it was also more variable









than m54 and it provided a measure of trunk movement in only one horizontal direction,

that parallel to the primary wind flow. Trunk movement (m54) provided a measure of

deflection in both horizontal directions, parallel and perpendicular to the primary wind

flow and it was most responsive of all individual measurements. Therefore, m54 was

chosen for use in the analysis and dya was considered redundant.

Trunk movement (m54) proved to be an effective measure of tree response

regardless of quality of the corresponding wind speed profile. Time series plots of m54

(Figure 4.6) showed better resolution of individual rpm than time series plots of wind

speed (Figure 4.3). They also showed effect of pruning dose as reduced movement with

each repeated sequence of motor rpm (Figure 4.6).

Tree testing date was expected to influence trunk movement. Trees were tested

from May 2004 to November 2004 so there was ample time for growth and development

of trunks. Trees tested early were blocked in time by type but additional lion's tailed

(numbers 4-7), raised (numbers 5-7), and reduced (numbers 5-7) trees and all structurally

pruned trees were not blocked in time. Figure 4.7 shows that trunk movement (m54), at

the highest motor rpm and no foliage removed, appeared to decrease with time from May

through November while wind speeds did not (R2 = 0.53). However, comparison of

lion's tailed trees tested early in the year with those tested later showed no differences (P

= 0.664) in m54 attributable to date tested. There was not enough evidence to conclude

that tree growth had a significant impact on trunk movement over the dates tested.

Longitudinal Young's modulus was expected to influence trunk movement as well.

Results from efforts to determine Young's modulus of whole trunks and coupons were

inconclusive (Trachet 2005), so efforts were abandoned and Young's modulus was









assumed to be uniform within and among trees tested. Structurally pruned tree number

two suffered significant water stress which provided reason to believe its Young's

modulus differed from that of other trees. It was left out of the analysis and there were

no other anomalies that knowingly might have caused variation in Young's modulus

among trees. As a result, the assumption of material homogeneity seems appropriate for

the trees tested.

Analytical Approach

A principal goal of this experimental study was to compare effects of pruning type

on tree response to wind load. Because pruning dose and wind speed were not recorded

at set levels, orthogonal levels of both were sought to simplify comparisons between

pruning types. Trunk movement (m54) was regressed against wind speeds (Wind) and

pruning doses (Dose) measured for each tree (Tables 4-6 and 4-7). A complete two

factor quadratic empirical model was used for all regressions as given in Equation 4.1,

where bo b5 are constants generated as the regression coefficients.

m54 = b0 + bWind + b2Dose + b3Wind2 + b4Dose2 + bWind x Dose (4.1)

A quadratic model was chosen because wind speed is accounted for in the standard

equation used to calculate drag by the fluid velocity term (which is squared) and pruning

dose results in a reduction in surface area (also a squared term). Two regressions were

conducted: one using all measurements of wind speed and m54 (Table 4-6), and one

using measurements of wind speed and m54 averaged within pruning type, tree, pruning

dose and rpm (Table 4-7). Figure 4.8 shows a graphical representation of all m54

measurements and average m54 measurements for raised tree number 5. Regressions

using averaged m54 had higher R2 values than those using all measurements and they









were simpler since pruning dose coefficients were almost entirely insignificant. All

regressions using averages of wind speed and m54 had R2 values in excess of 0.94.

Regression equations were used to predict trunk movement (pm54) for all trees at

orthogonal levels of pruning dose and wind speed. For some trees predicted values were

interpolated between large gaps in the data (Figure 4.9A). For others predicted values

were extrapolated some distance from the data (Figure 4.9B and C). Figure 4.10 gives a

graphical summary of the procedure used to generate predicted average m54 (pm54)

values for an individual tree.

The pm54 values were used to compare pruning type, pruning dose and wind speed

in a three-way analysis of variance (ANOVA). Table 4-8 shows that there was

essentially no difference between ANOVA results using pm54 values predicted from

equations using all measurements of m54 and wind speed, and those using the averages

m54 and wind speed. Similarly, there was little difference between ANOVAs using m54

versus those using deflected area (dya). Table 4-8 confirms that average m54 was as

good a measure of tree response as m54, deflected area (dya), or average dya. Average

trunk movement (m54) was used to complete the analysis.

The three-way ANOVA using predicted trunk movement (pm54) (Table 4-9)

showed that interaction between pruning type, pruning dose, and wind speed, and

interaction between pruning type and pruning dose were statistically insignificant (P =

1.00 and P = 0.74 respectively). However, interactions between pruning type and wind

speed (Figure 4.11) and between pruning dose and wind speed (Figure 4.12) were

significant (P < 0.0001 for both). Throwing out early lion's tailed, raised and reduced









trees that had extreme dose levels (Figure 4.1A) did not affect the results. Further

analysis was conducted to quantify the significance of the interactions.

Predicted trunk movement (pm54) was used to compare pruning types at each wind

speed averaged across pruning dose. Table 4-9 shows least squares means, adjusted with

Tukey's method, ordered by pruning type at each targeted wind speed. Predicted trunk

movement of thinned trees was statistically greater than the pm54 of structurally pruned,

raised, and lion's tailed trees at 45 mph and greater than all other trees at 60 mph. There

were no statistical differences between pruning types at lower wind speeds. These results

are seen clearly in the pruning type x wind speed interaction profile plot (Figure 4.11).

Thinning was the least effective pruning type for reducing wind load and there were no

differences between the other four types.

Predicted trunk movement (pm54) was then used to compare wind speeds within

each pruning type averaged across pruning dose. Table 4-10 shows least squares means,

adjusted with Tukey's method, ordered by pruning dose at each targeted wind speed.

Predicted trunk movement increased for all pruning types as wind speeds increased. The

increase in movement was statistically significant (P < 0.05) at all wind speeds for all

pruning types except structural pruning. In structurally pruned trees, pm54 was only

statistically different between 15 and 60 mph wind speeds. There was no statistical

difference in movement between 15 to 45 mph or between 30 to 60 mph wind speeds

Overall, increases in wind speed resulted in increased trunk movement. Regression

models (Table 4-7) also show this as coefficients for wind and wind2 terms were almost

always significant and positive.









The pruning dose x wind speed interaction (Table 4-11) was analyzed in the same

fashion as pruning type x wind speed. At 30 mph predicted trunk movement was only

reduced with a 60% pruning dose compared to no pruning. At the two highest wind

speeds, the 30% dose was similar to both unpruned an the 60% pruning dose, with only

pruning doses of 45% or greater predicting less trunk movement than no pruning. At

both highest wind speeds, 60% pruning dose reduced predicted trunk movement about

50% compared to no pruning. This is represented graphically in Figure 4.12. Therefore,

pruning dose affected tree response to wind load but the effect was not statistically

significant at the low wind speeds and it was only statistically significant at the highest

wind speeds once the 45% targeted pruning dose level was reached.

Separations of pm54 among wind speeds at each pruning dose level (averaged

across pruning type) showed a similar effect. Least squares means, adjusted with

Tukey's method, ordered by wind speed at each targeted pruning dose are shown in Table

4-12. The effect of pruning dose is first seen at the 30% targeted dose level as pm54 at

15 mph was not statistically different than pm54 at 30 mph. At the highest targeted dose

level pm54 at 30 mph was not statistically different than pm54 at 15 or 45 mph -

indicating that pruning dose increasingly offset the effect of wind speed. Prediction

models showed this as well (Table 4-7). Although the dose and dose2 coefficients are

mostly insignificant, the dose*wind coefficients are almost all significant and negative. It

is significant that pruning dose did not reduce trunk movement until 30% of the foliage

was removed, and that further reduction in pm54 required a doubling of that dose (60%

foliage dry mass removed). Both of these doses exceed current recommended pruning

dose levels (ANSI 2001).













Table 4-1. Foliage and stem weight for 13 trees harvested to quantify pruning dose.


Pruning Tree Remaining Pruned Sum foliage Remaining
typea no.b foliage (lbs) foliaged (lbs) (lbs) stemf (lbs)
LT 4 2.568 2.171 4.739 26.780
LT 5 2.789 2.150 4.939 36.792
LT 6 3.098 1.857 4.955 35.255
LT 7 1.773 2.140 3.913 31.058
RA 3 1.202 3.425 4.627 14.627
RA 4 0.808 4.233 5.041 14.140
RA 6 2.116 2.837 4.952 19.387
RA 7 2.207 2.189 4.396 23.709
RE 5 2.114 2.011 4.125 29.844
RE 6 2.410 1.860 4.270 33.096
RE 7 1.564 2.262 3.827 31.056
ST 1 2.162 2.356 4.519 30.900
ST 3 0.975 2.683 3.658 26.632
mean = 4.458
std. err.j = 0.131
a Pruning types: LT = lion's tailing, RA = raising, RE = reduction, ST = structural, TH = thinning.
b Tree no.: Number assigned to a tree within a pruning type (LT 4 was the fourth tree lion's tailed).
Remaining foliage: Foliage dry mass remaining on the tree after the heaviest pruning dose.
d Pruned foliage: Foliage dry mass removed with all pruning doses.
e Sum foliage: Sum of remaining and pruned foliage.
f Remaining stem: Stem dry mass remaining on the tree after the heaviest pruning dose.
g Pruned stem: Stem dry mass removed with all pruning doses.
h Sum stem: Sum of remaining and pruned stem.
1 Sum total: Sum foliage plus sum stem.
SStd. err.: Standard error of 1 standard deviation.


Pruned
sterm (lbs)
5.147
5.706
5.035
5.277
12.730
16.426
11.562
8.621
3.086
3.520
4.047
8.637
8.930
mean =
std. err =


Sum stem
(lbs)
31.927
42.498
40.290
36.335
27.357
30.566
30.949
32.329
32.930
36.616
35.102
39.536
35.562
34.769
1.199


Sum total'
(lbs)
36.666
47.436
45.246
40.248
31.984
35.608
35.901
36.725
37.055
40.885
38.929
44.055
39.220
39.227
1.201














Table 4-2. Percent foliage dry weight, stem dry weight, and total dry weight removed with each pruning dose.


Foliage dry mass removed


Pruning
type"
LT
LT
LT
LT
LT
LT
LT
LT
LT
LT
LT
LT
LT
LT
LT
LT
LT
LT
LT
LT
LT


Tree
no.b
1
1
1
2
2
2
3
3
3
4
4
4
5
5
5
6
6
6
7
7
7


Stem dry mass removed


Actual
dosef (%)


targeted
dose" (%)
15
30
45
15
30
45
15
30
45
15
30
45
15
30
45
15
30
45
15
30
45


Average
dose (%)
5
7
11
1
7
12


Actual
dose (%)


Total dry mass removed


Sumd
(lbm)
0.532
0.759
1.328
0.202
0.817
1.336
0.505
0.895
1.448
0.459
1.073
2.171
0.572
1.320
2.150
0.560
1.215
1.857
0.635
1.273
2.140


Average
dosee (%)
12
17
30
5
18
30
11
20
32
10
24
49
13
30
48
13
27
42
14
29
48


Actual
dose (%)


Sum
(Ibm)
1.598
2.496
3.933
0.476
2.465
4.051
1.544
2.624
3.625
1.231
2.844
5.147
1.878
4.117
5.706
1.975
3.740
5.035
1.902
3.579
5.277


Sum
(Ibm)
2.130
3.255
5.261
0.678
3.282
5.386
2.049
3.519
5.074
1.690
3.918
7.318
2.450
5.437
7.856
2.536
4.955
6.893
2.537
4.852
7.417


Average
dose (%)
5
8
13
2
8
14
5
9
13
4
10
19
6
14
20
6
13
18
6
12
19














Table 4-2. Continued


Foliage dry mass removed


Sumd
(lbm)
3.720
2.583
3.450
3.786
2.153
2.564
3.425
1.858
3.571
4.233
1.070
1.839
2.810
1.101
2.071
2.837
0.735
1.527
2.189


Average
dosee (%)
83
58
77
85
48
58
77
42
80
95
24
41
63
25
46
64
16
34
49


Stem dry mass removed


Actual
dosef (%)


Sum
(Ibm)
16.800
10.291
13.743
14.724
9.615
10.744
12.730
6.632
14.203
16.426
3.549
6.320
11.940
3.782
7.365
11.562
2.459
5.819
8.621


Average
dose (%)
48
30
40
42
28
31
37
19
41
47
10
18
34
11
21
33
7
17
25


Total dry mass removed


Actual
dose (%o)


Sum
(Ibm)
20.520
12.875
17.194
18.510
11.768
13.309
16.155
8.489
17.774
20.659
4.619
8.159
14.751
4.882
9.436
14.399
3.194
7.346
10.809


Average
dose (%)
52
33
44
47
30
34
41
22
45
53
12
21
38
12
24
37
8
19
28


Actual
dose (%)


Pruning
typea
RA
RA
RA
RA
RA
RA
RA
RA
RA
RA
RA
RA
RA
RA
RA
RA
RA
RA
RA


Tree
no.b
1
2
2
2
3
3
3
4
4
4
5
5
5
6
6
6
7
7
7


Targeted
dose" (%)
45
15
30
45
15
30
45
15
30
45
15
30
45
15
30
45
15
30
45














Table 4-2. Continued


Foliage dry mass removed


Pruning
typea
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE
RE


Tree
no.b
1
1
1
2
2
2
3
3
3
4
4
4
5
5
5
6
6
6
7
7
7


Stem dry mass removed


Actual
dosef (o)


Targeted
dose" (%)
15
30
45
15
30
45
15
30
45
15
30
45
15
30
45
15
30
45
15
30
45


Sumd
(lbm)
1.159
1.757
2.239
1.834
2.992
3.651
1.326
2.472
3.252
1.665
2.808
3.789
0.648
1.333
2.011
0.603
1.410
1.860
0.802
1.531
2.262


Total dry mass removed


Actual
dose (%)


Average
dosee (%)
26
39
50
41
67
82
30
55
73
37
63
85
15
30
45
14
32
42
18
34
51


Sum
(lbm)
0.939
2.225
4.717
1.948
4.830
8.808
1.052
3.034
6.440
1.347
3.514
7.826
0.641
1.595
3.086
0.659
2.197
3.520
0.929
2.235
4.047


Actual
dose (%)


Average
dose (%)
3
6
14
6
14
25
3
9
19
4
10
23
2
5
9
2
6
10
3
6
12


Sum
(lbm)
2.099
3.983
6.956
3.782
7.822
12.458
2.378
5.506
9.692
3.012
6.322
11.615
1.289
2.929
5.097
1.261
3.607
5.380
1.731
3.766
6.309


Average
dose (%)
5
10
18
10
20
32
6
14
25
8
16
30
3
7
13
3
9
14
4
10
16














Table 4-2. Continued


Tree Targeted
no.b dose' (%)


Foliage dry mass removed
Sumd Average Actual
(Ibm) dosee (%) dosef (%)
0.864 19 19
1.691 38 37


Pruning
typea
ST
ST
ST
ST
ST
ST
ST


Stem dry mass removed


Sum Average
(Ibm) dose (%)
2.408 7
4.931 14


52 8.637
- 3.564
7.501
- 9.972
24 2.825
51 5.706
73 8.930
- 1.365
- 3.046
- 4.203
- 0.969
- 2.171
- 4.213
- 1.504


Total dry mass removed


Actual
dose (%)
6
12


Sum
(Ibm)
3.272
6.621


25 22 10.993
10 4.437
22 -9.041
29 11.948
8 8 3.693
16 16 7.566
26 25 11.612
4 2.272
9 4.827
12 6.584
3 1.754
6 3.749
12 6.783
4 2.238


TH 3 30 1.639 37 2.715 8 4.355 11 -
TH 3 45 2.607 58 4.843 14 7.451 19 -
TH 4 15 1.007 23 1.796 5 2.803 7 -
TH 4 30 1.970 44 3.943 11 5.913 15 -
TH 4 45 2.994 67 7.256 21 10.250 26 -
a Pruning types: LT = lion's tailing, RA = raising, RE = reduction, ST = structural, TH = thinning.
b Tree no.: Number assigned to a tree within a pruning type (i.e. LT 4 is the fourth tree lion's tailed).
Targeted dose: Percentage of foliage dry mass intended to be removed by pruning.
d Sum: Total dry mass removed at specified pruning dose (foliage, stem, and total respectively). Dry mass was summed incrementally within a tree (i.e. sum at
targeted dose 45 = sum of the dry mass removed at 15, 30 and 45 % levels).
e Average dose: Dose calculated from an average tree canopy dry mass [foliage (4.458 Ibm), stem (34.769 Ibm), and total (39.227 Ibm) respectively]
f Actual dose: A value in this column indicates that dose was calculated from the actual tree canopy dry mass (foliage, stem, and total respectively). A dash
indicates the entire tree was not weighed.


45 2.356
15 0.873
30 1.539
45 1.976
15 0.869
30 1.860
45 2.683
15 0.907
30 1.781
45 2.381
15 0.786
30 1.578
45 2.570
15 0.735


Average
dose (%)
8
17


Actual
dose (%)
7
15
25




9







79


Table 4-3. Wind speeds (mph) recorded during testing and least squares means (P =
0.05) adjusted with Tukey's method.
Motor rpma
1 1251 2000 2750 1252 2
n 6480 12956 12929 12960 12960 6463
Meanb 2.70 ac 23.26 b 38.05 c 52.51 d 20.99 e 2.93 a
Std. err. 0.02 0.06 0.09 0.12 0.06 0.02
Max 10.47 34.62 53.40 75.77 32.83 12.26
a Motor rpm 1 and 2 are ambient conditions and 1251 and 1252 are 1250 rpm before and after the higher
rpm respectively.
b Mean is an average of all measurements within an rpm across pruning type, tree, and pruning dose.
c Means with the same letter within rows are not significantly different (P < 0.05) based on LS mean
separations adjusted using Tukey's method


Table 4-4. Wind speed (mph) by pruning type and motor rpm.
Pruning 1a Pruning 2750


Meanc Max.
2.72 + 0.03 10.47
2.16 + 0.04 9.57
2.43 + 0.03 6.89
4.03 + 0.06 9.57
2.93 + 0.05 8.68

1251
Mean Max
21.27 0.13 34.62
25.93 + 0.10 33.73
23.80 +0.11 33.73
19.10 + 0.20 32.83
24.91 + 0.10 33.73


type
LT
RA
RE
ST
TH


Pruning
type
LT
RA
RE
ST
TH


Mean Max
47.25 0.27 ae 75.77
58.72 0.19 b 73.08
52.71 0.24 c 72.19
50.37 0.41 d 72.19
53.64 0.19 c 72.19

1252
Mean Max
18.54 + 0.12 32.83
24.55 0.08 31.04
21.45 0.11 31.94
17.44 0.18 31.04
21.81 0.12 31.94


Pruning 2000 Pruning 2
type Mean Max type Mean Max
LT 34.92 0.19 53.40 LT 3.02 0.04 12.26
RA 42.22 0.15 52.51 RA 2.42 0.04 10.47
RE 37.58 + 0.17 53.40 RE 2.68 0.03 9.57
ST 34.80 + 0.29 50.72 ST 4.06 0.05 7.79
TH 40.53 + 0.15 52.51 TH 3.16 + 0.05 9.57
a Motor rpm 1 and 2 are ambient conditions and 1251 and 1252 are 1250 rpm before and after the higher
rpm respectively.
b Pruning types: LT = lion's tailing, RA = raising, RE = reduction, ST = structural, TH = thinning.
Mean: Mean standard error of one standard deviation with n from 720 to 3360 per pruning type.
dMax: Maximum value recorded.
e Means with the same letter within columns are not significantly different (P < 0.05) based on LS mean
separations adjusted using Tukey's method


typeb
LT
RA
RE
ST
TH

Pruning
type
LT
RA
RE
ST
TH











Table 4-5. Wind speed, trunk movement (m54), and deflected area (dya) at 0 pruning
dose and 2750 rpm.


Pruning Tree
type" no.b
LT 1
LT 2
LT 3
LT 4
LT 5
LT 6
LT 7
RA 2
RA 3
RA 4
RA 5
RA 6
RA 7
RE 1
RE 2
RE 3
RE 4
RE 5
RE 6
RE 7
ST 1
ST 2
ST 3
TH 1
TH 2
TH 3
TH 4


Wind speed (mph) m54 (in.)
Std. Std.
Meanc err.d Maxe Mean err. Max
47.50 0.50 60.56 3.05 0.03 4.19
66.59 0.54 73.98 2.96 0.04 4.05
36.74 0.57 50.72 2.48 0.02 3.15
33.63 0.91 54.30 2.76 0.02 3.33
28.63 0.76 58.77 1.46 0.02 2.16
41.00 0.93 63.24 1.39 0.01 1.75
51.50 0.46 60.56 1.87 0.02 2.51
62.69 0.32 69.51 4.08 0.03 4.74
61.42 0.50 70.40 2.29 0.01 2.67
63.68 0.61 72.19 4.57 0.02 5.14
65.06 0.25 71.29 3.62 0.02 4.04
58.96 0.79 70.40 2.36 0.01 2.76
66.00 0.41 71.29 2.81 0.02 3.23
52.74 0.91 66.82 2.76 0.03 3.45
65.11 0.35 70.40 4.13 0.02 4.54
48.62 1.03 70.40 3.56 0.03 4.65
61.14 0.32 67.72 2.98 0.01 3.30
56.49 0.55 65.93 2.17 0.02 2.69
39.01 1.18 61.46 1.91 0.01 2.23
63.01 0.28 70.40 2.57 0.02 3.05
65.97 0.20 69.51 2.21 0.01 2.50
32.59 0.87 56.98 1.21 0.01 1.42
42.74 0.93 62.35 2.11 0.04 2.87
55.33 0.21 63.24 4.23 0.03 5.28
52.13 0.35 60.56 3.21 0.02 3.85
40.28 0.31 48.93 3.00 0.02 3.67
61.73 0.77 72.19 3.60 0.02 4.09


Mean
106.73
102.27
87.50
95.81
51.37
48.89
66.01
150.95
80.84
157.42
127.52
85.08
99.81
97.76
145.99
124.42
106.51
76.33
67.90
91.18
77.85
42.44
72.53
146.49
111.81
104.18
126.35


dya (in. )
Std.
err. Max
1.22 145.91
1.26 137.73
0.69 111.86
0.66 118.30
0.54 75.40
0.37 61.93
0.84 90.24
1.10 177.49
0.48 95.89
0.80 177.72
0.62 144.44
0.53 99.33
0.60 116.27
1.13 122.59
0.62 161.65
1.10 162.83
0.52 117.54
0.89 94.85
0.40 81.81
0.69 107.84
0.53 88.33
0.37 50.20
1.27 99.75
1.19 184.15
0.71 137.40
0.75 129.55
0.87 143.72


a Pruning types: LT = lion's tailing, RA = raising, RE = reduction, ST = structural, TH = thinning.
b Tree no.: Number assigned to a tree within a pruning type (i.e. LT 4 is the fourth tree lion's tailed).
' Mean: Average of 120 measurements.
d Std err: Standard error of one standard deviation.
eMax: Maximum value recorded.













Table 4-6. Regression coefficients and R2 values generated using all measurements of trunk movement (m54) and wind speed (Wind)
within a pruning type, tree, pruning dose (Dose), and rpm treatment combination in a complete two factor quadratic
empirical model. R2 values for regressions of deflected area (dya) are given for comparison.


Pruning
type
LT
LT
LT
LT
LT
LT
LT
RA
RA
RA
RA
RA
RA
RE
RE
RE
RE
RE
RE
RE


Tree
b
no.
1
2
3
4
5
6
7
2
3
4
5
6
7
1
2
3
4
5
6
7


Dose
0.0047193
-0.0159439
-0.0065756
-0.0098421
-0.0026469
-0.0085027
-0.0065309
NS
0.0017010
NS
0.0021254
0.0049636
0.0024453
0.0051511
0.0054225
NS
0.0051231
-0.0044869
NS
NS


Wind2
0.0004759
0.0001968
-0.0001017
0.0000748
-0.0003689
0.0001034
-0.0000014
0.0002678
0.0000995
0.0003028
0.0003331
0.0001484
0.0001711
0.0001259
0.0004122
0.0001248
0.0002828
0.0001827
-0.0002288
0.0003363


Dose2
-0.0002174
0.0005834
0.0001145
0.0001746
0.0000586
0.0001678
0.0001463
0.0000184
0.0000141
-0.0000210
-0.0000616
-0.0000387
-0.0000527
-0.0001726
-0.0000259
-0.0000479
-0.0000796
0.0000754
NS
-0.0001186


Wind*Dose
-0.0010233
-0.0004388
-0.0007349
-0.0007114
-0.0003598
-0.0003292
-0.0006956
-0.0006647
-0.0004418
-0.0005730
-0.0003397
-0.0004692
-0.0003323
-0.0005428
-0.0004044
-0.0004766
-0.0003062
-0.0002095
-0.0002129
-0.0000891


Intercept
NSe
0.0521557
NS
0.0678023
-0.1399732
0.0562677
NS
NS
-0.0642992
0.1265204
NS
-0.0916750
NS
0.0473252
-0.0933176
NS
NS
0.0489778
NS
0.0492041


Wind
0.0394638
0.0283154
0.0590989
0.0590676
0.0564957
0.0228512
0.0351883
0.0478202
0.0296015
0.0482748
0.0345920
0.0293212
0.0311609
0.0432073
0.0357488
0.0596352
0.0333231
0.0269396
0.0530362
0.0508709


R 2
(m54)c
0.925
0.911
0.805
0.818
0.761
0.883
0.836
0.952
0.955
0.961
0.975
0.909
0.953
0.909
0.895
0.904
0.966
0.926
0.747
0.958


R2
(dya)d
0.924
0.917
0.805
0.818
0.759
0.884
0.834
0.949
0.950
0.962
0.974
0.909
0.951
0.905
0.894
0.901
0.962
0.924
0.747
0.956













Table 4-6. Continued.
Pruning Tree R2 R2
typea no.b Intercept Wind Dose Wind2 Dose2 Wind*Dose (m54)c (dya)d
ST 1 -0.1543650 0.0288068 0.0037946 0.0001217 -0.0000384 -0.0003226 0.959 0.957
ST 2 NS 0.0405986 -0.0057574 -0.0002261 0.0000806 -0.0004180 0.751 0.744
ST 3 -0.0680923 0.0411714 NS 0.0001394 NS -0.0004593 0.870 0.865
TH 1 0.0695847 0.0402283 -0.0091875 0.0005953 0.0001810 -0.0006801 0.935 0.932
TH 2 -0.0586411 0.0366951 NS 0.0004529 0.0000266 -0.0004189 0.956 0.952
TH 3 NS 0.0468575 NS 0.0003885 0.0000353 -0.0005330 0.897 0.897
TH 4 -0.1534129 0.0405974 0.0035574 0.0002915 NS -0.0004013 0.933 0.935
a Pruning types: LT = lion's tailing, RA = raising, RE = reduction, ST = structural, TH = thinning.
b Tree no.: Number assigned to a tree within a pruning type (LT 4 was the fourth tree lion's tailed).
R2 (m54): regressions using all wind speed and m54 measurements within pruning type, tree, pruning dose, and rpm.
d R2 (dya): regressions using all wind speed and dya measurements within pruning type, tree, pruning dose, and rpm.
e NS: Not statistically significant at P = 0.05.












Table 4-7. Regression coefficients and R2 values generated using averages of trunk movement (avm54) and wind speed (Wind) within
a pruning type, tree, pruning dose (Dose), and rpm treatment combination in a complete two factor quadratic empirical
model. R2 values for regressions of average deflected area (avdya) are given for comparison.


Pruning
type
LT
LT
LT
LT
LT
LT
LT
RA
RA
RA
RA
RA
RA
RE
RE
RE
RE
RE
RE
RE


Tree
b
no.
1
2
3
4
5
6
7
2
3
4
5
6
7
1
2
3
4
5
6
7


R2
(avm54)d
0.991
0.986
0.949
0.951
0.994
0.980
0.983
0.980
0.979
0.976
0.995
0.979
0.994
0.975
0.948
0.990
0.989
0.983
0.947
0.986


R2
(avdya)e
0.991
0.989
0.950
0.952
0.994
0.981
0.983
0.978
0.976
0.979
0.995
0.979
0.994
0.975
0.949
0.990
0.989
0.984
0.947
0.986


Dose2
NS
NS
NS
NS
NS
0.0001765
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
-0.0000629
NS
NS
NS


Wind*Dose
-0.0012533
-0.0004317
-0.0011734
-0.0008994
-0.0004278
-0.0005026
-0.0004130
-0.0006488
-0.0004305
-0.0005782
-0.0003064
-0.0004774
-0.0002541
-0.0006638
-0.0002876
-0.0004837
-0.0003246
NS
-0.0003410
-0.0001350


Intercept
NSt
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS


Wind
0.0346330
0.0249140
0.0496499
0.0588409
0.0203197
0.0206290
0.0326324
0.0460744
0.0272470
0.0467960
0.0300226
0.0247701
0.0232733
0.0441449
0.0251932
0.0463334
0.0326210
0.0200714
0.0363065
0.0200764


Dosec
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS


Wind2
0.0006744
0.0002559
NS
NS
0.0010574
0.0002513
NS
0.0002787
0.0001357
0.0003508
0.0004070
0.0002553
0.0003050
0.0002051
0.0005598
0.0004676
0.0003099
0.0003093
0.0004618
0.0003667













Table 4-7. Continued.
Pruning Tree R2 R2
typea no.b Intercept Wind Dosec Wind2 Dose2 Wind*Dose (avm54d) (avdyae)
ST 1 NS 0.0266158 NS 0.0001453 NS -0.0003157 0.985 0.986
ST 2 NS 0.0345746 NS NS NS -0.0006789 0.968 0.969
ST 3 NS 0.0275358 NS 0.0005033 NS -0.0004889 0.988 0.988
TH 1 NS 0.0365198 NS 0.0006660 NS -0.0006502 0.988 0.989
TH 2 NS 0.0337152 NS 0.0005155 NS -0.0004285 0.993 0.993
TH 3 NS 0.0313188 NS 0.0008477 NS -0.0006126 0.982 0.983
TH 4 NS 0.0299773 NS 0.0005137 NS -0.0004197 0.986 0.986
a Pruning types: LT = lion's tailing, RA = raising, RE = reduction, ST = structural, TH = thinning.
b Tree no.: Number assigned to a tree within a pruning type (LT 4 is the fourth tree lion's tailed).
Dose: Percentage of foliage dry mass removed from the tree with pruning.
d avm54: Wind speed and m54 measurements averaged within pruning type, tree, pruning dose, and rpm.
e avdya: Wind speed and dya measurements averaged within pruning type, tree, pruning dose, and rpm.
f NS: Not statistically significant at P = 0.05.













Table 4-8. ANOVA of predicted trunk movement (p avm54) and predicted deflected area (p avdya).
p m54a (in.) p avm54b (in.) p dyac (in.2) p avdya (in.2)
Source of variation F P>F F P>F F P>F F P F
Pruning type 41.49 < 0.001 21.94 < 0.001 42.16 < 0.001 21.98 < 0.001
Pruning dosef 83.49 < 0.001 66.18 < 0.001 84.22 < 0.001 67.11 < 0.001
Pruning type pruning dose 0.58 0.898 0.75 0.739 0.62 0.870 0.78 0.708
Wind speedg 395.29 < 0.001 399.48 < 0.001 396.02 < 0.001 402.90 < 0.001
Pruning type wind speed 7.37 < 0.001 3.69 < 0.001 7.26 < 0.001 3.54 < 0.001
Pruning dose wind speed 5.96 < 0.001 5.02 < 0.001 5.93 < 0.001 5.05 < 0.001
Pruning type pruning dose 0.07 1.000 0.12 1.000 0.07 1.000 0.12 1.000
wind speed
a p_m54: predicted trunk movement based on all measurements within pruning type, tree, pruning dose and motor rpm.
b p_avm54: predicted trunk movement based on measurements averaged within pruning type, tree, pruning dose and motor rpm.
c p_dya: predicted trunk deflected area based on all measurements within pruning type, tree, pruning dose and motor rpm.
d p_avdya: predicted trunk deflected area based on measurements averaged within pruning type, tree, pruning dose and motor rpm.
e Pruning types: LT = lion's tailing, RA = raising, RE = reduction, ST = structural, TH = thinning.
f Pruning dose (percentage of foliage dry mass removed) levels at which average trunk movement was predicted from regression models.
g Wind: wind speed (mph) levels at which average trunk movement was predicted from regression models.










Table 4-9. Least squares means of predicted average trunk movement (p_avm54) due to interaction of pruning type and wind speed -
type by wind speed.
p avm54a (in.)
Typeb 15 (mph)' 30 (mph) 45 (mph) 60 (mph)
ST NSd NS 1.31ae 2.05a
RA NS NS 1.50a 2.25a
LT NS NS 1.51a 2.39a
RE NS NS 1.77ab 2.70a
TH NS NS 2.07b 3.32b
a p_avm54: Average trunk movement predicted from regression models.
b Pruning types: LT = lion's tailing, RA = raising, RE = reduction, ST = structural, TH = thinning.
Wind speed (mph) levels at which average trunk movement was predicted from regression models.
d NS: Not statistically significant at P = 0.05.
e Means with the same letter within columns are not significantly different (P < 0.05) based on LS mean separations adjusted using Tukey's method.


00
Table 4-10. Least squares means of predicted average trunk movement (p_avm54) due to interaction of pruning type and wind speed C
wind speed by type.
p avm54a (in.)
Windb Lion's tailing Raising Reduction Structural Thinning
15 0.34a' 0.40a 0.41a 0.28ab 0.41a
30 0.82b 0.89b 1.00b 0.72abc 1.10b
45 1.51c 1.50c 1.77c 1.31bcd 2.07c
60 2.39d 2.25d 2.70d 2.05cd 3.32d
a p_avm54: Average trunk movement predicted from regression models.
b Wind speed (mph) levels at which average trunk movement was predicted from regression models.
Means with the same letter within columns are not significantly different (P < 0.05) based on LS mean separations adjusted using Tukey's method.










Table 4-11. Least squares means of predicted average trunk movement (p_avm54) due to interaction of pruning dose and wind speed
dose by wind speed.
p avm54a (in.)
Doseb 15 (mph)' 30 (mph) 45 (mph) 60 (mph)
60 NSd 0.51abe 1.Olab 1.71ab
45 NS 0.70abc 1.31abc 2.11abc
30 NS 0.90abcd 1.62bcd 2.53bcd
15 NS 1.llbcd 1.94cde 2.96cde
0 NS 1.32cd 2.27de 3.40de
a p_avm54: Average trunk movement predicted from regression models.
b Dose: Pruning dose (percentage of foliage dry mass removed) levels at which average trunk movement was predicted from regression models.
' Wind speed levels at which average trunk movement was predicted from regression models.
d NS: Not statistically significant at P = 0.05.
e Means with the same letter within columns are not significantly different (P < 0.05) based on LS mean separations adjusted using Tukey's method.


00
Table 4-12. Least squares means of predicted average trunk movement (p_avm54) due to interaction of pruning dose and wind speed --
wind speed by dose.
p avm54a (in.)
No foliage dry mass 15 % foliage dry 30 % foliage dry 45 % foliage dry 60 % foliage dry
Windb removed' mass removed mass removed mass removed mass removed
15 0.56ad 0.46a 0.36a 0.27a 0.19a
30 1.32b 1.11b 0.90a 0.70a 0.51ab
45 2.27c 1.94c 1.62b 1.31b 1.01b
60 3.40d 2.96d 2.53c 2.11c 1.71c
a p_avm54: Average trunk movement predicted from regression models.
b Wind speed (mph) levels at which average trunk movement was predicted from regression models.
"Pruning dose (percentage of foliage dry mass removed) levels at which average trunk movement was predicted from regression models.
d Means with the same letter within columns are not significantly different (P < 0.05) based on LS mean separations adjusted using Tukey's method.














90

80


S70

o 60

U)
C 50
E
2
0 40 -

30-
U-o

20 -


10 -

0

0 xqx 0< 0, 0<, 0, O Ob J0 0 0 0 0^ xx ^

Tree
A


Figure 4.1. Pruning dose represented as percentage of foliage dry mass removed (A) and percentage of total dry mass removed (B)
with respect to an average tree canopy mass. Black bar = 15%, gray bar = 30%, and white bar = 45% targeted pruning
dose levels. LT = lion's tailing, RA = raising, RE = reduction, ST = structural, TH = thinning pruning types. Number
represents a tree assigned to the specific pruning type (27 trees total).