Abstract
Bark-included junctions are frequently encountered defects within the aerial structures of trees. The presence of included bark within a branch junction can substantially reduce the junction’s factor of safety. Recent research has found naturally occurring bracing to be a primary cause of the formation of included bark within branch junctions. This study tested the load-bearing capacity of branch junctions in hawthorn (Crataegus monogyna Jacq.) using rupture tests and compared the mechanical performance of “control” branch junctions, bark-included junctions with the natural bracing retained, and bark-included junctions where we had intentionally removed their natural braces by cutting them out. Substantial variability was observed in the failure kinematics of bark-included branch junctions when their natural braces were retained. The type of natural brace present affected the mode of failure of the branch junctions when pulled apart. A single specimen with fused branches presented the strongest form of natural brace in this study, followed by entwining branches, whereas crossing branches were found to provide the least mechanical resistance. This study provides initial evidence that the type of associated natural brace is an important consideration when an arborist is trying to assess the likely mechanical performance of a bark-included junction within a tree and its likelihood of failure.
INTRODUCTION
A tree’s ability to withstand both static and dynamic loading is developed through acclimation to the strains exerted upon components within its structure, much of which is explained by the process of thigmomorphogenesis (Jaffe 1973; James 2003; Jungnikl et al. 2009). This mechano-perception directs the alteration of physiological, biochemical, and morphological characteristics of a tree in response to mechanical perturbations (Telewski and Pruyn 1998). Common responses in trees when components are exposed to elevated levels of strain are increased radial thickening of limbs, formation of an elliptically shaped limb cross section in the direction of flexure, and decreased overall tree height. Such acclimations act to counterbalance the levels of strain experienced and reduce the likelihood of structural failure in a healthy tree (Pruyn et al. 1998).
Branch junctions are a necessary component within nearly all trees’ aerial structures. Some authors have often broadly categorised branch junctions into two forms: branch-to-stem attachments and stem-to-stem attachments (Gilman 2003; Smiley 2003; Kane 2007). However, this dichotomy cannot be a formal botanical division as both forms of branch junction are topologically equivalent despite their morphological differences (Slater and Harbinson 2010). The conjoining of branches to stems and stems to stems is a developmental continuum typically differentiated externally by measuring the diameter ratio between the two attached limbs and internally differentiated by the extent of embeddedness of the base of the smaller-diameter branch into the larger-diameter stem (i.e., the length of the knot forming the distal end of the smaller branch that is occluded within the xylem of the larger stem)(Figure 1). The extent to which one branch’s base is occluded into the tissues of the other is reported to give the branch junction greater load-bearing capacity than where no occlusion occurs (Kane et al. 2008).
Branch junctions from the crown structure of a silver birch (Betula pendula Roth.). A: Codominant branch junction; B: Branch-to-stem branch junction. Note that both types of branch junction exhibit a visually observable branch bark ridge (BBR) at the apex of the junction (white arrows), so a BBR is a shared anatomical feature. Internally, the branch-to-stem junction will exhibit significant occlusion of the base of the smaller branch into the larger stem, whereas the codominant branch junction will most often exhibit little to no occlusion of the tissues of one branch into the other.
A spectrum of diameter ratios occurs at branch junctions, and no strict cut-off point can be easily justified to differentiate between two distinct forms of branch junction in a scientific discourse. For the purposes of this study, we use the general term “branch junction” and treat the diameter ratio of the tested specimens as a continuous variable, rather than applying an arbitrary numeric division between “low” and “high” diameter ratio groups.
As well as the potential embedding of one branch’s tissues into the limb it is attached to, another important strengthening tissue at a branch junction is located within its axil, consisting of specialised centrally placed xylem tissues, the presence of which is most often expressed externally by a branch bark ridge (Figure 1). Dense, tortuous xylem is typically formed centrally within the axil of a branch junction in a woody plant (Shigo 1985; Lev-Yadun and Aloni 1990; Kramer 1999; Slater et al. 2014; Slater and Ennos 2015a), which this study refers to as “axillary wood” hereafter. The heightened density and the orientation of the tortuous wood grain pattern in this axillary wood confer substantial mechanical support but forego most hydraulic conductivity in this part of the branch junction (Lev-Yadun and Aloni 1990; Slater et al. 2014).
Bark-Included Branch Junctions
Bark inclusions are commonly encountered defects found at the junction apex in a wide range of tree species (Slater 2018b). A bark-included branch junction (hereafter, “BI junction”) is characterised by a seam of bark, which is enveloped within a branch junction, whereby the usual formation of axillary wood is disrupted (Figure 2). The presence of included bark has been found to significantly lower the load-bearing capacity of such branch junctions (Smiley 2003; Slater and Ennos 2015b).
A comparison of the distinct morphologies of branch junctions: normally formed; cup union; wide-mouthed bark inclusion; cracked.
A substantial proportion of above-ground failures has been attributed to BI junctions; for example, a survey conducted after the Great Storm of 1987 within southeast England reported that 20% of crown failures recorded across 18 species in parkland locations related to BI junctions (Gibbs and Grieg 1990).
Slater and Ennos (2015b) identified three distinct categories of BI junction (named “wide-mouthed,” “cup union,” and “embedded bark” in that study), based upon their morphology and the location of the included bark in relation to the apex of the junction (Figure 2). Their study found that small zones of embedded bark (a patch of included bark that has been fully occluded by secondary growth) did not significantly reduce the bending strength of BI junctions when compared with normally formed branch junctions in hazel (Corylus avellana L.) and that cup-shaped BI junctions were substantially stronger than those found with included bark at their apex. They conclude by suggesting that the gradations in bark-inclusion morphology (from wide-mouthed to cup-union or embedded bark) represent the sequential “repairing” of a defective BI junction. These findings may serve as an additional tool to the arborist in a visual tree risk assessment when determining the relative risk of a BI junction. Slater (2016, 2018b) provides a table giving a generalised qualitative likelihood of failure for each of these junction types, based on this research.
Various authors have suggested that there is an association between the angle of inclination of arising branches and the proclivity of a junction to contain included bark (e.g., Mattheck and Vorberg 1991; Mattheck and Breloer 1994; Lonsdale 1999). Though there is not a substantial body of evidence substantiating this claim, it is logical that there should be a relationship between an acute angle of branch attachment and the formation of included bark. The frequent near-vertical disposition of the branches arising from BI junctions would produce a much lower level of static load acting across the junction, in comparison with that of a branch junction where either arising branch (or both branches) was set at a more horizontal inclination.
Vertically aligned arising branches, however, are still likely to be affected by dynamic movement induced by the wind; therefore, it seems unlikely that solely the angle of inclination is the determining factor in the formation of a BI junction (Slater 2018a).
A more plausible explanation for the formation of BI junctions is due to a much lower “loading regime” as a branch junction develops compared with a normally formed branch junction. This can occur in the presence of “natural braces,” which substantially restrict movement at the branch junction set below them and subsequently may cause the branch junction to develop weakly. Natural braces, as defined by Slater (2018a), have been found to be strongly associated with the occurrence of BI junctions in several tree species. This finding provides a rational explanation for the association between BI junctions and near-vertical arising parallel stems, as it would be more likely that there would be an occurrence of physical contact between lateral branches arising from such stems, or the entwining of those stems, leading to the formation of one or more natural braces.
Natural Bracing in Trees
Slater (2018a) conducted a study of a cohort of broadleaved trees and reported a very strong association (> 93%) between the occurrence of BI junctions and the presence of “natural braces” set above these BI junctions in the trees surveyed (Figure 3).
The occurrence of a natural brace (top arrow) formed directly above a BI junction (bottom arrow). Here, the natural brace is a fusion of two adjacent limbs, which is acting to arrest the movement of the junction below, substantially reducing the mechanical perturbations experienced across the BI junction.
Natural braces are typically a configuration of touching/rubbing/fusing branches and/or stems, which act to restrict the movement of a branch junction lower down in the tree, although natural braces can also be formed that involve climbing plants, aerial roots, or foreign objects (Slater 2018b). If there is firm contact between the interacting limbs forming the natural brace, this may substantially lessen the strain experienced at the branch junction apex set below this natural brace. This lack of strain can result in a lowered rate of growth of the denser (Slater and Ennos 2015a), tortuously grained (Slater et al. 2014) axillary wood—so much so that instead of this important connective tissue being grown, bark-onbark contact replaces it (Slater 2016).
Where a natural brace is lost, either through self-shading or pruning, compensatory growth in the form of bulging around the bark inclusion is induced in response to the new loading regime that acts upon the BI junction (Figure 4). Slater (2018a) identified a very strong association (> 93%) between the loss of a natural brace and the occurrences of discernible bulging in xylem tissues proximal to the bark inclusion.
An illustration of the typical process where a natural brace is present and then subsequently lost through self-shading (its loss may otherwise be attributed to pruning). The BI junction begins to repair by forming compensatory tissue proximal to the bark inclusion where the vascular cambium is responding to mechanical perturbations. The formation of a sufficiently strong branch junction is likely if the continuation of compensatory secondary growth is achieved.
Natural bracing in trees can take many forms. Slater (2016) describes nine distinct types of natural brace (to which the aerial rooting of some species of tropical/subtropical trees could be added):
Fused limbs
Entwining stems
Entwining branches
Crossing branches
Crossing lateral branches that touch an adjacent stem
Intermeshing twigs
Resting stems
Climbing plants acting as braces
Complex braces (incorporating any other object)
Every natural brace found in a tree is different in its configuration, which gives rise to very different considerations and outcomes when assessing the likelihood of failure of the associated BI junction. Four natural brace types were tested within this study: “fused branches”; “entwining stems”; and “crossing branches”/“crossing lateral branches that touch an adjacent stem” (the latter two categories hereafter grouped together as “crossing branches”)(Figure 5).
Illustration of four different forms of natural brace: A: Fused limbs; B: Entwining limbs; C: Crossing limbs; D: Lateral branch crossing to be in physical contact with an adjacent stem.
Fused branches (Figure 5A) can be described as the fusion of sapwood between two limbs, which occurs through the process of “anastomosis” or “inosculation” (Cook and Northey 2012). The connecting of the xylem in the two branches in this way typically makes a mechanically strong join in a short time after fusion, and thus it is regarded as the most sustainable form of natural brace, often persisting in a tree’s structure for its lifetime. Slater (2018b) regards the typical likelihood of failure of a BI junction associated with fused branches set above it as “very low” and its typical sustainability rating as “very good.”
Entwined stems (Figure 5B) are adjacent stems that have entwined around each other by 90° or more, such that they have “swapped positions” as the stems extend upward and outward. The entwining of the stems prevents major tensile forces from wind loading acting at the apex of the associated branch junction. Slater (2018b) regards the typical likelihood of failure of a BI junction associated with this brace as “low” to “very low” and the brace’s sustainability as “very good.”
Crossing branches (Figure 5C and D) are described as tree limbs that are interacting with one another either in a pressing or rubbing relationship, substantially restricting the typical swaying motion of the associated limbs. Slater (2018a) found this form of natural brace to be the most common within a large cohort study (n = 575). The sustainability of crossing branches as a natural brace is heavily dependent upon the size and position of the interacting components, as the branches’ survival are linked to effective light-capture by their foliage. The form of contact also influences the longevity of this type of natural brace, as a rubbing relationship can result in mechanical damage and the exposure of sapwood, the subsequent weakening of the damaged areas of such branches, and potentially the ingress of decay organisms into those damaged areas. For these reasons, a natural brace consisting of crossing branches that exhibits a firm bark-on-bark contact is regarded as more sustainable than one where movement and friction are causing substantial physical damage to either branch. This form of natural brace, referred to as “crossing/rubbing branches” in industry guidance, is often cited as a defect that should be removed when pruning a tree (e.g., BSI 2010; Gilman 2011; ANSI 2017). Without an appropriate assessment of the associated junction’s morphology and the sustainability of the natural brace, the removal of crossing branches may induce a subsequent BI junction failure lower down in the tree’s structure. Slater (2018b) categorises the typical likelihood of failure of a BI junction associated with this type of natural brace as “medium” to “low” and the brace sustainability as “average” to “very poor”—so it is advised that this type of natural brace typically performs poorly compared with fusion or entwining. However, this assessment is currently only based upon the observations of one author with no empirical data to support or refute it.
Study Rationale
Further research into BI junctions and their associated natural braces would inform tree risk management decisions and assist in developing our understanding of which configurations of natural braces in trees are more sustainable and which are likely to be problematic.
This study assessed the bending strength of BI junctions in Crataegus monogyna Jacq. (common hawthorn) in the presence and absence of natural braces. Crataegus has been used in this study because it is a frequently encountered genus, so the results will be of some direct relevance to arborists. In addition, the species frequently contains BI junctions associated with natural braces (Slater 2018a), and the harvesting of the required branch junctions was sustainable due to the abundance of semimature C. monogyna within the chosen site and these trees’ ability to recuperate from such pruning.
This study sought to address the following three research questions:
Does the presence of a natural brace impart a higher-than-average load-bearing capacity to a BI junction in comparison to one without a natural brace?
Do differing forms of natural braces result in differing modes of failure for the associated branches and BI junctions?
Is the density of the wood located at the apices of normal branch junctions and BI junctions significantly different from the density of the wood in the adjacent stem (which may explain some differences in load-bearing capacity)?
MATERIALS AND METHODS
Sample Collection
On 13 January and 26 February 2018, a total of 74 C. monogyna branch junctions (no more than 3 specimens taken from any individual tree) were harvested from Scutcher’s Acres, Ormskirk, Lancashire, England, by permission of the owner (Grid reference: SD452106). Branch junctions were cut from young to semimature hawthorn trees of ages ranging between 7 to 30 years.
All branch junctions collected gave rise to two branches of approximately equal diameter emerging from a single parent stem (i.e., a bifurcation). Thirty-seven BI junctions were harvested for use as treatments (i.e., these were naturally braced and had a discernible bark inclusion at the junction apex), and thirty-seven branch junctions were harvested for use as controls (i.e., normal branch junctions with no bark inclusion evident, a normally formed branch bark ridge and not associated with any natural braces).
The branch junction specimens were harvested using a Japanese pull-saw, retaining at least 100 mm of the parent stem below each bifurcation collected and retaining at least 180 mm of each arising branch above the bifurcation. Arising branches were cut with this minimum length to allow for drill-hole placement prior to rupture testing. For the same purposes, naturally braced specimens were cut to retain at least 120 mm of the arising branches above the natural brace present.
Quantities of each natural brace type collected and details of how treatment groups were assigned are provided in Table 1. After harvesting, specimens were placed in folded plastic bags with identity records and kept within an unheated outbuilding (c. 2 to 3 °C) to prevent sap loss prior to rupture testing within 10 days of their collection.
Natural brace classifications, quantity of samples, and groupings of branch junction specimens collected from Scutcher’s Acres.
Preparation of Samples for Mechanical Testing
BI junction specimens with natural braces present (n = 37) were randomly selected for the two treatment groups: rupture testing with the natural brace retained or with the natural brace removed. Twenty BI junctions were selected for testing with the natural brace retained (although for “crossing branches,” lateral branches forming the brace often had to be shortened to make the rupture testing practicable), and seventeen BI junctions were selected for testing with the natural brace removed. The distance from the natural brace to the junction apex was recorded before the natural braces were removed from the seventeen specimens in this treatment group. The removal of natural braces was achieved using either a pair of secateurs or a Japanese pull-saw to cut away lengths of smaller-diameter lateral branches that were acting as a natural brace.
All branch junctions were prepared for rupture testing by drilling a 6-mm-diameter hole in each of the two arising branches, set perpendicular to the plane of the bifurcation. Branch junctions in which the natural brace had been removed had their two arising branches drilled at approximately 200 mm from the junctions’ apices; where the natural brace was retained, the arising branches were drilled approximately 100 mm above the position of the natural brace. Any excess branch length above the drill holes was removed. A steel rule was used to measure the distances between the two drill holes (A to B) and from each drill hole to the apex of the branch junction (A to C and B to C) to calculate the internal angle of the bifurcation (Figure 7).
Diagram of a specimen bolted to two U-shaped brackets in the testing machine, illustrating the measurements of branch lengths that facilitated the calculation of the internal angle of each branch junction and other parameters needed to calculate the peak relative load (Pr) from each rupture test.
Rupture Testing
Each branch junction was secured between the crosshead and the base of an Instron™ universal testing machine 3369, using two bespoke U-shaped brackets (Figure 7). Steel bolts of 5-mm diameter were inserted through the pre-drilled holes to secure the arising branches within the centre of each U-bracket. During the test, the crosshead of the testing machine was set to continuously move vertically at a speed of 35 mm min−1 until the branch junction or one of the associated branches failed. The force exerted was measured using a 50-kN load cell while an interfaced computer plotted a graph of force against displacement. The peak force and displacement at peak force were both recorded.
Observations were made during testing to assess the mode of failure for each branch junction. Three failure categories were recorded, as defined and recorded by Slater and Ennos (2013): Type 1, whereby the specimen fails first by yielding under compression at the outer edge of the smaller-diameter branch and then a crack is initiated in the branch junction’s apex; Type 2, whereby the branch junction fails under tension directly at the branch junction’s apex and a central crack is initiated; and Type 3, whereby failure occurs only in the smaller-diameter branch (initially yielding under compression and then failing in tension) and the associated branch junction does not fail during testing. Any anomalies observed in specimens or during testing were also recorded.
Measurements Taken Post Rupture Testing
Following rupture testing, key dimensions of each branch junction were measured (Figure 6). Measurements of the over-bark diameters of the parent stem (PS1 and PS2) were taken just below the branch bark ridge using digital callipers, in line with and perpendicular to the plane of the bifurcation. Similarly, the diameters of each arising branch were measured, also using digital callipers, both in line with and perpendicular to the plane of the bifurcation proximal to the branch junction (A1, A2, B1, and B2). Each arising branch from the junction was designated as either “Branch A” or “Branch B,” Branch A being the larger-diameter branch and Branch B being the smaller-diameter branch. The diameter ratio was calculated by comparing the diameters of Branch A and Branch B in line with the bifurcation (i.e., the ratio of B1 to A1). The distance between the bifurcation of the pith and the apex of the branch junction (h) and the diameter of the stem at the point of the bifurcation of its pith were also measured using digital callipers.
Measurements of the branch junctions taken in this study to calculate peak relative load (Pr). The larger-diameter branch was defined as “Branch A” and the smaller-diameter branch as “Branch B.” The diameter of the parent stem was measured in line with the bifurcation (PS1) and perpendicular to it (PS2) at the base of the visible branch bark ridge. The diameters of both arising branches were also measured in line with the bifurcation (A1; B1) and perpendicular to it (A2; B2) above any swelling proximal to the branch junction and away from any swelling associated with it. The external height of the branch bark ridge (H2) was also recorded for each specimen.
Specimens that had a natural brace retained had additional measurements taken that involved recording the diameters of the components of the natural brace at their point of contact.
Estimating Peak Loading
The moment needed to break each sample, Mp, was calculated using equation (1):
where Fp is the peak force recorded, b is the distance from the centre of the drill hole in Branch B to the apex of the bifurcation, and Ɵ is the angle at which the peak force was applied relative to the bearing of length b (Figure 6).
The angle Ɵ was calculated using equation (2):
where a + ∂a is the distance between the two drill holes at the time when the peak force was recorded, and b and c are the linear distances between the apex of the branch bark ridge and the drill holes in the smaller and larger arising branches of the sample, respectively (Figure 6). The peak relative load (Pr) experienced at the branch junction was estimated using equation (3):
where B1 and B2 are the diameters of the smaller-diameter branch proximal to the branch junction in line and perpendicular to the bifurcation (Figure 7). This estimation of peak loading has been used previously for the tensile testing of branch junctions (e.g., Gilman 2003; Kane et al. 2008; Slater and Ennos 2015b), but it is now recognised that this estimation method does not satisfactorily quantify the mechanical stress at the branch junction apex at the point of failure (Slater and Ennos 2013). To allow for meaningful comparisons with previous studies, equation (3) has been used in this study and normalises the loading of the branch junctions tested in relation to their size and dimensions.
Wood Density Testing
Small cuboid wood samples (approximately 6 × 6 × 6 mm) were excised from the apices of tested branch junctions and from the periphery of the associated parent stems for 15 control specimens, then weighed. Similarly, for the treatment groups, a small wood sample was cut from the periphery of the parent stem; however, included bark was present at the junction apices of all the treatment specimens, so wood samples from those junctions had to be excised from the connecting tissue formed to the side of the included bark. Again, a total of 15 cuboid samples from stems and sides of junctions were harvested and weighed for the BI junctions with natural braces originally present.
The samples were then oven-dried for 96 hours at 60 °C. After the elapsed 96 hours, a random selection of 10 samples was removed from the oven and weighed. Samples were then returned to the oven for a further hour and then weighed again to ensure they were fully dry. The dry weights of all the sample cuboids were measured using a weighing scale accurate to 10 mg. The volume of each wood sample was calculated by measuring the displacement weight when each sample was immersed in a beaker of distilled water at room temperature on a weighing scale (hydrostatic weighing) with the density of the water estimated at 998 kg/m−3. Wood density (ρ) was then calculated using equation (4):
The gravimetric water content of each sample was calculated by subtracting the sample’s dry weight from its fresh weight and dividing that value by the sample’s fresh weight.
Statistical Analysis
Minitab v. 18 was used for statistical analysis of the data from this study. Mean values are reported with their associated standard error (SE). Where the means from two groups were compared and data was normally distributed, two-sample t-tests were used. The relationship between the continuous variable of diameter ratio and peak loading was analysed using a linear regression. ANOVAs were used to identify differences in the sample means of more than two groups. Significant differences between sample groups were tested using post hoc Tukey tests after each ANOVA, setting the confidence level at 95%. Residuals from the ANOVAs were assessed for normality using the Kolmogorov-Smirnov test. Where data was not normally distributed, a Kruskal-Wallis test was used to assess whether there were significant differences in sample medians, and post hoc Mood’s median tests were used to identify significant differences between groups.
RESULTS
Parameters of Specimens Tested
The diameter ratio of the arising branches in control specimens (mean: 81% ± 2% SE) and treatment specimens (mean: 80.4% ± 3.5% SE) was not found to be significantly different between test groups (F2, 71 = 0.73, P = 0.487).
A one-way ANOVA showed that there were significant differences in the mean internal angle of control and treatment specimens (F2, 71 = 41.11; P < 0.001). The internal angle of branch junctions (between A and C and B and C [Figure 7]) ranged from 13.3° to 71.3° in control specimens (mean: 42.4° ± 2.4° SE), ranged from 4.3° to 33.7° (mean: 20.9° ± 1.7° SE) for BI junctions with the natural brace removed, and ranged from 2.9° to 32.8° (mean: 13.2° ± 2.15° SE) for BI junctions with a natural brace retained. A post hoc Tukey test confirmed that the mean internal angle of control specimens was significantly higher than either BI junction group.
Rupture Testing
Regression analysis found a significant relationship between peak relative load (Pr) and diameter ratio (R2 = 0.219; P < 0.001) for all specimens tested, with higher peak loads recorded for branch junctions exhibiting lower-diameter ratios. The residuals were tested using a Kolmogorov-Smirnov test and were found to be normally distributed (KS = 0.047; P > 0.150).
The results of a one-way ANOVA identified that there was a significant difference in mean Pr between the three groups subjected to rupture tests (F2, 71 = 3.22; P = 0.046)(Figure 8). The residuals were found to be normally distributed (KS = 0.051; P > 0.150). Post hoc comparisons using a Tukey test (CI = 95%) indicated that the mean value for the control group (44.19 MPa ± 1.69 SE) differed significantly from that of the BI junctions tested with the natural brace removed (35.76 MPa ± 3.35 SE). The mean value of those BI junctions where the natural brace was retained (39.69 MPa ± 3.06 SE) did not differ significantly from the mean Pr of the other two groups.
Mean Pr of the three groups of branch junctions that were rupture tested: control (n = 37), natural brace removed (n = 19), and natural brace retained (n = 16). Error bars represent standard error of the mean. Letters denote heterogeneity between groups. Groups sharing the same letter were not significantly different from each other.
Failure Mode
Data for the type of failure that occurred for each branch junction subjected to rupture tests were recorded in relation to sample groups (Table 2). Statistical analysis for this data set was not undertaken due to insufficient replicates within four of the categories.
Different failure modes observed in specimens tested by treatment group, reported by count (N) and percentage of specimens in each group that failed in that mode.
Of the 12 BI junctions rupture tested with the natural brace “crossing branches” retained, 8 exhibited Type 2 failures whereby a crack was initiated directly at the junction apex, whereas 2 specimens exhibited Type 1 failures and a further 2 exhibited Type 3 failures.
The Effect of Natural Bracing on Failure Kinematics
Crossing branches was the most common and also the only natural brace type that was numerous enough to use for statistical analysis to determine if there was a significant difference in mean Pr between the two groups: “natural brace retained” and “natural brace removed.” A two-sample t-test identified that there was not a significant difference (T26 = −0.65; P = 0.518) between mean Pr of BI junctions with crossing branches/stems retained (39.8 MPa ± 3.7 SE) and crossing branches/stems removed (36.5 MPa ± 3.4 SE).
However, the related bending load/displacement graphs (examples provided in Figure 9) identified distinct variation in the failure kinematics when the natural brace was present compared to when it had been removed. When the natural brace was absent, the graph plots followed a simple curvilinear path until peak force was reached, followed closely by the point of fracture. In contrast, when the natural brace was retained on the sample, the failure behaviour was markedly more complex, the jagged plotlines indicating where the natural brace caused additional resistance to the pulling apart of the branch junction (Figure 9B).
Load/displacement graphs of typical specimens in each category. A: The testing of a control branch junction that failed by cracking at its apex (red), contrasted with a branch failure (green), where the plastic deformation phase is much longer and peak load higher. B: The testing of three BI junctions, one with no natural brace present (blue), one with crossing branches bracing the junction (grey), and one with entwining branches (orange). For specimens with natural braces left in situ, the plotlines juddered as the natural brace provided resistance to displacement.
Figure 9 shows the variability in the failure behaviour in response to the different forms of branch junctions tested. The bending apart of a control junction invariably followed a curvilinear plot until peak relative load (Pr) was reached, preceding the point of fracture, with a relatively short plastic yielding phase (Figure 9A). Where a branch failure occurred in a control specimen, this was typically at a higher relative load and after a comparably longer plastic yielding phase (Figure 9A).
The failure of BI junctions with their natural braces removed also gave rise to a curvilinear plot, again typically exhibiting only a short plastic yielding phase before a Type 2 failure occurred in most specimens (Figure 9B).
The example shown of the rupture testing of entwining stems (Figure 9B) began with a normal linear elastic phase until firm contact occurred between the arising branches. This contact generated a nonlinear, jagged plot caused by friction acting between the branches. After much displacement, the entwined branches were eventually prised apart, and the associated BI junction failed almost immediately afterward in this specimen. In all other specimens with their entwinement retained (n = 4), one of the arising branches failed instead of the BI junction, as the branches were too intertwined to be separated by this testing method.
The example for the crossing branch retained specimen provided in Figure 9B also demonstrates a normal linear elastic phase until firm contact with the natural brace increased resistance to displacement. The jagged plotline indicates where the natural brace was providing strong resistance to the pulling apart of the BI junction. This resistance was lost once the crossing branch slipped away from the branch it touched, and the plotline then returns to one typical of the failure of a BI junction. Due to machine-testing parameters, the crossing branches on these specimens had to be shortened, and this likely reduced their effectiveness as a natural brace compared with if they could have been kept at their full length. This mechanical behaviour during testing explains why mean Pr was not significantly different between the specimens with and without crossing branches present. An assessment of the kinetic energy needed to break these specimens may have been more appropriate to differentiate the behaviour of these two contrasted groups.
Only one of the BI junctions subjected to rupture testing had a fused natural brace retained, and this specimen also exhibited entwining stems (Figure 10). The fusion of the two arising stems was positioned approximately 30 mm below the placement of the drill hole and the arising branches entwined at least 90° around each other. The concerted resistance of these two natural braces against the BI junction being pulled apart resulted in failure occurring at the drill hole instead of at the natural brace or at the BI junction. The peak relative load (Pr) was 133.9 MPa, the highest recorded from this testing. Unfortunately, no other fused natural braces could be found within the hawthorn trees in Scutcher’s Acres to achieve any statistically meaningful comparisons for this natural brace type.
Specimen that exhibited fusion of branches between the two limbs (indicated by an arrow), which we found to be uncommon in hawthorn.
Wood Density Testing and Water Content
The median wood density for samples excised from the control apex was 876.0 kg/m−3, whereas the median density of the sample cut from the stem was 774.5 kg/m−3. Within BI junctions, wood samples removed from the apex had a median value of 777.9 kg/m−3, whereas median wood density of samples from the stem were 743.6 kg/m−3. The data collected for wood density was found not to be normally distributed, therefore, a Kruskal-Wallis test was used. This test identified that there were significant differences in mean wood density between the groups tested (H = 16.01; P = 0.001). A post hoc Mood’s median test identified that there were significant differences in the medians (χ2 = 15.73; P = 0.001), as identified in Figure 11.
Box and whisker plot of wood density sampled from four different locations (n = 15 × 4). The whiskers denote the range, limits of the boxes show the upper and lower quartiles, asterisks indicate outliers, and the midline is the median value. Letters above groups identify heterogeneity as identified using a post hoc Mood’s median test.
The gravimetric water content of test samples was 45.54% ± 0.19 SE for stem samples and 40.14% ± 0.17 SE for samples from the axils of the branch junctions, identifying that the specimens had a high moisture content when rupture tested. A two-sample t-test identified that the moisture content of the denser axillary wood was significantly lower than that of the xylem extracted from the stems (T57 = 20.93; P < 0.001).
Observations on Bark-Included Junctions
Observations of the exposed fracture surfaces of the BI junctions after the rupture testing revealed that these branch junctions had been normally formed for at least a few growing seasons before the formation of included bark within each junction. This was determined by counting the annual rings formed from the point of the bifurcation of the pith to the occurrence of the included bark in each specimen (Figure 12). None of the 37 BI junctions tested had started its development with included bark formed inside it.
Fracture surfaces of three BI junctions after rupture testing. Note the occurrence of the included bark (red arrows) appears after the junction being normally formed for several years; indicated by the annual increments formed from the bifurcation of the pith (green arrow) up to the occurrence of the included bark. Also note that the specimen on the left-hand side of this image had one minor area of embedded bark and grew normally again for a few years before more bark was included in the join.
DISCUSSION
The significantly wider internal angle found for the normally formed branch junctions collected for this study when compared with the naturally braced specimens is a predictable outcome based on the observations of Slater (2018a). The relationship between the angle of branch attachment and the increased likelihood of natural bracing occurring when branch junctions form tighter internal angles helps to explain this finding.
An analysis of all three junction groups subjected to rupture testing identified that the diameter ratio of the two arising branches had a significant effect on branch junction strength; regression analysis established that the diameter ratio explained 21.9% of the variation in the peak relative load (n = 73). Branch junctions, which had a higher diameter ratio (close to 1:1 ratio for the two branch diameters A1 and B1), broke at a lower relative load than those with a lower diameter ratio. This accords with the findings of other authors (Gilman 2003; Kane 2007; Kane et al. 2008).
Statistical analysis of the mechanical strength of the branch junction groups established that there were significant differences in their load-bearing capacity. Control branch junctions were found to fail at peak relative loads on average 24% higher than for BI junctions with the natural brace removed, comparable to previous studies (Smiley 2003; Slater and Ennos 2015b). BI junctions where the natural brace was retained exhibited an intermediate level of peak load, answering our first research question. These results identify that if an arborist chose to retain a natural brace within a tree, it is likely to help prevent the failure of the BI junction set below it. It then follows that the death and subsequent loss of a natural brace, or its removal by an unwitting arborist, are likely to increase the likelihood of failure of the BI junction set below that natural brace.
The typical failure mode observed in control branch junctions and those BI junctions where the natural brace was removed prior to testing provided similar results to the observations made by Slater and Ennos (2015b), who tested common hazel (Corylus avellana L.). However, there was substantial variability in the failure kinematics of BI junctions when the natural brace was retained, which related to the type of natural brace present on each specimen. Of the five branch junctions tested with entwining limbs as the natural brace retained, four failed at points along the smaller-diameter branch (Type 3 failure) rather than at the BI junction. This result supports the “very good” sustainability rating ascribed by Slater (2018b) to this form of natural brace and addresses our second research question: the type of natural brace can affect the failure mode of its associated BI junctions. The single specimen with an entwined natural brace that failed at the BI junction (Figure 9B) consisted of one smaller-diameter branch entwined with a substantially larger branch, suggesting that the diameter ratio of both entwined branches is an important factor in this natural brace’s resistance to bending forces—something that an arborist should take into account (Slater 2016).
Specimens with crossing branches left in place also displayed a complex pattern in their kinematics of failure with their bending load/displacement graphs exhibiting a high degree of variability, which related to the position and morphology of the associated natural brace. Failure at the BI junction occurred only after the crossing branches were displaced by the pulling action of the testing machine (made easier because these branches had been shortened due to the testing constraints of this study)—so the peak relative loads for these specimens were typical of a BI junction without a natural brace present. However, the resistance to branch displacement supplied by this form of natural brace is likely to act differently under wind loading than in these static tests. Displacement caused by strong winds does not act at a constant rate in line with the bifurcation, as it did in our rupture tests. Under dynamic wind loading, the friction between crossing branches may more often provide enough resistance to gusts of wind and prevent failure at the BI junction set below them. When the crossing branches were of large diameter relative to the size of the BI junction and in a firm pressing relationship, there was substantial resistance to the pulling apart of the arising branches. Those braces, which conferred considerable resistance to the splitting of the BI junction, generated a similar force/displacement graph to BI junctions with entwining limbs, whereby a jagged plotline was produced as the friction between the two crossing branches varied with displacement. The variability of these “crossing branch” specimens substantiates the range of sustainability ratings of “average” to “very poor” ascribed by Slater (2018b) to advise arborists assessing this type of natural brace in a standing tree.
The single BI junction where two adjacent limbs were fused was unique within this study: we observed that the fusion of branches was a rare occurrence in these specimens of C. monogyna, as no further examples of fused limbs could be found across a population of over 150 hawthorn trees. This specimen failed at one of the drill holes after an estimated peak relative load nearly three times as high as that of a normal branch junction, illustrating the potential strength of a fused connection between adjacent limbs. This finding corroborates the evaluation made by Slater (2018b), who proposed that the typical likelihood of failure of a natural brace consisting of fused limbs was “very low.” However, a further study is needed to put a statistically meaningful figure for the average bending strength of fused natural braces in C. monogyna or in other tree species.
Comparisons of excised wood samples confirmed that the axillary wood was significantly denser than xylem excised from the parent stem, answering our third research question. The heightened wood density found within branch axils in this study supports similar findings in previous studies using common hazel (Slater and Ennos 2013, 2015a). It is likely that the higher density of this axillary wood will impart a higher mechanical strength, whereby the dense, tortuous arrangement of the wood grain is less likely to yield from radial or tangential loading (Dresch and Dinwoodie 1996; Anten and Schieving 2010; Slater and Ennos 2013; Slater and Ennos 2015a; Ozden et al. 2017).
The observation of a period of normal development in the naturally braced BI junctions identifies that the inclusion of bark within these junctions is typically not initiated at the junction’s formation. A plausible explanation for this observation is that these branch junctions experienced normal loading at the early stages of their development but started to include bark within their structure after they became naturally braced.
Overall, the outcomes of this experiment strongly support the recommendation by Slater (2016) that arborists should take into account the presence, the type, the relative size, and the configuration of any natural bracing that is found set above a BI junction in an amenity tree. For an arborist to regularly prune out natural braces by working under an absolute rule of “remove rubbing branches,” it is likely to induce many unnecessary BI junction failures in the trees that such an arborist works upon.
RECOMMENDATIONS FOR FURTHER RESEARCH
No previous research has been conducted comparing the load-bearing capacity of BI junctions with their natural braces retained and removed, so this study provides initial data to which further studies could usefully add. This study has verified to some extent the ratings of three types of natural brace described by Slater (2018b). A wider range of natural braces across a wider range of tree species could be assessed through further static pulling tests and by assessing their “survivorship” within the crowns of standing trees under storm conditions. Testing the impact of different configurations of natural bracing selected by their type, relative diameter, and angle of interception during the elastic phase of bending could inform practitioners of any significant differences in the biomechanical performance of natural braces that may further inform tree assessment.
ACKNOWLEDGMENTS
Our thanks go to Dr. John Watt for allowing the use of his woodland for the supply of branch junctions from his hawthorn trees. We also acknowledge the kind help of Paul Critchley at UCLAN for providing training and support in the use of the Instron™ testing machine.
Footnotes
Conflicts of Interest:
The authors reported no conflicts of interest.
- © 2020, International Society of Arboriculture. All rights reserved.
Jump to section
Related Articles
Cited By...
- No citing articles found.