Abstract
Background Urban trees are commonly impacted by construction activities. Often, damage occurs below ground, as trenching, grading, and excavation activities disrupt tree root systems.
Methods In this study, 30 Acer rubrum L. were intentionally damaged by simulated trenching treatments. Treatments were randomly assigned at 1× (n = 10), 3× (n = 10), or 5× (n = 10) stem diameter away from the base of the tree. Root systems and severed roots were excavated, cleaned, and digitized using photogrammetry to assess losses in root volume. This measure, as well as changes in bending stress, were compared among the 3 treatments.
Results The 1× and 3× treatments exhibited the highest bending stress loss (%), correlating with the highest observed root loss. A positive relationship was noted between root volume loss and bending stress reduction (r2 = 0.5466).
Conclusions These results support previous studies that utilized similar trenching treatments on different tree species.
Introduction
Urban trees are generally long-lived organisms— persisting for decades despite growing in highly altered and constantly changing environments (Petri et al. 2016; Hilbert et al. 2019; Hauer et al. 2020). Given their longevity, mature urban trees located near built infrastructure such as roadways, sidewalks, or underground utilities will likely be impacted by construction or repair activities at least once during their lifetime (Hauer et al. 2020). While best management practices exist to help arborists and urban foresters make informed decisions regarding the impact of such conflicts (Matheny et al. 2023), the extent to which construction activities reduce tree health and structural integrity is not fully understood (Watson et al. 2014).
Much of the damage associated with development and redevelopment occurs belowground. Roots can be severed through trenching and grading activities, smothered by soil compaction or the addition of fill soil, or otherwise damaged as the surrounding site is altered to accommodate building activities (Matheny et al. 2023). In one of the earliest studies on the effects of construction damage on tree health, Miller and Neely (1993) observed the response of 98 trees damaged by utility trenching on a university campus. The authors noted that the negative impacts of trenching on tree growth were diminished as the distance between the root disruption and the trunk of the tree increased (Miller and Neely 1993). Moreover, the authors believed their data supported guidelines proposed by the American Society of Consulting Arborists which suggested adopting a minimum buffer of 0.3 m (1 ft) per 2.5 cm (1 in) in order to limit damage caused by utility trenching (American Society of Consulting Arborists 1989).
More recent research has offered further support for current industry guidelines (Costello et al. 2023; Matheny et al. 2023). Benson et al. (2019b) simulated linear trenching damage near mature Quercus virginiana Mill. and found that long-term water stress could be avoided if trenching was conducted at a distance ≥ 6× the stem diameter. Benson et al. (2019a) simulated grading damage around trees by trenching full circles around mature Q. virginiana at various distances from their base. In this work, extending the distance between the trunk and the trenching to 12× avoided any noticeable differences in stem diameter growth or twig elongation, though some short-term physiological stress was detected (Benson et al. 2019a).
Beyond these controlled trials, a nearly 40-year observational assessment of tree health in Milwaukee (Wisconsin, USA) has quantified the benefit of belowground space in limiting root system disruption associated with sidewalk and curb replacement (Koeser et al. 2013; Hauer et al. 2020). As tree terrace width was increased from 60 cm to 120 cm, a tree was 3 times more likely to still be present after the study’s initial 10-year assessment period (Koeser et al. 2013). As the city refined its preservation strategies, even trees located close to infrastructure replacement were spared from lasting damage thanks to alternative construction practices that limited root severance (Hauer et al. 2020).
In addition to the impacts of belowground construction damage on tree health, researchers have assessed the more immediate impacts related to whole tree stability (Smiley 2008; Ghani et al. 2009; Smiley et al. 2014; Fini et al. 2020). In one of the earlier works on this subject, Smiley (2008) used static pull testing to investigate the impacts of linear trenching on young Q. phellos L. (willow oak) stability. In this study, multiple iterations of trenching occurred at 5×, 4×, 3×, 2×, and 1× the trunk diameter away from the tree before a final trench was made at the tree base (measured at 1.4 m [4.5 ft]). The study found the force required to pull the trees to 1° was reduced when trenching occurred within 2 diameters of the tree’s trunk. In this same study, the author removed individual roots contiguously from 50% of the root flare circumference for another set of Q. phellos. A significant linear relationship existed between the percentage of the roots cut and the reduction in force required to pull the tree to 1° (r2= 0.80). Smiley et al. (2014) replicated this latter treatment regime with Acer rubrum L. and found a similar relationship (r2 = 0.74).
More recently, Fini et al. (2020) looked at the impacts of trenching on the uprooting resistance of Aesculus hippocastanum L. and Tilia × europaea L. In this study, the authors used static pull tests to compare the stability of trees that were severely damaged (trenched 40 cm/~4.5× DBH from the trunk on 2 sides) and moderately damaged (trenched 40 cm from the trunk on one side) to undisturbed controls. As expected, the trenching treatments reduced the bending moment required to pull the trees to 0.2°. More interestingly, the impacts of the moderate damage treatments were detectable in the T. × europaea more than 4 years after initial disruption (Fini et al. 2020). The severe damage treatment negatively affected uprooting resistance in both species for the duration of the monitoring period (51 months).
This paper builds on past research that utilized static pull tests to assess the effect of trenching on loss in stability by utilizing structure from motion (SfM) photogrammetry to accurately model tree root systems and quantify the effect of root volume loss due to trenching on changes in tree stability. The aim is for these findings to add to the existing body of research on belowground disruption and tree stability and are intended to bolster current industry best management practices focused on managing trees on construction sites (Matheny et al. 2023).
Materials and Methods
This trial was conducted at research plots at the University of Florida Environmental Horticulture Teaching Lab in Gainesville, Florida, USA (29.62423, -82.35425) in January of 2018. A stand of 30 established A. rubrum with stem DBH (measured at 1.4 m [4.5 ft]) measurements ranging from 9.9 cm (3.9 in) to 17.0 cm (6.7 in) served as the test subjects. The trees were planted at 3.7-m (12-ft) spacing intervals.
At the beginning of the trial, each tree was pulled until the base of the trunk tilted to a 1° angle toward the direction of the winch. Pull tests were conducted 3 times per tree to allow for the averaging of the calculated stress measurements. Once baseline stability was established, trees were randomly assigned 1 of 3 linear trenching treatments: trenching 1× DBH away from the base of the trunk; trenching 3× DBH away from the base of the trunk; and trenching 5× DBH away from the base of the trunk. Trenches were made with a pneumatic air excavator (AirSpade; Guardair Corporation, Chicopee, MA, USA). Trenches were excavated as a straight line on the side of the tree opposite the winch. Trench lengths were approximately 1.8 m (6 ft) long and approximately 46 cm (18 in) deep. Exposed roots were manually severed using a pruning hand saw.
Once the trenching treatments were applied, the trees were pulled again as described above. For all static pull tests, tree tilt, rope angle, and force were measured simultaneously with an integrated, purposebuilt measurement system (PiCUS TreeQinetic; IML Electronic, Rostock, Germany). The system’s inclinometer was secured to each tree’s base opposite the winch at a height of 15 cm (6 in). A 2,268 kg (5,000 lb) capacity electric winch was secured to a tractor and used to generate the pulling force required for the tests. The load cell and winch cable were attached to each tree at a height of 1.4 m (4.5 ft) using a sling. After each pull, the tension on the winc cable was released and trees were allowed to settle back to their initial inclination. Tree tability was compared in terms of bending stress (σ) which was calculated using the equation:
where P = the force (averaged over 3 pulls) required to reach 1° of inclination; θ = the angle of the winch cable from the horizontal; l = the trunk length from the ground to the height of the sling attachment; and D = the trunk diameter measured 15 cm above ground level (height of inclinometer attachment, which was directly above the root flare).
Upon completion of the static pull tests, tree root systems were dug using a 244-cm (96-in) tree spade. Soil was then removed from the root balls using a pneumatic air excavator. As the soil was being removed, care was taken to make sure that loose roots that were severed during trenching were numbered along with the corresponding root within the intact portion of the root ball so that they could be reattached. Once all the soil was removed, small diameter roots (< 1-cm diameter) were removed with a hand pruner. All of the severed root segments that were cut during trenching were then reattached to the root system from where they were originally severed using a thick-gauge stapler and tape. Thus, we were able to reconstruct the root system in full as it was prior to trenching.
Sample Preparation
Once the root systems were reassembled, digital 3D models were created using photogrammetry software following the methods described in Miesbauer and Koeser (2020). The root systems were placed upside down in a bucket and stabilized with sand. Coded targets were attached around the root system at measured intervals to provide scale for the software to generate the final models. Photographs of each root system were then taken at 3 different perspectives (120 photographs per perspective, 360 photographs per root system): above the root system angled downward; level with the root system; and below the root system angled upward, all using a digital single-lens reflex camera (Pentax K-1; Ricoh Imaging, Tokyo, Japan) mounted on a tripod and trolley (see Figure 1). All photos were taken indoors in a closed warehouse against a black backdrop.
Panel A shows an excavated and reassembled root system in the warehouse photography studio. Black and white targets were used to assist in the photogrammetry stitching process and to establish scale. A camera on a rolling tripod mount was moved to each tape line to capture imagery from above, below, and at root level. Panel B shows a 3D rendering of a similar root system. This scale model was used to measure root volume.
After the photographing of the root systems was complete, the digital images were downloaded to a computer for 3D model construction using a photogrammetry software program (PhotoScan; Agisoft LLC, St. Petersburg, Russia). The first step in the process was to import the photosets into the photogrammetry software to establish camera positions and create a sparse point cloud. Subsequently, a dense point cloud was generated from the same set of photos using the sparse cloud and the specified camera positions. Using the dense point cloud as a framework, a network of 3D polygons was superimposed onto the digitized root surface, followed by the creation of a watertight model (ensuring all holes in the mesh are sealed). The photos were then overlayed onto the textured, watertight model to achieve a realistic appearance for the final 3D model. The scale determined from the coded targets was utilized by the software to allow for estimate measurements of root volumes and surface area in the models.
Utilizing the software, we were then able to isolate the portion of the root system that was severed during trenching and calculate the volume of each root segment. The volumes of severed root segments from each root system were added together to get the total volume of severed roots. We then calculated the percent root volume loss due to trenching using the following formula:
where Total root volume of root system = Volume of severed roots + Volume of intact root system.
Design and Analysis
Relative differences in root volume loss (%) and percent bending stress loss among treatments were compared through an Analysis of Variance using the aov() function in R (R Core Team, Vienna, Austria). Mean comparisons were performed using the TukeyHSD() function. Means and standard errors for box plots and error bars were calculated with the DescribeBy() function from the psych package in R (Revelle 2023). Regression analysis comparing bending stress loss to root volume loss and total root volume to DBH was conducted using the statistical analysis and plotting application DataGraph (Visual Data Tools, Inc., Chapel Hill, NC, USA). We assessed all underlying assumptions for the statistical models through residual plots. All decisions were made at a significance level of α = 0.05. Figures were generated using DataGraph.
Results
When assessing the loss of bending stress after trenching, we observed a significant treatment effect (P-value = 0.007). Trees trenched at a distance of 1× DBH experienced a larger loss in bending stress (45.2%) than trees that were trenched 5× DBH from the trunk (22.3%). Trees trenched 3× DBH (35.2%) were not different than the other treatments (Figure 2).
Bending stress loss (%) across trenching treatments. Thirty Acer rubrum L. trees were randomly assigned to one of three trenching treatments: trenching at a distance of 1× tree’s diameter at breast height (DBH), 3× DBH, and 5× DBH away from the base of the trunk. Trees were pulled to 1° in the direction opposite the trenching damage. Error bars depict 95% confidence intervals.
Utilizing photogrammetry software, we were then able to isolate the portion of the root system that was severed during trenching and calculate the volume of each root segment that was detached from the root system during trenching. When looking at the percentage of root volume loss, we found a significant treatment effect (P-value < 0.0001). Trees that were trenched at 1× DBH had an average root volume loss of 13.3%, which was greater than for trees that were trenched at 3× and 5× DBH (4.1% and 2.5%, respectively)(Figure 3).
Root volume loss (%) across trenching treatments. Thirty Acer rubrum L. trees were randomly assigned to one of three trenching treatments: trenching at a distance of 1× tree’s diameter at breast height (DBH), 3× DBH, and 5× DBH away from the base of the trunk. The trees were excavated using a 2.4-m (96-inch) diameter tree spade, cleaned with an air excavator, and then the volumes of both the main and severed roots were calculated (in cm3) using photogrammetry. Error bars depict 95% confidence intervals.
When evaluating the relationship between the loss of bending stress and the loss of root volume, we observed a relatively strong positive correlation between these 2 variables. The coefficient of determination for these measures was 0.5466 (P-value < 0.0001). As depicted in the model fit presented in Figure 4, our data suggests that with each percentage of root loss, there is a corresponding 2% reduction in bending stress.
Relationship between bending stress loss (%) and root volume loss (%). Y = 19.68 + 2.072X. Model r 2 = 0.5466. Dashed lines depict 95% confidence intervals.
An even more pronounced relationship was evident between total root system volume and trunk diameter.
The coefficient of variation for these 2 measures was 0.7524 (P-value < 0.0001). Once again, this relationship was positive, with root volume increasing as tree DBH increased (Figure 5).
Relationship between root volume and diameter at breast height (DBH). Y = –0.0301 + 0.0059X. Model r 2 = 0.7524. Dashed lines depict 95% confidence intervals.
Discussion
This study adds to the body of research investigating the effects of trenching on tree stability. Smiley (2008) reported a significant decrease in force required to tilt trees to 1° when linear trenching was made closer than 3× DBH on young Q. phellos trees. Our findings are in partial agreement, in that trenching at 1× DBH reduced bending stress more than at 5× DBH, but not at 3× DBH. Moreover, when roots were cut at a distance equivalent to 1× DBH from the tree’s base, Smiley (2008) observed a 23% decrease in the force required to pull trees to 1°. In our study, at the same treatment level, we observed a 50% reduction in force (along with a 45% decrease in bending stress). Differences in methodology—Smiley (2008) successively trenched closer to the tree at 1× DBH increments, whereas we applied treatments of 1×, 3×, and 5× DBH randomly assigned to trees within the block—may partially account for seeming disparities. Another possible explanation for the observed differences in sensitivity to trenching regarding tree stability may be attributed to variations in tree species and soil conditions between the 2 studies. Tree root systems can be broadly categorized into 3 distinct forms: tap root, heart root, and flat root (Stokes 2002; Stubbs et al. 2019). Acer rubrum root systems are typically categorized as flat root systems, characterized by substantial lateral roots alongside some smaller vertical roots (Lyford and Wilson 1964). However, some researchers have observed that A. rubrum can exhibit more of a heart root system, featuring a combination of large vertical and horizontal roots (Smiley et al. 2014). In our study, when analyzing the root systems of the sampled trees, the majority aligned with the former description. In contrast, Smiley (2008), who expressed surprise at the lower force reduction numbers, posited that this discrepancy was likely due to the presence of deep, uncut roots associated with the heart root form of Q. phellos.
While the methods used provide greater insights into the volume of roots lost during trenching operations, there are some notable limitations to the method. The first is the use of the large diameter tree spade to conduct the initial harvest. While this allows for more efficient removal of soil from the dug root system, we lost any insights into what roots were present outside of this harvested area. The second is the loss of fine roots (which must be removed to facilitate the photogrammetry process).
Despite the limitations of our harvesting method, we observed a relatively strong correlation between DBH and root volume. DBH is one of the most frequently employed tree measurements, both in traditional forestry and urban forestry. Consequently, allometric research has frequently centered on determining the predictability of both aboveground and belowground tree components based on DBH (Shinozaki et al. 1964a; Shinozaki et al. 1964b). Wang (2006) observed that the belowground biomass of 10 common forest species in China displayed a strong linear relationship with logarithmically transformed diameter measurements, whereas Peichl and Arain (2007) noted an exponential correlation between root biomass and stem diameter in Pinus strobus L., encompassing trees aged from 2 to 65 years old. Our simple linear relationship may be a reflection that our root volume measurements were limited to only the roots captured in the tree spade, or it may simply reflect our relatively uniform sample group of even-aged A. rubrum.
With our methods, we were able to measure the amount of root volume within the excavated root ball that was lost due to trenching. Trenching at 1× DBH from the trunk resulted in a much larger percent of total root loss than the other treatments. Figure 4 shows a fairly strong correlation between the amount of root loss and reduction in bending stress. Trees that were trenched 5× DBH showed a reduction in bending stress of 22.3%, which is similar to the results reported by Fini et al. (2020), who trenched at a similar distance from the trunk (4.5× DBH) on A. hippocastanum and T. × europaea. For trees that experienced moderate damage (i.e., trenching on just one side), there was a 21% reduction in bending stress compared to the control group.
Conclusion
Our study aligns broadly with previous work demonstrating the adverse effect of root cutting on stability for different species. In particular, cutting roots at a distance from the trunk less than DBH reduces stability significantly more than cutting roots at a distance 5× DBH. But cutting at 3× DBH reduced stability as much as 1× DBH, suggesting that, for management purposes, cutting roots within 5× DBH on one side of the tree is more likely to result in instability.
Reconstructing root systems for photogrammetric analysis allowed us to quantify root loss more rigorously. Although the volume of roots lost to cutting was directly proportional to the reduction in bending stress needed to tilt trees 1°, the relationship was not exactly 1:1. The reduction in bending stress was more than 3 times greater than the volume of roots lost when they were cut at 1× DBH and more than 8 times greater when roots were cut at 3× and 5× DBH. Practitioners should understand that even a small volume of root loss can meaningfully reduce tree stability. Future work should conduct tests to failure to assess whether pulling tests in the elastic range accurately represent the loss in load-bearing capacity needed to cause uprooting.
Acknowledgements
This project received funding from the Florida Chapter of the International Society of Arboriculture Grant Program. We extend our gratitude to Chris Harchick, Drew C. McLean, John Roberts, Deb Hilbert, Marvin Lo, and Abby Tumino for their contributions in harvesting the roots, photographing them, and constructing the 3D models.
Conflicts of Interest
The authors reported no conflicts of interest.
- © 2025 International Society of Arboriculture
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.