Potential tree species for use in urban areas in temperate and oceanic climates

This study aims to assess the potential of trees for integration in urban development by evaluating the damage caused by trees in relation to various tree characteristics. Tree damage to permeable pavement systems and other urban structures such as impermeable pavements, kerbs, roads, retaining walls, footpaths, walls and buildings were assessed to identify the most suitable trees for the urban environment. One hundred square sites of 100 m × 100 m were randomly selected in Greater Manchester for this representative example case study to demonstrate the assessment methodology. Among tree species in this study, Acer platanoides L. (Norway maple) occurred most frequently (17%); others were Tilia spp. L. (Lime; 16%), Fraxinus excelsior L. (common ash; 12%), Acer pseudoplatanus L. (sycamore; 10%) and Prunus avium L. (wild cherry; 8%). The study concludes that 44% of the damage was to impermeable pavements and 22% to permeable pavements. Other damage to structures included kerbs (19%), retaining walls (5%), footpaths (4%), roads (3%) and walls (3%). Concerning the severity of damage, 66% were moderate, 21% light and 19% severe. Aesculus hippocastanum L. (horse chestnut) caused the greatest damage (59%) expressed in percentage as a ratio of the tree number related to damage over the corresponding tree number that was found close to structures.


Rationale
Trees play major roles in creating healthy urban ecosystems and sustainable environments. However, some trees may cause damage to urban structures such as permeable pavements, impermeable pavements, kerbs, roads, footpaths, buildings and retaining walls. For the purpose of this study, permeable pavements are defined as a sustainable system comprising a base and subbase allowing the movement of storm water through the surface, reducing runoff. Randrup et al. (2003) indicated that in some cities substantial amount of money has been allocated to address conflicts between the rooting system of trees and urban infrastructure. An assessment of this sort becomes important as part of a decision support tool for the fitting and retrofitting of sustainable drainage systems (SuDS), and in the planning of tree planting projects at urban development sites, regeneration projects, and sustainable drainage projects (Scholz and Uzomah, 2013). A SuDS is designed to lessen the potential impact of construction developments with concerning surface water drainage discharges (Scholz, 2010(Scholz, , 2015. Table 1 summarised tree development characteristics (Pliûra and Heuertz, 2003;Defra, 2007;Arbor Day Foundation, 2015;British Hardwood Tree Nursery, 2015;Garden Centre, 2015). The presence of urban trees often increases property prices (Sander et al., 2010;Scholz, 2010Scholz, , 2015. Trees may be regarded as desirable by residents, because they increase the aesthetic value of a place, and provide ecosystem services including amenities (Scholz and Uzomah, 2013).

Brief literature review
The root system of vegetation such as most trees provides the essential functions of anchorage (known as structural stability), absorption of runoff water and nutrients as well as storage of vital food reserves (Scholz, 2010(Scholz, , 2015. Tree roots may cause damage to underground utility services by direct pressure on conduits as roots grow and expand in diameter, or by entry to hydraulic services such as sewers and storm water pipes occasionally causing destruction and more frequently blockage (Mather and Morton, 2008). Tree species that have large and vigorous root systems in terms of their growth rate may result in significant destruction of public infrastructure elements including roads, kerbs, footpaths, paved areas and underground utility services. Trees with these characteristics should preferably be avoided or at least controlled. Mechanical forces exerted by radial growth of tree roots can lift relatively light structures such as paths, curbs, paving slabs and Table 1. Tree development characteristics (after Garden Centre (2015), Arbor Day Foundation (2015), British Hardwood Tree Nursery (2015), Pliûra and Heuertz (2003), and Defra (2007)). Tree  boundary walls as well as occasionally single story buildings including porches and garages (Biddle, 1998;Mather and Morton, 2008). However, poor construction of pavements can also cause structural failure (Sydnor et al., 2000).
Some research studies have found strong correlations between the size of specific trees and serious conflicts with infrastructure demands (Mather and Morton, 2008).
Large trees usually cause more conflicts than small trees. Damage to pavements correlate with nearby tree diameters. Most trees are linked to damage when they are between 11 and 20 cm in diameter (measured at breast height). However, most Quercus spp. (Oak) and A. hippocastanum do not cause harm, unless they are greater than 20 cm in tree diameter at breast height (DBH) according to Randrup et al. (2003). Randrup et al. (2003) pointed out that a concrete or asphalt pathway can act as a barrier preventing soil moisture loss by evaporation. This artificial evaporation barrier creates a more humid environment on the underside of the pavement surface, because of temperature differences between the soil and the above pavement. Tree roots are therefore naturally attracted to the condensation water at the soil and impermeable pavement interface (Randrup et al., 2003). This may eventually lead to pavement surface destruction through the radial forces generated during tree root growth.
When tree roots encounter dense soil layers, they usually change direction, stop growing, or adapt by remaining unusually close to the surface. This superficial rooting makes urban trees more vulnerable to drought and can cause destructive pavement heaving (Randrup et al., 2003). The highly compacted soils commonly required for constructing pavements do not allow tree root penetration (Scholz, 2013). Viswanathan et al. (2011) undertook a research study concerned with the performance of Liquidambar styraciflua L. (American sweetgum) roots under permeable and impermeable pavements. Their results suggested that the standing live root lengths for the American sweetgum were longer in impermeable concrete than in permeable concrete for the first 0 to 20 cm of soil depth. Beyond this depth, the standing live roots were more abundant in permeable than in impermeable pavements. However, they came to the conclusion that pervious concrete does not give a quantifiable root production benefit in comparison to impervious concrete. Giuliani et al. (2015) used modelling tools to analyse tree growth in street pavements. The findings indicate the progressive reduction of deformations with the increase of the depth of root penetration. However, these studies require a lot of data for individual trees and sites.

Aim, objectives and significance
This study aims to provide a simple method to assess the damage caused by urban tree roots in relation to the corresponding tree characteristics such as species, distance from structures, DBH, tree height, crown spread (diameter), and tree hangover characteristics. The outcomes should be used to focus on planting the most suitable tree species near specific urban structures in the future.
The objectives are (1) to outline a method for random selection of representative sites in Greater Manchester to study the tree damage characteristics; (2) to identify the predominant trees causing damage to urban structures in Greater Manchester (example case study); and (3) to define a rapid methodology to assess the damage to structures such as permeable pavements, impermeable pavements, roads, kerbs, footpaths, and retention walls for which individual tree species are responsible for. This study provides valuable information for the retrofitting of structures such as permeable pavements in combination with existing trees and to developers in choosing the most suitable trees for the right urban environment minimising damage.

Site selections
In order to address objective (1), a total of 100 sites were randomly selected in the Greater Manchester area (North-west England) using the Google Earth map and tools, but restricted within the area of an ellipse covering the main urban areas around Manchester city centre for ease of assessment and to reduce transportation costs ( Fig. 1). A square of 100 m × 100 m was drawn around the centre of each selected site to identify an outer boundary for the tree assessment studies. The coordinates, grid references, longitudes, latitudes and post codes of all sites were [ ( F i g . _ 1 ) T D $ F I G ] Fig. 1. Overview of the assessed case study sites located in the Greater Manchester area.
Article No~e00154 subsequently determined. The Greater Manchester area is located between 53°28′0′′N and 2°14′0′′W. The estimated population of Greater Manchester is around 2,680,000.

Tree damage data collection
Data related to tree damage assessments were collected to address objective (2).
The data set included variables such as site number, tree number, tree species, common name and genus, tree DBH (1.5 m from ground level), estimated tree height, estimated tree crown diameter, structures near the tree, distance (no maximum threshold) of all aboveground nearby structures from the tree (defined at species level where possible), type of damage (if any) to structures and their severity as well as subjective aesthetic considerations. Site visits were carried out during 2013 and 2014. Predominantly summer periods were chosen because during these periods, trees have their full leaf canopies, which makes tree identifications and corresponding crown spread determinations easier.
All trees within the marked 100 m × 100 m boundaries with a DBH of greater than 10 cm were assessed, except where a site was inaccessible for a valid reason; e.g., some relatively small areas within restricted (private) access areas such as gated private gardens were not assessed, and were subsequently marked as inaccessible sites. Other sites without any tree data entries were without any trees, had only trees where the DBH was less than 10 or were predominantly of a different land use category (without trees) such as water.
The DBH was calculated by measuring the circumference at breast height using a tape measure, and dividing the value by π (approximately 3.14159). Trees less than 10 cm in diameter were not recorded as they were considered too young to cause any measureable damage. Tree heights were estimated using methods based on goniometry (Vernier, 2013), and also by comparing the tree height with nearby structures such as houses as well as electric and telephone poles. Goniometry involves walking away from the base of the trunk until the observer sees the top of the tree from an angle of 45°(which the observer can check using his or her arm). The height of the tree roughly equates to the distance from the tree to where the observer is standing plus his or her eye height from the ground.

Tree damage assessment method
In order to address objective (3), the information and data collected and processed to show the methodology to assess damage by trees is summarised in Table 2,  Severe   Damage to retaining walls Fagus sylvatica L. 2/3 2/4 2/9 0/3 0/1 1/3 0/3 Table 5. Presentation of relative tree rankings concerning the structural damage. The rankings indicate increases of 'potential for damage', where 1 represents the least potential for damage and where 10 represents the highest potential for damage. Weighting factors reflect the relative importance of structures based on civil engineering expert judgement by the authors.  Table 6. Predicted future damage potentials for tree species based on their growth and development characteristics. Relative rankings (RR) of the 'potential for future damage', where 1 represents least potential for damage, and 10 represents highest potential for damage. Weighting factors reflect the relative importance of structures based on civil engineering expert judgement by the authors.

Article No~e00154
The footpath structure refers to a walkway though areas such as parks. The damage that was taken into account is lifting-up of structures, disjointing of structures by roots, sinking-in (depression) of structures and cracking-up of structures. The assessment was undertaken with care to distinguish between damage due to trees and/or poor construction (Sydnor et al., 2000). However, all assessments were based on civil engineering expert opinion considering that no disruptive and/or destructive tests could be undertaken on private and public land. Pictures of actual root damage were taken and analysed.
The severity of damage was determined by assigning numbers between 1 and 3, where 1 represents an emerging damage at an early stage ('light damage'), 2 indicates a damage that is gradually advancing or already well-established ('moderate damage'), and 3 equates to 'severe damage', which is an advanced damage (e.g., pavements completely separated or kerbs completely disjointed) or a well-advanced damage that has become a safety hazard to users requiring immediate attention (or a damage that has already been repaired). It follows that essentially a rather coarse three-category damage scale (Table 3) has been used to reflect the fact that damage to structures by trees is rare despite the large data set collected. Furthermore, the absolute majority of trees did not cause any damage, and could be seen as the control group, which was assigned 0 (no damage recorded). Table 4 and Table 5 indicate the proportion of tree species that caused structural damage and the relative tree rankings concerning the structural damage, respectively (see Section 3.4 for detailed descriptions and interpretations). Weighting factors reflect the relative importance of structures to the local infrastructure (key criterion of assessment), and have been determined by civil engineering expert judgements expressed by the authors (Table 5). For example, damage to a building receives a higher weighting than damage to pavements.
However, pavements are more important than their corresponding kerbs. Finally, Table 6, which has also been informed by the literature review and statistical analyses of the data set was limited due to the small sample size for most species causing damage and the dynamic nature of the urban environment.

General overview
A total of 536 mature trees were assessed in detail. After applying the criteria given above, the tree species percentage occurrence reduced accordingly. Table 2 shows a summary of the most frequently occurring trees and their corresponding damage recordings to key urban structures. The range of tree size has been limited by including only trees that have DBH entries of at least 10 cm to avoid skewing the data set towards small trees that might not survive. Furthermore, small and young trees have not yet developed sufficient size and strength to cause damage to the surrounding infrastructure. The fact that the standard deviations of DBH are high reflects the point that even mature trees are highly variable in size, which is natural.
The application of the proposed methodology has been demonstrated in Table 2,

Structural damage
The proportions of structures that were linked to damage from trees can be found in Table 2. Of the total 231 damaged structures observed, the following proportions expressed in percentages can be calculated: impermeable pavements (44%), permeable pavements (22%), kerbs (19%), retaining walls (5%), footpaths (4%), roads (3%) and walls (3%). No damage to buildings (0%) has been recorded. The patterns associated with damage linked to impermeable pavements compared to permeable pavements are in line with the findings by Randrup et al. (2003). This suggests the need for more retrofitting of robust SuDS techniques (Scholz and Uzomah, 2013). However, it is expected that the severity of damage will advance further with time.
No damage to buildings was recorded. This is possibly due to the fact that the assessment was only based on an external visual observation. An internal structural assessment may reveal damage to buildings. Moreover, most buildings have formidable foundations and may not be easily damaged as compared to road structures and pathways.
By dividing the number of a particular species causing damage by the number of the corresponding species occurrence, the proportion of damage caused by this species can be calculated from In order to achieve maximum ecosystem service benefits, the most suitable trees that could be combined with SuDS are the one that (a) are as close to structure as possible; (b) have a large diameter; (c) cause the least or no damage; and (d) are readily available and desirable by residents. The closer trees are to the structures or residents, the more the effects are felt; e.g., reducing localised extreme temperatures. The greater a tree diameter, the more mature the tree is likely to be and, therefore, the more noticeable will be the tree benefits (Leuzinger et al., 2010). Trees linked to a low damage potential are usually preferred both for new construction or retrofitting of SuDS sites. Vegetation that is desirable by residents is usually associated with high aesthetic values such as mature and beautiful trees with a perceived rich character.
3.3. Damage to structures linked to tree diameter and distance 3.3.1. Overview Fig. 2, Fig. 3, Fig. 4, Fig. 5, Fig. 6, Fig. 7, Fig. 8  For x(y/z), where x represents the DBH (cm), which is also signified by the relative size (diameter) of the circle. The diameters expressed by circles give a visual indication of the maturity of the average tree species, which makes visual comparisons between trees easier. The entry z indicates the number of the tree species within 10 m of the structure, out of which y trees caused moderate to severe damage.

Permeable pavement
For permeable pavements, most significant (p < 0.05) damage to permeable pavements was caused by trees located within 0-1.0 m away from a structure, except for those from F. excelsior. About 33% of F. excelsior located close to permeable pavements caused damage to these pavements if their average diameter was 66 cm and if their average distance was 2.3 m away from the permeable pavements (Fig. 2). The trees with the highest percentage of moderate and severe damage to permeable pavements (up to 50%) were F. sylvatica, A. pseudoplatanus and B. pendula. However, the corresponding sample sizes were rather small. The average distance of F. sylvatica and B. pendula to permeable pavements was 0 m, indicating that most of these two species were planted too close to the pavement.
The average DBH of these trees was 68 cm and 20 cm, respectively (Fig. 3).
[ ( F i g . _ 2 ) T D $ F I G ]  Fig. 3. Relationships of tree diameters at breast height (DBH (cm); represented by circles), average distances of trees away from impermeable pavements, and the proportion of trees within 10 m to this structure subjected to moderate to severe damage. Note: x(y/z), where x represents DBH and z indicates the number of the tree species out of which y trees caused damage.

Article No~e00154
Acer platanoides caused the most overall damage (2 light, 4 moderate and 4 severe damage) to permeable pavements (Table 3), which was statistically significant (p < 0.05). The average diameter of the tree was 56 cm and the mean distance from the permeable pavements was 0.8 m. However, Fig. 3 shows a comparison for only moderate and severe damage. Seven out of sixteen T. cordata trees located close to permeable pavements caused major damage. The corresponding average tree diameter was 26 cm and the mean distance from the structures was 0.8 m. Acer pseudoplatanus caused six major damage incidents to permeable pavements. The average DBH of this tree was 52 cm and it was located about 0.4 m away from structures. Tilia platyphyllos caused five major damage occurrences to permeable

Impermeable pavement
Concerning impermeable pavements (Table 2), the majority of the damage occurred to these pavement structures (44%), which was statistically significant (p < 0.05). The reason for this is that impermeable pavements do not allow free circulation of moisture and air into and out of the pavement surface (Randrup et al., 2003;Day et al., 2010;Scholz, 2013). Because of this, pockets of moisture buildup below the surface of impermeable surfaces, causing the roots of trees below the impermeable surface to be attracted to these pockets of moisture, and thereby lifting-up of the corresponding pavement surface. This may have accounted for the relatively high number of damage to impermeable pavements.
For an impermeable pavement, the further away the tree (up to a distance of 1.4 m), the higher is the percentage of this tree causing damage irrespective of the tree DBH (Fig. 3). This is rather unexpected, considering that a large tree DBH is usually linked to large roots, which would cause damage at close range. However, the tree DBH variability is relatively small and the sample sizes are rather small as well. Regression analysis did not reveal any significant findings.
Wherever tree roots are deprived of air and moisture, they start to grow back towards the surface to obtain these resources. Morgenroth (2011) studied root distribution in relation to paved and normal surfaces in the top 30 cm of soil. He found that root abundance in the top 30 cm is greater in impermeable pavements than in normal soil.

Article No~e00154
This study revealed that the pavements of Greater Manchester roads consist of more impermeable pavements than permeable pavements. Considering the findings of Morgenroth (2011) and Viswanathan et al. (2011), the Greater Manchester case is more likely linked to the phenomenon of insufficient moisture in the compacted soil strata below the impermeable pavements, and the tendency of roots to remain close to the surface for oxygen and moisture availability. Hence, this is the reason for greater damage to impermeable pavements than permeable pavements. This phenomenon seems common where there are more impermeable pavements than porous surfaces. Acer pseudoplatanus caused the most damage to impermeable pavements (78%) from an average distance of 1.3 m and an average DBH of 64 cm.

Kerb
Kerb damage comprised 19% of all recorded structural damage. Acer platanoides caused the most damage to kerbs (10 out of 16 trees were located close to kerbs) from an average distance of 0.6 m and with a mean DBH of 41 cm (Fig. 4). Similar to A. platanoides was the impact of T. platyphyllos (7 out of 13 nearby T. platyphyllos), A. pseudoplatanus (2 out of 11 surrounding trees) and F. excelsior (7 out of 14 surrounding trees) caused damage to kerbs from the furthest average distance of 1 m (Fig. 4).
Other trees that caused damage were less than 1 m from the kerb as shown in Fig. 4. Prunus avium was the best tree suitable for kerbs: only 1 in 14 trees caused moderate to severe damage (Fig. 4). However, most P. avium were very closely located (0 m) to kerbs, and their average DBH was 62 cm. This was closely followed by F. sylvatica. Although for F. sylvatica of an average DBH of 93 cm (indicating trees well-advanced in age) and an average distance of 0.23 m from kerbs, only 2 out of 9 trees caused moderate to severe damage to kerbs (Fig. 4).
The worst tree to be located close to kerbs is A. platanoides. For trees of this species with an average DBH of 41 cm (indicating middle age) and located about 0.6 m from the kerbs, about 10 out of 16 A. platanoides caused moderate to severe damage to kerbs (Fig. 4).

Other structures
The percentages of damage to roads and retaining walls were 3% each. Only three trees (F. excelsior, A. platanoides and A. pseudoplatanus) caused moderate to severe damage to roads (Fig. 5). Trees that caused damage to roads were located within an average distance of 2-5 m away from roads, indicating that the majority of them were planted close to the pavements.
Not many trees were found close to retaining walls. For B. pendula with a DBH of 125 cm and planted at an average distance to structures of close to 0 m, only 1 out of 2 trees caused moderate to severe damage (Fig. 6). Aesculus hippocastanum Article No~e00154 caused the most damage to retaining walls. Three out of four A. hippocastanum with an average DBH of 61 cm and located at a mean distance of 0.25 m away caused moderate to severe damage to retaining walls (Fig. 6).

Trees
Acer platanoides occurred the most frequently (17%) among other trees that were found in this survey (Fig. 1). Furthermore, A. platanoides caused the most severe damage to structures (Table 2 and Table 3). The damage done to structures by A.
platanoides did not follow any particular pattern. In this survey, 38 out of 73 (52%) A. platanoides caused damage to various structures (Table 2). About 35% of all A.
platanoides planted close to permeable pavements with an average DBH of 56 cm and an average distance of 0.75 m from the permeable pavements caused severe to moderate damage to the pavement structures. This average DBH represents maturing A. platanoides.
On average, A. platanoides caused more damage (42%) to impermeable pavements than to permeable pavements. These 42% of A. platanoides had an average DBH of 42 cm with an average distance of 0.3 m from impermeable pavements. This DBH represents A. platanoides, which are still in their relatively fast growth phase. This indicates that A. platanoides has a greater potential to cause more damage to impermeable pavements than to permeable pavements.
About 60% of A. platanoides with an average DBH of 41 cm caused severe to moderate damage to kerbs from an average distance of 0.6 m. Acer platanoides of this DBH are still in the growing stage, indicating a future potential to cause more damage to kerbs. It follows that A. platanoides should not be recommended for planting near kerbs, as it is ranked the least suitable tree for planting close to kerbs (Table 4).
Only 7% of A. platanoides with an average DBH of 50 cm caused severe and moderate damage to roads. Their average distance from roads was 2.0 m. Roads are normally well-compacted during construction to bear heavy traffic and haulage loads, and will therefore resist most damage from tree roots. Moreover, trees are normally at least 2.0 m located from roads, because of spaces for permeable or impermeable pavements and kerbs. Therefore, roads were linked to rather few damage incidents by tree roots.
There were no records of severe and moderate damage to retaining walls by A.
platanoides. About 20% of A. platanoides planted close to walls of average DBH of 51 cm caused severe and moderate damage to these wall structures. Those that caused damage were placed at an average distance of 1.0 m from the walls.
About 10% of A. platanoides close to footpaths with an average DBH of 63 cm caused severe and moderate damage to footpaths. Those that caused damage were at an average distance of 3 m to the footpaths. Damage to footpaths by Article No~e00154 A. platanoides even at a distance of 3 m are possible, because the underlying soils at footpaths are not as compacted as those associated with other road structures. Despite that A. platanoides caused the most damage, and was also ranked the lowest in the potential for retrofitting (Table 5).
Concerning T. platyphyllos, most trees that caused damage to urban structures (for example, impermeable pavements, retaining walls and footpaths) were very closely located to these structures compared to other trees (Fig. 4, Fig. 6, Fig. 7 When assessing the damage to structures caused by T. platyphyllos with the relative importance of these structures, T. platyphyllos came second in terms of choice (Table 5). Furthermore, T. platyphyllos did not rank high in terms of future potential for damage ( excelsior tree of this DBH is considered to be fully grown (Dobrowolska et al., 2011). Fraxinus excelsior was the tree furthest away that caused damage to permeable pavements. This may be due to its great size.
About 70% of F. excelsior that were close to impermeable pavements caused severe to moderate damage to these pavement structures. The trees were of an average DBH of 30 cm and were located at a mean distance of 0.7 m from the impermeable pavements. Fraxinus excelsior trees of such DBH are considered to be young and developing, and are likely to cause more damage to any urban structures in the future.
About 50% of the F. excelsior trees that were located closely to kerbs (average distance of 0.9 m) caused severe to moderate damage. Their average DBH was 62 cm. Most of these trees could be considered as mature. About 25% of F. excelsior close to roads with an average DBH of 72 cm caused severe to moderate damage to these road structures. They were located at an average distance of 2.0 m to the roads. About 50% of the F. excelsior found close to retaining walls with an a mean DBH of 53 cm caused severe to moderate damage to the retaining walls.
They were placed at an average distance of 0.3 m from the retaining walls.
Fraxinus excelsior had the highest average distance from the retaining walls amongst other trees that caused damage to retaining walls. The percentage of F. excelsior that caused damage to footpaths was the least among damage to other structures. The percentage of the trees that caused damage to footpaths was about 18% with a mean DBH of 53 cm and located an average distance of 0.5 m from the footpaths.
Fraxinus excelsior ranked very high (8/10) in terms of potential for damage (Table 5), but ranked lower (5/10) in terms of potential for future damage. Most F. excelsior trees recorded in this survey were already mature, but reached less than half of their life span when compared with data shown in Table 1. Secondly, none of the F. excelsior trees were located very close to any structure. Fraxinus excelsior received average scores (51%) in terms of aesthetics in spring and summer, but very low scores (24%) for aesthetics in autumn.
Acer pseudoplatanus caused damage to structures, even if planted at distances that could be considered as far away from structures such as permeable pavements, impermeable pavements, kerbs and roads. However, there were no recorded damage by A. pseudoplatanus to footpaths and retaining walls. The average diameter of A. pseudoplatanus that caused damage to structures ranged from 52 cm for permeable pavements to 73 cm for both roads and walls. Findings indicated that 6 out of 12 A. pseudoplatanus (50%) with a mean DBH of 52 cm caused damage to permeable pavements at an average distance of 0.4 m (Fig. 2). Acer pseudoplatanus was the only tree that consistently caused damage from the furthest distance concerning kerbs, impermeable pavements and roads (Fig. 3,  Fig. 4, Fig. 5). Acer pseudoplatanus was responsible for the most damage to impermeable pavements from the furthest average distance of 1.2 m with a mean DBH of 64 cm (Fig. 3).

Article No~e00154
Because of the potential to cause damage even from a relatively far distance, A. pseudoplatanus ranked very high (9/10) in the potential for damage (Table 5), and also ranked very high (10/10) in the potential for future damage (Table 6).
Prunus avium caused moderate to severe damage only to kerbs and footpaths. The corresponding damage to kerbs was the lowest (1/14 trees) among other trees.
Prunus avium had an average DBH of 62 cm and were located very close (touching distance) to kerbs (Fig. 4). The number of P. avium that caused damage to footpaths was also very small (2/14). The DBH was 45 cm and the average distance from the footpaths was 2.5 m.
Concerning future damage, P. avium ranked third ( Aesculus hippocastanum caused moderate to severe damage to permeable pavements, impermeable pavements, kerbs and retaining walls, but none to roads, footpaths and walls. About 32% of A. hippocastanum were responsible for moderate to severe damage to these structures. Most A. hippocastanum that caused damage were mature in size with a mean DBH ranging from 51 to 71 cm (Fig. 2,   Fig. 3, Fig. 4, Fig. 5, Fig. 6, Fig. 7, Fig. 8).
Aesculus hippocastanum was ranked as the second (2/10) best tree with regard to damage to structures, and ranked fourth best in the potential for future damage, because most of the assessed trees were already mature. However, A.
hippocastanum leaves generally lead to considerable volumes of leaf litter on streets during autumn.
cordata to structures were generally within a mean distance of 0 m (as for impermeable pavements) to 0.7 m (as for permeable pavements). Most T. cordata that caused damage could be classed as still being very young, since their average DBH were between 26 to 38 cm, compared with those of 146-200 cm for a mature T. cordata tree (Fig. 2, Fig. 3, Fig. 8 and Table 1).
Tilia cordata was responsible for damage already at young age (indicated by a small DBH). Therefore, this tree was considered to have a high potential to cause damage both in the present but particularly in the future.

Article No~e00154
Betula pendula caused moderate to severe damage to permeable pavements, kerbs, walls and retaining walls, but no harm to impermeable pavements, roads and footpaths. The DBH for B. pendula that caused damage varied widely: 20 cm for those trees near permeable pavements, 45 cm for those near kerbs, 73 cm for those near walls, and 125 cm for those near the retaining walls. Most B. pendula that caused harm were very close to the structures they damaged, except for those close to walls, which were located at an average of 0.9 m away from trees. Due to B.
pendula being able to cause damage even at small DBH, it ranked very high in the potential for structural damage both at presence and in the future (Table 5 and  Table 6).
Crataegus monogyna caused moderate to severe destruction to only impermeable pavements and retaining walls at an average DBH of 25 cm and 20 cm, respectively (Fig. 2, Fig. 3, Fig. 4, Fig. 5, Fig. 6, Fig. 7, Fig. 8), indicating that these were still relatively small trees. However, 2 out of 4 C. monogyna caused moderate to severe harm to impermeable pavements from an average distance of 1 m, while 1 out of 3 trees caused damage to retaining walls from a mean distance of 0 m ( Fig. 3 and Fig. 6).
Crataegus monogyna ranked fifth in terms of potential for damage. This tree was located close to most structures, but damaged only two (Table 5). However, it ranked second in terms of potential for future damage ( Table 6). The overall size of this tree may not increase significantly in the future due to its natural size, which is rather small compared to other trees such as A. pseudoplatanus (Table 1).
Fagus sylvatica was linked to moderate and severe destruction to permeable pavements, impermeable pavements, kerbs and footpaths, but no damage to roads, walls and retaining walls. The average DBH of most F. sylvatica trees causing damage was relatively large, ranging from 68 to 93 cm (Fig. 2, Fig. 3, Fig. 4 and Fig. 8), indicating that they are already large and mature (Table 1). In all cases of harm to structures, F. sylvatica trees appeared to be the largest trees in terms of DBH wherever they featured (Fig. 2, Fig. 3, Fig. 4 and Fig. 8).

Study limitations
The study has limitations due to the complex nature of the dynamic urban environment. In order to make sure that tree species, which have a good spread in Greater Manchester and that are typical for urban areas are well-reflected in this analysis, and also that recorded damage were actually caused by trees and not by other causes such as soil settlements due to unforeseen heavy traffic, the following criteria were applied: (1) Tree species that had less than 10 occurrences in total were not included in the analysis; (2) Tree species that occurred in less than five different sites were also discarded.
(3) All damage classed as 'light' was also not included in the detailed analysis to reduce the likelihood of making trees responsible for damage when in fact other causes of damage are potentially also likely. Alternative reasons for damage might be natural settling of structures, fatigue of old constructions and physical damage linked to road accidents.
(4) For the analysis of structural damage, only the structure types with at least ten damage reports linked to a specific tree species were considered to decrease the likelihood of high variability linked to small data sets to lead to spurious findings.
Furthermore, not all trees were located in areas where they had the opportunity to influence all categorised structures in the same manner; e.g., some of the trees were located, for example, in parks with no major structures (e.g., buildings and roads) around, and that may have reduced the proportion of trees that have caused damage to these types of structures. Therefore, it was important to base the study on a relatively large data set collected at random and to implement the above criteria limiting the risk of spurious findings.

Conclusions
Considering the damage to structures by trees, obvious patterns may not have been demonstrated due the small sample size for a given species combined with the variability of the growing conditions for each species and site. Nevertheless, based on the Greater Manchester case study, the 'best trees' (in terms of relatively low risk to infrastructure) to be recommended for temperate and oceanic climates are T.
The project also concludes that impermeable pavements were subject to the highest number of damage from trees (44%), followed by permeable pavements and kerbs (22% and 19%, respectively). Trees planted close to impermeable pavements will cause more damage to the structure compared to those planted close to permeable pavements under the same conditions, which should be considered by town planners in the future.
Other structural damage to roads, retaining walls and houses ranged from 0 to 5%.
These rather low figures can be explained by the high compaction of the underlying media during their construction. Planners should consider that the more compacted underlying materials are, the greater is the likelihood that tree roots will spread close to the surface, and thereby damaging roads and SuDS structures.
Roots of trees planted in not compacted underlying soil media, for example, in parks, fields and footpaths, did not spread along the ground surface, but went Article No~e00154 deeper into the soil causing little or no damage to these structures. It follows that trees to be planted along streets in the future require more space and less compacted soil to reduce the risk of damage to nearby structures.
Considering that the proportion of trees causing damage to infrastructure is always relatively small, the corresponding sample size per species is also rather small.
This makes a statistical analysis rather challenging. Therefore, the authors recommend to undertake further studies on a much larger scale, and to focus only on a specific group of trees. Such studies should also assess tree trunk and root flare developments to provide a better understanding of root growth and development, particularly under structures such as pavements as well as the interaction of roots and trunk flare with pavements.

Declarations
Author contribution statement Miklas Scholz: Conceived and designed the experiments; Analyzed and interpreted the data; Wrote the paper.
Vincent Uzomah: Performed the experiments; Analyzed and interpreted the data; Wrote the paper.