Abstract
Background Urban forests are increasingly recognized as important tools in climate change mitigation and adaptation, prompting many cities to set tree canopy cover targets. However, current gaps in knowledge include understanding relationships and the feasibility of maximizing benefits between urban greening and other climate actions, such as densification. This study offers a data-driven and manageable framework for assessing current and anticipated future urban forestry conditions using spatial tree and built-form models.
Methods We spatially modelled 4 planting scenarios for increasing tree canopy cover by 2050 in a densifying neighbourhood in Vancouver, Canada, with low (< 10%) existing tree canopy.
Results Based on mortality assumptions, we aged out and replaced 1,853 to 2,445 trees since 2020. We added 6,079 to 11,726 trees across the 4 scenarios (10,228 to 15,823 total), increasing canopy cover from 7% in 2020 to a maximum of 16% by 2050. Despite rigorous tree planting, we were unable to achieve a 30% canopy cover target at neighbourhood scale. Tree replacement due to mortality was a major contributor to decreased canopy cover and volume in future scenarios. The 31% to 34% reduction in future canopy cover due to the replacement of aged-out trees was driven by changes on private parcels.
Conclusion Our systematic framework for generating and spatially modelling trees in a simulated future neighbourhood provides an opportunity for iteratively assessing multiple potential tree planting configurations. Future work for this project includes investigating social-ecological, outdoor shading, and building energy implications of various modelled urban forest strategies.
Introduction
Urban Forests as Tools in Climate Change Mitigation and Adaptation
Trees can improve the resilience and livability of urban environments through benefits such as heat reduction, flood mitigation, air purification (Livesley et al. 2016), and carbon sequestration (Nowak and Crane 2002), as well as improve the physical, mental, and social wellbeing of individuals (Hartig et al. 2014; Nesbitt et al. 2017; van den Bosch and Ode Sang 2017). At neighbourhood and property scales, trees can provide measurable climate and livability co-benefits, including improved thermal comfort of residents, natural cooling (indoor and outdoor), and increased climate resiliency (Bowler et al. 2010; Gupta and Gregg 2013; Ziter et al. 2019). Strategic tree planting may also improve walkability through increased comfort levels by defining pedestrian spaces and shading walkways (Langenheim et al. 2020). Greener areas of cities have been associated with higher levels of physical activity and active living amongst residents, even during extreme heat events (Villeneuve et al. 2018). In response to the recognition of the many benefits urban trees and vegetation provide, enhanced planning guidelines such as the “3-30-300 rule” (Konijnendijk 2022) articulate how cities can provide equitable and consistent access to trees and green spaces across scales.
Conflicts Between Urban Greening and Densification
It is crucial for cities to strategically plan for climate adaptation while efficiently managing their limited space and resources to meet local needs (Hamin and Gurran 2009). There is a gap in our understanding about how different urban forest strategies may conflict with other climate actions. For example, higher density development can decrease per person greenhouse gas emissions (Burton 2000; Norman et al. 2006) but makes it more challenging to provide greenery and parks (Haaland and Konijnendijk van den Bosch 2015). Space is often limited for urban trees (Koeser et al. 2013; Jim et al. 2018) and tree removals are also common during land development (Koeser et al. 2013; Guo et al. 2018). The densification of cities can also lead to unfavorable growing conditions for trees, such as limited root space due to conflicts with underground infrastructure, low surface permeability, and compacted soil (Diamond Head Consulting 2016; Jim et al. 2018). City trees are at greater risk of dehydration (Bassuk and Whitlow 1987), storm damage (Staudhammer et al. 2011), and spread of disease (Poland and McCullough 2006). To maximize urban greening benefits, urban forestry practitioners need to set policies and strategies within a context of increasing population densities and development (Cheng et al. 2021).
Opportunities for Tree Management and Planting with Densification
Despite the challenges that urban environments and densification exhibit on tree planting and maintenance, new developments can provide opportunities for increased future tree covers. Large comprehensive developments often require public realm and landscape plans, which include street and park trees. For example, the Southeast False Creek Community Public Realm development (includes the 2010 Olympic Village) in Vancouver, Canada, is a 32.5-ha mixed-use community with up to 13,000 residents and 10 ha of public parks and waterfront access (Mooney 2014; City of Vancouver 2023b; PWL Partnership [date unknown]). The development site was part of a larger industrial area since the late 1800s; in the 1960s, industry began to leave, and the land was rezoned for housing and parks by the late 1990s (City of Vancouver 2023b). As part of Phase 1 of the Southeast False Creek Community Public Realm development, 302 trees were planted (Mooney 2014; more current information is unavailable). Orthophoto imagery of the the Olympic Village area of the Southeast False Creek Community Public Realm development—collected before development (2006)(City of Vancouver Open Data Portal 2010) and several years after (2015)(City of Vancouver Open Data Portal 2020)—shows substantial increases in building density as well as greenspace and tree coverage (Figure S1).
Tree protection bylaws, if rigorous and regulated, can also ensure new developments do not needlessly remove mature existing trees (Clark et al. 2020; Ordóñez Barona et al. 2020). Haaland and Konijnendijk van den Bosch (2015) also suggest that compact city environments can facilitate the provision of trees and other vegetation by preserving and enhancing the quality of existing green spaces as well as allocating greenery in ways that increase visibility and visual quality.
Urban Forest in Vancouver, Canada
Over the past 30 years, Vancouver (British Columbia [BC], Canada) has maintained a relatively moderate-high level of vegetative greenness (Czekajlo et al. 2020) and exhibited an overall canopy cover of about 22% in 2020 (City of Vancouver and Vancouver Park Board 2020a). However, tree canopy is not distributed evenly across Vancouver. More mature neighbourhoods in the western side of Vancouver have higher canopy cover due to larger/older trees and more retention of forested lands, while the eastern portion of the city has smaller/younger trees as well as more sites with poor growing conditions due to lower quality or shallow soils, as well as more impervious surfaces (City of Vancouver and Vancouver Park Board 2018; City of Vancouver and Vancouver Park Board 2020a). Most of Vancouver’s tree canopy cover is associated with private land (37%)—mainly due to its large land base and not its expansive tree coverage. Trees on public streets also provide a substantial amount of canopy coverage (35%), while the remainder of Vancouver’s tree canopy is associated with parks (particularly select large, forested parks, like Stanley Park and Everett Crowley Park) and other civic property (City of Vancouver and Vancouver Park Board 2018).
The current discrepancies in neighbourhood-level tree canopy in Vancouver highlight the need for better planning for density and urban forests. Due to the City of Vancouver’s physical constraints (i.e., bordered by coastal waters and an adjacent municipality) and the lack of remaining open brownfields, development primarily occurs on parcels that have been built-up since 1971 (Statistics Canada 2016). From 1995 to 2014, Vancouver’s tree canopy had decreased primarily due to private developments; the declining trend may have been reversed by urban greening policies (City of Vancouver and Vancouver Park Board 2018; City of Vancouver and Vancouver Park Board 2020a). The City of Vancouver has set a canopy cover target of 30% by 2050; an 8% canopy cover increase from the 2020 level (City of Vancouver and Vancouver Park Board 2020a). Proposed planting measures to reach a city-wide 30% canopy cover by 2050 include tailoring tree canopy targets per land and housing types (City of Vancouver and Vancouver Park Board 2020a), as well as doubling street tree density in underserved and priority neighbourhoods (City of Vancouver and Vancouver Park Board 2018). However, protecting existing canopy is also crucial and addressed by the city’s urban forestry plan through action items such as developing policy for retaining soil and growing space for private trees, tracking pre- and post-construction tree canopy, and updating tree protection standards for public trees.
With an aspiration to reach 30% city-wide canopy cover by 2050, areas with low tree cover may provide the most feasible planting locations and could benefit the most from additional urban forestry resources (Nesbitt et al. 2019b). The City of Vancouver has also adopted an equity lens into its urban forestry programming and plans to focus tree planting initiatives in lower canopy and low-income neighbourhoods (City of Vancouver and Vancouver Park Board 2020a, 2020b). Additionally, to support regional growth, Vancouver will require substantial development to house an additional 260,000 persons (approximately 2,250 persons/km2 increase; City of Vancouver 2022). To ensure Vancouver does not experience increasing vegetation loss due to urbanization and climate change impacts such as droughts (Lantz et al. 2021), urban greening and densification must be assessed and planned for holistically.
Introduction to Study
The aim of this study was to assess the outcomes of various tree planting configurations in an urban area with low existing canopy cover that is projected to densify. Specifically, we investigated the following question: how do varying tree planting and management strategies impact the size and location of trees (and canopy cover and volume) by 2050 for a neighbourhood with low (< 10%) canopy cover in 2020 that expects a substantial increase in population (+30%)? We also investigated a secondary question, in alignment with City of Vancouver city-wide canopy cover targets and their prioritized planting initiatives for low canopy neighbourhoods (City of Vancouver and Vancouver Park Board 2020a), as well as the “30% canopy cover per neighbourhood” guideline for urban forestry and urban greening (Konijnendijk 2022). We asked: can the study neighbourhood achieve a 30% canopy cover by 2050 through any of the developed future “what-if” tree planting scenarios? The methods outlined in this paper provide a semi-automated approach to spatially model different tree planting strategies alongside urban development at the neighbourhood scale. By using existing research and evidence-based tree and building data, this work provides comparable projections of future (2050) urban forestry outcomes between various feasible tree planting strategies.
Materials and Methods
We developed spatial proxy models of current (2020) and 4 future (2050) urban forest scenarios for the neighbourhood-scale study area to assess outcomes of different tree planting strategies. Our study included: (i) generation of proxy tree data for the baseline 2020 condition; (ii) development of future scenario models using a semi-automated, rule-based procedure; and (iii) defining urban forestry indicators to assess outcomes.
Sandbox Study Area
The study area (i.e., “sandbox”) is a 1,600 by 1,600 metre (m) spatial proxy model that approximates spatial and non-spatial attributes of an area in Southeast Vancouver (Figure 1). Sandbox models, which typify urban form—including population density, street patterns, block sizes, parcel density and sizes, and land use proportions—were developed to represent the 2020 (baseline) and 2050 (future) conditions following methods by previous work conducted through the Energy Efficiency in the Built Environment project (Pacific Institute for Climate Solutions 2012; detailed by Salter et al. [2020], Lu et al. [2021], and Lu et al. [2023]). Specifically, the 2020 sandbox model was created by extracting simplified yet characteristic spatial representations of the real neighbourhood in Vancouver using available land use data (acquired from Metro Vancouver Open Data Portal [2023]), parcel and building information, including type, age, occupancy, height, and construction type and date (acquired from BC Assessment data [2019]), and local census population data (Statistics Canada 2017, 2019), as well as informed by current city plans and policies (City of Vancouver 2022). The developed future sandbox model for this study resembles an area that is expected to increase in population by approximately 30% by 2050, based on mid-point provincial projections with input from local planners (BC Stats 2023). The Vancouver study area population growth rate was estimated to be 10% per decade, which determined increases in residential and commercial floor areas (with subsequent increases in standardized resident values) in future scenarios.
Maps of the real neighbourhood (top left) located in the City of Vancouver (Southeast Vancouver region), British Columbia, Canada (bottom left), and representative “sandbox” model for baseline 2020 built-form condition (top right). The baseline 2020 sandbox model included 7,125 buildings, while building count in the 2050 model decreased by approximately 100 buildings (7,019 buildings total) due to increased dwelling density. Imagery from Planet Team (2017), land use data from Metro Vancouver Regional District (2020), parcel and dwelling unit counts based on BC Assessment (2019) data, and resident counts based on Statistics Canada (2017, 2019) data.
Baseline Tree Data
Aerial laser scanning (ALS; also referred to as LiDAR) data collected by the City of Vancouver in 2013 (City of Vancouver Open Data Portal 2013) provided information about tree planting location, height, and canopy spread for the baseline (2020) model. Individual tree canopy crowns in the ALS data were delineated using an object-based image analysis approach developed by Matasci et al. (2018) and visualized as crown radius-buffered circles. Tree heights were derived directly from the ALS data, whereas baseline 2020 tree locations in the sandbox were approximated from locations of trees in the real neighbourhood via ALS data. The Vancouver street tree inventory (City of Vancouver and Vancouver Park Board 2023) was used to identify the species of street trees for the baseline (2020) model. Species information for private and park trees were randomly assigned using a list of common trees observed in the real neighbourhood using Google Street View imagery.
Development of Future Planting Scenarios
Four future scenarios were developed based on policy and best practices research and sequentially increased in their boldness of tree planting goals (Table 1). The following policy actions informed the design of each future scenario: (i) tree planting rates on public and private lands; (ii) replacement of “aged-out” trees (i.e., trees removed between baseline and future scenario models due to mortality or development purposes); (iii) locations of trees on private parcels; (iv) proportions of tree size classes (i.e., large = mature tree height > 15 m; medium = mature tree height 10 to 15 m); and small = mature tree height < 10 m); (v) proportions of deciduous-type trees (i.e., conduct leaf abscission and typically broadleaf) versus coniferous-type trees (i.e., do not conduct leaf abscission and typically needleleaf; includes one deciduous conifer, Metasequoia glyptostroboides); and (vi) tree diversity proportions based on genus and species. Scenario 1 (S1) was the most conservative, as it adhered to the current existing species list (City of Vancouver 2011) and planting goals (City of Vancouver and Vancouver Park Board 2018) set by the city, while scenario 2 (S2) followed the same planting goals but used climate-adapted tree species (Diamond Head Consulting 2019a, 2019b). Scenario 3 (S3) utilized the same trees as scenario 2 but strategically placed newly added deciduous trees at the southwest corner of private buildings to target their shading potential. The boldest tree planting strategy was scenario 4 (S4), as it incorporated climate-adapted species and additional greenspace features, including partially forested parks and additional tree rows on blue-green infrastructure streets, as well as added more private trees with targeted south-side planting for building shading. Blue-green streets were designed following City of Vancouver (2023a) initiatives, which include trees and other green infrastructure intended to manage water cycles (e.g., water quality and flood risk management), enhance access to urban green space, and provide other ecosystem services, such as heat mitigation and air-quality improvement.
Summary of future planting scenario rules, including mortality rates, rates of replacement and additional trees, species lists and mixes, as well as the general placement of additional trees. “Aged-out trees” refers to trees removed between baseline and future scenarios. “Replacement trees” are new trees provided to future scenarios in place of aged-out trees. “Additional trees” are new trees that are provided to the future models to achieve scenario and location-based tree counts. CA (climate-adapted); L (large); M (medium); m (metre); S (small); SE (southeast); SW (southwest).
Future planting scenarios were systematically modelled through the application of a semi-automated workflow to produce spatial tree data (Figure 2). To model tree losses, trees in the baseline (2020) scenario were selected to be “aged out” (i.e., removed from future scenarios) based on location-based mortality rates, parcel development, and tree species’ climate adaptability and size. For each scenario, mortality rates were assigned to street, park/civic, and private trees, respectively, based on a recently completed literature review study on tree mortality rates (Hilbert et al. 2018). Mortality rates were based on location and adjusted for each scenario to reflect the effects of policy changes on tree survival and health conditions. For example, scenario 1 (i.e., representing “business as usual”) used the higher median annual mortality rates per planting location as reported in Hilbert et al. (2018; outlined in Table 1). Scenarios 2 and 3 were given mortality rate decreases (by 0.5%) to reflect the policy changes that would provide better tree planting and tree care, while scenario 4 had the lowest mortality rates to reflect the maximal level of efforts by the municipal government and other stakeholders (e.g., property owners) for planting and managing urban trees. Table S1 provides the climate-suitability of baseline tree species (used in determining the aged-out trees).
Semi-automated modelling procedure of tree quantifications, species selections, and locations. “Rules” refers to scripting through R software, except for manual placement (which was conducted in ArcGIS software). “Aged-out” trees refers to trees removed between baseline and future scenarios. “Existing” trees are trees that were not removed since baseline and remain in future scenarios. “Replacement” trees are new trees provided to future scenarios in place of aged-out trees. “Additional” trees are new trees that are provided to the future models to achieve scenario and location-based tree counts.
Following scenario and location-based mortality rates, trees retained from the baseline model in the future scenario models (i.e., “existing” trees) maintained their location and species attributes. Next, aged-out trees were replaced with new trees in future scenario models (i.e., “replacement” trees) following at least a 1:1 ratio to aged-out trees. Extra new trees were also added to future scenario models (i.e., “additional” trees) to achieve tree counts associated with scenario and location-based rules as well as current tree protection bylaws for redeveloped parcels (City of Vancouver 2010). Using the calculated number of replacement and additional trees required per location, we developed virtual tree “nurseries” following scenario-specific requirements of species and size. New trees’ species information, including mature tree height and deciduous/coniferous status, were compiled from various sources (primarily Breen 2022; also Kwantlen Polytechnic University [2015], Missouri Botanical Garden [date unknown], or Plants For A Future [2022] for missing information). We standardized 30-year tree heights per species and calculated them as two-thirds (i.e., 67%) of average mature tree height. Due to limited resources about mature crown spread, 30-year crown diameter was estimated as 45% of average mature tree height (i.e., crown radius = 22.5% of average mature tree height). Trees with standardized 30-year sizing, generated in the virtual tree “nurseries,” were then placed in the sandbox systematically; existing tree sizes were adjusted to reflect new tree sizing standards. A dataset of species information, including height, crown diameter, size class, and coniferous/deciduous status, for each scenario are also provided in Tables S2–S5.
Urban Forestry Indicators
Urban forestry indicators that were assessed for the baseline and future scenarios include tree statistics, canopy cover, and canopy volume. Tree statistics included counts and percentages of trees overall, as well as per location and size class. Canopy cover (ha, %) was calculated by using a merged polygon of crown radius-buffered trees. For location-specific analyses (i.e., street vs. private vs. park/civic), we clipped merged canopy cover per each parcel/street segment (i.e., trees from all planting locations contributed). Canopy coverage was reported as a total for the entire sandbox area, as well as per tree planting location, for each scenario. Total canopy volume (m3) was calculated for all trees of each scenario using the sum of individual tree volumes. The volume of each tree was calculated based on type: the volume of a hemisphere for deciduous trees (a) and the volume of a cone for coniferous trees (b), as follows:
a
b
where V is volume, r is crown radius, and h is height. The difference (%) in canopy cover (total, per planting location) between aged-out trees (with 30-year sizing) and replacement trees was also calculated per scenario to assess the impact of replacing existing trees. We performed Kruskal-Wallis tests (Kruskal and Wallis 1952; McKight and Najab 2010)—reported via chi-square statistic (χ2)—and pairwise Wilcoxon rank sum tests (Wilcoxon 1945) to assess variation in tree height, crown radius, canopy cover, and canopy volume per location and per tree size across scenarios. A similar analysis was also performed to compare aged out and replacement trees in future scenarios. Scenarios 2 and 3 were grouped together for tree height, crown radius, and canopy volume for Kruskal-Wallis and pairwise Wilcoxon tests because of identical tree characteristics. Significant difference was identified using P-value (p) < 0.05.
Systematic procedures were automated using R software (R Core Team 2017), including the following packages: ggplot2 (Wickham 2016), raster (Hijmans et al. 2022), rgdal (Bivand et al. 2022), sf (Pebesma et al. 2022), and tidyverse (Wickham et al. 2019). Manual editing and subsequent spatial analysis and mapping were performed using ArcGIS Pro (version 2.9.2).
Results
Tree Statistics
The systematic removal, replacement, and addition of trees in the sandbox model resulted in 10,228 to 15,823 trees across future scenarios 1–4, as seen in Table 2. The increase of trees between future scenarios and baseline was driven by additional trees: +103% from baseline to scenarios 1–3 and +181% from baseline to scenario 4. Most new trees (replacement and additional) were private—about double the amount of new street trees and 7 to 10 times the number of park/civic trees, for all future scenarios. Results presented as well as additional visuals related to this research can be explored on the Urban Greening & Urban Densification dashboard here: https://experience.arcgis.com/experience/011b5cf2809a4c79b02b112629a826ae.
Annual mortality rates (% per year) and tree counts of aged-out, existing, replacement, and additional trees per location for the baseline and each future scenario (S1–4). “Aged-out trees” refers to trees removed between baseline and future scenarios. “Existing trees” are trees that were not removed since baseline and remain in future scenarios. “Replacement trees” are new trees provided to future scenarios in place of aged-out trees. “Additional trees” are new trees that are provided to the future models to achieve scenario and location-based tree counts. S2 and S3 share tree counts and characteristics, other than S3 having some strategically placed private trees to target their shading potential for buildings. CA (climate-adapted); NA (not applicable).
Tree type, size, and planting location varied across scenarios as a result of pre-determined mortality rates and planting strategies for additional trees (Figure 3; more details provided in Table S6). As more trees were added to private parcels than any other location, the dominant location of trees shifted from streets in baseline (62%) to private in future scenarios (53% to 54%). More specifically, the percentage of street trees decreased by 22% from baseline to scenarios 1–4, while the proportion of private trees doubled (+27% of total). The proportion of park/civic trees remained relatively constant across all scenarios. The amount of coniferous trees increased by 6% to 10% in scenarios 2–4 compared to baseline.
Count and percentage of trees per location (private, park/civic, street) and size (S – small, M – medium, L – large) for baseline 2020 (B) and each future scenario (S1–S4).
Overall, tree heights and crown radii significantly decreased between baseline and future scenarios (χ2tree height [3, N = 40,369] = 3,949.5, p < 0.05; χ2crown radius [3, N = 40,369] = 2,248.1, p < 0.05). The distribution of crown radii per planting location for each scenario is illustrated in Figure 4; similar patterns were exhibited by tree heights (Figure S2). Most notably, private tree heights decreased from baseline to future scenarios by approximately 5.0 m but increased from scenario 1 to scenarios 2–4. After baseline, scenario 4 street tree heights were greatest. Baseline park tree heights were greater than all future scenarios by approximately 5.5 m, followed by scenario 1 and then scenarios 2–4. Similar patterns were found for tree crown radii in baseline and scenarios. Private tree crown radii varied the most, with baseline the greatest, followed by scenario 4, scenarios 2 and 3, and scenario 1, respectively. Park tree crown radii were greatest for baseline, then scenario 1 and scenario 4, and lowest for scenarios 2 and 3. Street tree crown radii followed the same pattern as for private tree crown radii and street tree heights. Mean, median, standard deviation, and interquartile range values for tree height and crown radius of each planting location, per scenario, are provided in Table S7.
Boxplot of crown radius (m) for baseline and future scenarios (S1, S2/S3, S4) per planting location (park/civic, private, street). Within boxes, solid black lines denote median values while large solid black dots indicate mean values. Boxes extend from the 25th to the 75th percentile of each group’s distribution of values (i.e., interquartile range [IQR]); vertical lines extend to the “minimum” (i.e., 25th percentile – [1.5 × IQR]) and “maximum” (i.e., 75th percentile + [1.5 × IQR]); small black dots denote observations outside the range of adjacent values (i.e., outliers). Mean, median, and IQR values are outlined in Table S7.
Both small and medium trees were dominant in future scenarios, except for scenario 4, which had more medium (49%) than small (37%) trees (Figure 3). Dominant tree sizes followed dominant locations across scenarios, whereby most small and medium trees were located on streets in baseline (small = 26% and medium = 23% of total) and private for future scenarios (scenarios 1–3: small = 27% and medium = 25% of total; scenario 4: small = 24% and medium = 29% of total). Large trees in baseline were mainly found on streets (13% of total), and many were replaced/supplemented with small or medium trees in future scenarios (5% to 7% new large trees in scenarios 1–4).
Across all size classes, baseline generally had the tallest trees (meansmall = 9.89 m; meanmedium = 11.6 m; meanlarge = 15.8 m). Within the large size class only, scenario 1 contained taller trees compared to other future scenarios (meanS1 = 17.6 m; meanS2 ≈ meanS3 = 14.0 m; meanS4 = 13.7 m). However, scenario 1 had shorter small (meanS1 = 4.47 m; meanS2 ≈ meanS3 = 5.38 m; meanS4 = 5.41 m) and medium (meanS1 = 8.26 m; meanS2 ≈ meanS3 = 8.95 m; meanS4 = 8.98 m) sized trees compared to scenarios 2 and 3 and scenario 4. Baseline observed the greatest crown radii for small and medium trees (meansmall = 3.03 m; meanmedium = 3.41 m), while scenarios 2 and 3 and scenario 4 had greater crown radii than scenario 1 for small (meanS1 = 1.48 m; meanS2 ≈ meanS3 = 1.77 m; meanS4 = 1.78 m) and medium (meanS1 = 2.73 m; meanS2 ≈ meanS3 = 2.95 m; meanS4 = 2.94 m) sized trees. Crown radius for large trees was greatest for scenario 1, although not significantly different from scenario 4 (p = 0.09; meanB = 4.08 m; meanS1 = 4.72 m; meanS2 ≈ meanS3 = 4.56 m; meanS4 = 4.36 m).
Canopy Cover and Volume
The dominant contributor to canopy cover shifted from street trees in 2020 to both street and private trees in scenarios 1–3, and mainly private trees in scenario 4 (Table 3). Through the addition of 6,138 trees in scenario 1, the total canopy cover and volume grew by 2% and 39% since baseline, respectively. Compared to baseline, scenario 2 and scenario 3 canopy covers increased by 1%, while their total canopy volumes increased by 18%. Scenario 4 resulted in the greatest total canopy cover (+9% from baseline) and canopy volume (+92% from baseline) with the addition of almost 12,000 trees. An increase in overall tree density was evident from baseline through future scenarios, with hotspots concentrated in park/civic parcels as well as select streets (Figure 5). Scenario 4 shows the most and highest tree densities, specifically in its partially forested parks and along blue-green streets.
Summary of canopy cover (ha and % of total sandbox area) attributed to trees planted in each location (street, park/civic, private) and total for each scenario. CA (climate-adapted); NA (not applicable).
Maps of baseline (2020) and 4 future scenario (S1–4; 2050) models showing tree density (gaussian kernel per 1 m2 grid size; weighted by crown radius) and individual trees sized by crown radius.
Total canopy cover (per all parcels and street segments) significantly increased from baseline to future scenarios (χ2 [3, N = 30,896] = 5,894.6, p < 0.05), as well as between scenarios (meanB = 3.42%; meanS1 = 6.59%; meanS2 ≈ meanS3 = 7.79%; meanS4 = 13.20%). Canopy cover also increased on the street and parcel (private and park/civic) level from baseline and through scenarios 1–4. The greatest mean canopy coverage was for scenario 4 park/civic parcels (meanB = 15.30%; meanS1 = 23.80%; meanS2 ≈ meanS3 = 22.80%; meanS4 = 33.40%); however, there was no significant difference between any scenario. Scenario 4 street-specific canopy cover was also the greatest across scenarios (meanB = 13.30%; meanS1 = 17.20%; meanS2 ≈ meanS3 = 18.1%; meanS4 = 25.5%; no significant difference between scenario 1, scenario 2, and scenario 3). Private parcel canopy covers were the lowest overall (meanB = 2.70%; meanS1 = 5.76%; meanS2 ≈ meanS3 = 7.00%; meanS4 = 12.2%).
Individual tree canopy volumes significantly decreased from baseline to all future scenarios (χ2 [3, N = 40,369] = 2165.1, p < 0.05), with further decreases from scenario 1 to scenario 4 to scenarios 2 and 3 (meanB = 157.0 m3; meanS1 = 87.1 m3; meanS2 ≈ meanS3 = 74.1 m3; meanS4 = 78.0 m3). Similar to tree height and crown radius, individual private tree canopy volume decreased from baseline to future scenarios, with scenario 4 observing the greatest future private tree canopy volume (meanB = 169.0 m3; meanS1 = 34.4 m3; meanS2 ≈ meanS3 = 49.4 m3; meanS4 = 50.4 m3). Individual street tree canopy volume was greatest in baseline (meanB = 126.0 m3), and then future scenarios in order of scenario 1 (meanS1 = 109.0 m3), scenario 4 (meanS4 = 94.6 m3), and scenarios 2 and 3 (meanS2 ≈ meanS3 = 84.7 m3). Individual park tree canopy volume was variable and not statistically significant between future scenarios. Large trees increased in per-tree canopy volume from baseline (meanB = 278.0 m3) to future scenarios, with the greatest increase attributed to scenario 1 (meanS1 = 499.0 m3; meanS2 ≈ meanS3 = 335.0 m3; meanS4 = 288.0 m3). Medium trees observed the greatest decrease in per-tree canopy volume from baseline to future scenarios (meanB = 148.0 m3; meanS1 = 49.3 m3; meanS2 ≈ meanS3 = 67.7 m3; meanS4 = 64.8 m3). Small trees also decreased in individual tree canopy volume from baseline to future scenarios (meanB = 90.8 m3; meanS1 = 7.44 m3; meanS2 ≈ meanS3 = 12.7 m3; meanS4 = 13.0 m3).
Aged-Out and Replacement Trees
Although aged-out trees were replaced at a rate of at least 1:1, canopy cover associated with replacement trees was lower than that of the original aged-out trees (using same 30-year sizing; Table 4). All scenarios experienced a 31% to 34% decrease in total canopy cover with the replacement of aged-out trees. A decrease in canopy cover associated with aged-out and replacement trees was greatest on private parcels for all future scenarios. This decrease in private tree canopy cover was up to 7 times greater than other locations (i.e., scenario 4 private vs. street). Scenario 4 also experienced the greatest loss of canopy cover due to aged-out trees overall (−34%). The lowest decrease in park/civic canopy cover occurred in scenarios 2, 3, and 4. Street canopy cover change was minimal across all future scenarios (10% to 12%). We found significant decreases in parcel- and street-segment–level mean canopy cover attributed to replacement of aged-out trees across scenarios (χ2 [7, N = 30,896] = 96.9, p < 0.05; meanS1 ≈ meanS2 ≈ meanS3 = −41%; meanS4 = −48%). Across scenarios, we found the greatest reduction in mean canopy cover on private parcels (meanS1 = −59%; meanS2 ≈ meanS3 = −56%; meanS4 = −60%), although with variable significant differences. Park/civic parcel mean canopy cover reductions were also substantial (meanS1 = −37%; meanS2 ≈ meanS3 = −56%; meanS4 = −57%). Again, we found variable significant differences in mean canopy cover of aged-out and replacement trees on the park/civic parcellevel. Street-level mean canopy cover was the most consistent between aged-out and replacement trees (meanS1 = −17%; meanS2 ≈ meanS3 = −14%; meanS4 = −13%).
Summary of canopy cover (ha and %) of total sandbox area, as well as parcel- and street segment-based mean, associated with trees aged out since baseline (2020) and replacement trees for each scenario. “Aged-out trees” refers to trees removed between baseline and future scenarios. “Replacement trees” are new trees provided to future scenarios in place of aged-out trees. The percent decrease (%) between aged-out and replacement trees’ canopy cover was also calculated: ([Aged-out – Replacement] ÷ |Aged-out|) × 100%.
Following canopy cover patterns, we generally found significant decreases in tree height (χ2 [17, N = 13,062] = 9,987.9, p < 0.05), crown radius (χ2 [17, N = 13,062] = 9,495.9,p < 0.05), and canopy volume (χ2 [17, N = 13,062] = 9,716.9, p < 0.05) between aged-out and replacement trees across size classes for each scenario (baseline, scenario 1, scenarios 2 and 3, and scenario 4). Similarly, we also noted significant differences across locations for tree height (χ2 [17, N = 13,062] = 1,884.0, p < 0.05), crown radius (χ2 [17, N = 13,062] = 1,500.5, p < 0.05), and canopy volume (χ2 [17, N = 13,062] = 1,599.6, p < 0.05). Mean tree height, crown radius, and canopy volume of aged-out and replacement trees per scenario are summarized per location in Table 5 and per size class in Table 6. Private trees and small trees were each consistently and significantly smaller in terms of tree height, crown radius, and canopy volume for replacement trees compared to aged-out trees across scenarios. For private trees, scenario 1 observed the greatest differences in tree height (−39%), crown radius (−35%), and canopy volume (−86%) between aged-out and replacement trees. Park/civic trees in scenarios 2 and 3 and scenario 4 also observed similar significant losses in tree height, crown radius, and canopy volume. Scenario 1 also observed the greatest differences for small aged-out/replacement trees in terms of tree height (−34%), crown radius (−30%), and canopy volume (−78%). Large trees also showed substantial loss in tree height for scenario 1 and scenarios 2 and 3.
Summary of mean tree height (m), crown radius (m), and canopy volume (m3) for aged-out trees and replacement trees per location for each scenario. “Aged-out trees” refers to trees removed between baseline and future scenarios. “Replacement trees” are new trees provided to future scenarios in place of aged-out trees. The percent change (%) between aged-out and replacement trees’ measures was also calculated: ([Aged-out – Replacement] ÷ |Aged-out|) × 100%.
Summary of mean tree height (m), crown radius (m), and canopy volume (m3) for aged-out trees and replacement trees per size class for each scenario. “Aged-out trees” refers to trees removed between baseline and future scenarios. “Replacement trees” are new trees provided to future scenarios in place of aged-out trees. The percent change (%) between aged-out and replacement trees’ measures was also calculated: ([Aged-out – Replacement] ÷ |Aged-out|) × 100%.
Discussion
We presented a systematic and evidence-based method to spatially model current and potential future urban forest conditions. We aged out and replaced 1,853 to 2,445 trees since 2020 and also added 6,079 to 11,726 trees, totaling 10,228 to 15,823, across the 4 future (2050) scenarios. The dominant location of trees shifted from streets in baseline to private lands in future scenarios, with most trees being small or medium across all future scenarios. Starting with a 7% canopy cover in 2020, we were able to achieve a maximum canopy cover of 16% by 2050 (scenario 4). Substantial canopy loss was associated with mature trees aging out and being replaced by small trees. The framework of tree data creation and spatial modelling presented allows for assessing multiple potential futures within simulated real-world and projected constraints at the neighbourhood level.
Analytical Approach
A highlight of this research is the development of a semi-automatic procedure to spatially model tree mortality and replacement/additional tree counts and placement using previous research and open-source data and software. Similar work incorporated aspects of spatial-functional design to assess various benefits (e.g., Bodnaruk et al. 2017; Langenheim et al. 2020) and/or feasibility (e.g., Locke et al. 2011; Danford et al. 2014; Choi and Lee 2022) of potential tree planting strategies. Most previous literature is grid- and/or tract-based, which restricts the assessment of individual tree-based planting decisions. On the other hand, Choi and Lee (2022) developed a tree-selection and planning optimization model to minimize cost while maximizing CO2 absorption and species diversity for a development area. However, the complexity of the tree-selection and planning optimization model requires powerful proprietary optimization software, and the case study area is relatively small compared to city size. Additionally, Locke et al. (2011) developed needs-versus suitability-based urban tree canopy spatial models of New York City at the parcel level. Others, such as Langenheim et al. (2020), modelled individual trees in three-dimensions using idealized forms—a method that is too detailed and computationally intensive for neighbourhood-scale analysis presently.
The sandbox model of this study includes locally calibrated built-form typologies based on previous work (Salter et al. 2020; Lu et al. 2021), which provides a consistent and approachable method to spatially model intricate building and tree changes at a neighbourhood scale. As our model generalizes a real neighbourhood, outcomes of this research cannot be directly applied. Instead, the sandbox model provides opportunity for efficient iterative simulation of different policy options for comparable neighbourhoods of differing sizes, geographic, climatic, and potential future contexts. We also employed semi-automatic methods to systematically apply tree-related planting rules, which improves tuning ability and replicability of the tree models for various other areas, planting strategies, and tree characteristics.
This research also relied on existing datasets for baseline and future tree models, including tree planting location, height, and crown radius from ALS data, for developing a simplified yet representative baseline urban forest condition. We also used standardized and open-source tree information for a consistent approach to size tree species in future scenarios—similarly to Scholz et al. (2018). As we were unable to perform dynamic tree growth modelling, we were limited to constant 30-year sizing for all future trees (existing and new), which may have underestimated future measures like canopy cover. Despite this generalization of individual tree growth, canopy cover achieved by all future scenarios was comparable with existing Vancouver neighbourhoods of similar tree counts (proportional to neighbourhood area), particularly Grandview-Woodland, Oakridge (scenarios 1–3), and Kitsilano (scenario 4; Lu et al. 2022).
Outcomes of Future Urban Forest Strategies
We modelled 4 potential tree planting strategies, including more progressive approaches with bluegreen streets and partially forested parks (scenario 4), for a densifying neighbourhood and assessed tree type, location, and size, as well as canopy cover and volume. Overall, canopy volume increased from baseline to future scenarios by as much as 67% in scenario 4. However, future scenarios received at least double (almost triple for scenario 4) the number of trees in the baseline condition. Also, our most progressive scenario (scenario 4) attained a canopy cover of only 16%, falling short of the 30% neighbourhood canopy cover goal outlined by the “3-30-300 rule” (Konijnendijk 2022). Additionally, if a Vancouver neighbourhood like the sandbox model (including same area) underwent the boldest planting strategy (scenario 4), the scale of changes would not be large enough to substantially increase the city-wide canopy cover (i.e., approximately 1% increase). As outlined by the City of Vancouver and Vancouver Park Board (2020a), achieving the 30% city-wide canopy cover target will require tree planting efforts beyond parks and already-occupied low-cost planting locations (e.g., no boulevard pavement or utility conflicts). Example tree planting and retention solutions that the city and park board provide includes (but not limited to) the conversion to climate-adapted tree species, installation of new planting cut-outs in predominantly paved areas, and the integration of trees in blue-green infrastructure—all aspects covered by scenario 4. Therefore, if the City of Vancouver desires to reach a canopy cover target of 30% by 2050 and focus planting efforts in lower canopy cover neighbourhoods (City of Vancouver and Vancouver Park Board 2020a), considerable investments into tree planting and management as well as design and planning are required across a large proportion of the city’s land base.
The distribution of trees within the sandbox model for each scenario depended largely on pre-determined decisions (e.g., city-wide planting rates and greater focus on private planting). Specifically, the shift in dominant planting location from street (baseline) to private (future) emphasizes the opportunity private landowners have in shaping urban forests. Cities may need to emphasize more private planting and management for greater shading and other benefits, as well as to reach their tree canopy target (Daniel et al. 2016; City of Vancouver and Vancouver Park Board 2020a; Morgan and Ries 2022). The need for private tree planting interventions is particularly noteworthy. For example, despite scenario 4’s ambitious private tree planting strategy (i.e., 100% of private parcels with at least 2 trees), it still fell short of the neighbourhood-level tree canopy target.
An underlying assumption of the scenarios (especially scenario 4) is that residents will accept more trees in their yards given various benefits that urban trees provide. However, not everyone likes trees (Kirkpatrick et al. 2012). Our study highlights previous literature that states the importance of community engagement on tree planting and management, especially of private trees (Ordóñez-Barona et al. 2021). Cities will need creative and effective mechanisms to encourage private tree planting and protection in order to reach their aggressive urban forest goals. For example, cities could take advantage of homeowner satisfaction with neighbourhood trees (or dissatisfaction with the lack thereof) to implement private planting initiatives that property owners support and will contribute to maintaining (Coleman et al. 2023). Other private tree planting incentives could include rebates, such as the ‘Treebate’ program in Portland (Oregon, USA), which provides a utility bill credit for stormwater fee payees that plant trees on their residential private properties (Carney et al. 2021; Carney et al. 2022). Greater integration of recognitional equity strategies in municipal tree planting and management initiatives may also aid in more successful receptions of private trees (Nesbitt et al. 2019a).
Lower than expected canopy cover may also be due to some tree size choices for future scenarios, as small and medium trees were favored in future scenarios due to their greater climate adaptability (Diamond Head Consulting 2019a, 2019b) and higher planting feasibility in space-limited densified neighbourhoods (Guo et al. 2019; Koeser et al. 2022). However, tree density hotspots were still most dominant on streets and public parcels for all future scenarios—highlighting the importance of suitable public tree management. Additionally, as many street trees were aged out and replaced with new small or medium trees by 2050, best practices for urban tree planting and maintenance would be required for healthy establishment (Roman et al. 2015; Khan and Conway 2020).
A crucial topic of debate in urban forest management is the divergent responsibilities of managing existing trees to increase likelihood of survival or investing in replacement trees when trees age out. Due to the replacement of aged-out trees since baseline 2020 (30-year sizing for both), we found a decrease in all measured urban forestry indicators across scenarios, including a 31% to 34% reduction in overall canopy cover. Canopy cover losses due to the replacement of aged-out trees were driven by changes on private parcels, with a 45% to 60% reduction on private parcels (mean reduction of 56% to 60%) across all future scenarios. In contrast, canopy cover reductions on park/civic parcels varied much more per scenario, while street-based canopy cover reductions were relatively minor. Our results emphasize the vital importance of existing mature trees for reaching canopy cover targets and providing other benefits. We found that differences in canopy cover attributed to aged-out trees and replacement trees were due to overall decreased crown radii, particularly of small trees, but also large trees in scenarios 2 and 3. The high impact of canopy cover changes attributed to small aged-out/replacement trees is likely due to their disproportionate number across scenarios compared to other size classes (details provided in Table S8; a summary of existing, replacement, and additional trees per location is also provided in Table S9). It also highlighted the importance of integrated and early planning of urban development and urban forests, as the decisions urban foresters make now will likely have significant legacy impacts on urban forests of the future (Healy et al. 2022).
Limitations and Future Directions
This work incorporated species suitability and diversity in future tree planting scenarios using climate-adapted trees for scenarios 2–4 and diversity rules of ≤ 30% per family, ≤ 20% per genus, and ≤ 10% per species (Santamour 1990). However, less attention was provided for some species-specific planting decisions, resulting in less favored species mixes per location for some scenarios (e.g., some large coniferous trees planted on streets in scenario 1). Mortality rates, specific to planting location and scenario, were generalized using existing literature (Hilbert et al. 2018) and applied based on tree species mature size and climate suitability; however, other finer-scale or locationbased factors not considered in this research could also impact tree mortalities. Future research should explore sensitivity testing of different input variables and rules (e.g., tree species/sizes and specific planting locations, as well as mortality rates and prioritization) that could change the modelling outcomes. Due to the generalized modelling approach, tree heights and crown radii—as well as calculated canopy covers and volumes—provide rough estimates. First, tree heights and crown radii were standardized per species using average mature size ranges provided by open-source datasets. Realistically, tree heights and crown radii would vary within species groups and may also vary due to different growing conditions. Additionally, due to limited species-specific allometric information, we did not stagger tree mortality and planting between 2020 and 2050 to model variable tree ages (and sizes). Pretzsch et al. (2015) note the difficulty in acquiring age-diameter-crown relationships from field samples and the inaccuracy of constructing generalized allometric relationships due to the high variability of growth rates and crown spreading within species and genera. Baseline 2020 tree sizes are most realistic, as they were directly acquired from ALS data of trees found in the real neighbourhood. Second, simplified tree crown shapes used to calculate canopy cover (i.e., crown radius buffered circle) and volume (cones for conifers, hemispheres for deciduous trees) also contributed to generalized calculations. Lastly, 2013 ALS data were used as input information for the baseline 2020 tree model, as more current data were unavailable at the time of this study; more up to date data could provide more representative baseline tree conditions and potentially modify outcomes in comparison with future scenarios.
Although canopy cover and volume provide important insight into the state of potential future urban forests, these measures alone cannot provide a comprehensive understanding of the function and benefits of urban forests. Additional measures such as relative tree diameter and condition of publicly owned trees should be considered in future work (Kenney et al. 2011). Future studies should also evaluate more benefits and implications of various tree planting strategies, including biodiversity and ecosystem services (Mooney 2014), shading and outdoor comfort (Bowler et al. 2010; Aminipouri et al. 2019), and indoor energy and building emissions (Salter et al. 2020). Additional modelling using other greening methods, such as vertical and rooftop gardens, could be assessed for climate and other benefits, as well as their potential contribution to achieving at least 30% canopy cover at the neighbourhood scale.
Conclusions
Through this research we present a data-driven and manageable framework for assessing current and anticipated future urban forestry conditions using spatial tree and built-form models. Despite progressive interventions, the “sandbox” model neighbourhood achieved a maximum 2050 canopy cover of 16%—about half of the 30% neighbourhood canopy cover goal as per the “3-30-300 rule” (Konijnendijk 2022). The relative impact of the most ambitious tree planting strategy (scenario 4) was minimal with respect to the city-wide canopy cover target (City of Vancouver and Vancouver Park Board 2020a). Limitations to attaining greater canopy cover included existing low tree counts in 2020, the existing reservation to tree planting by private landowners, the mortality and replacement of trees, and physical space post-densification. Other indicators such as relative tree diameter and tree condition could provide a more complete understanding about potential future urban forest conditions. Additional research using the sandbox and tree models is planned, particularly to investigate implications of differing urban forest strategies on social-ecological systems, outdoor shading, as well as building energy.
Conflicts of Interest
The authors reported no conflicts of interest.
Acknowledgements
The authors thank Dr. Yuhao Lu, Noora Hijra, Samantha Miller, Nicholas Martino, Taelynn Lam, Kanchi Dave, Emma Gosselin, and Jennifer Reid for their research and visualization assistance. This project is supported by the Social Sciences and Humanities Research Council (SSHRC; grant number #892-2020-1038).
Appendix
Climate-suitability of baseline tree species used in the determination of tree mortality for future scenarios. Based on Metro Vancouver climate-adapted tree species database prepared by Diamond Head Consulting (2019a, 2019b). Size class was based on species-specific mature tree height: > 15 m (large); 10-15 m (medium); and < 10 m (small). Size information was primarily sourced from Breen 2022; missing information sourced from Kwantlen Polytechnic University (2015), Missouri Botanical Garden ([date unknown]), or Plants For A Future (2022).
Summary of baseline tree data, including species name (scientific), tree height and crown diameter (m; split into lower and upper range values), tree type (coniferous or deciduous), size class, and total count, as well as count per planting location (park/civic, private, street). Size class was based on species-specific mature tree height: > 15 m (large); 10-15 m (medium); and < 10 m (small). Size information was primarily sourced from Breen 2022; missing information sourced from Kwantlen Polytechnic University (2015), Missouri Botanical Garden ([date unknown]), or Plants For A Future (2022). Tree status is not included for the baseline tree data summary, as all trees are considered existing.
Summary of S1 tree data, including species name (scientific), tree height and crown diameter (m), tree type (coniferous or deciduous), size class, and total count, as well as count per planting location (park/civic, private, street) and count per tree status (existing, replacement, additional). Size class was based on species-specific mature tree height: > 15 m (large); 10-15 m (medium); and < 10 m (small). Size information was primarily sourced from Breen 2022; missing information sourced from Kwantlen Polytechnic University (2015), Missouri Botanical Garden ([date unknown]), or Plants For A Future (2022).
Summary of S2 tree data (same data for S3), including species name (scientific), tree height and crown diameter (m), tree type (coniferous or deciduous), size class, and total count, as well as count per planting location (park/civic, private, street) and count per tree status (existing, replacement, additional). Size class was based on species-specific mature tree height: > 15 m (large); 10-15 m (medium); and < 10 m (small). Size information was primarily sourced from Breen 2022; missing information sourced from Kwantlen Polytechnic University (2015), Missouri Botanical Garden ([date unknown]), or Plants For A Future (2022).
Summary of S4 tree data, including species name (scientific), tree height and crown diameter (m), tree type (coniferous or deciduous), size class, and total count, as well as count per planting location (park/civic, private, street) and count per tree status (existing, replacement, additional). Size class was based on species-specific mature tree height: > 15 m (large); 10-15 m (medium); and < 10 m (small). Blue-green streets include trees and other green infrastructure designed to manage water cycles (e.g., water quality and flood risk management), enhance access to urban green space, and provide other ecosystem services, such as heat mitigation and air quality improvement (City of Vancouver 2023a). Size information was primarily sourced from Breen 2022; missing information sourced from Kwantlen Polytechnic University (2015), Missouri Botanical Garden ([date unknown]), or Plants For A Future (2022).
Summary of tree count (and percentage of total trees) for each location, size class, and tree type per scenario (baseline and S1-4). Size class was based on species-specific mature tree height: > 15 m (large); 10-15 m (medium); and < 10 m (small). CA (climate-adapted); CoV (City of Vancouver).
Summary statistics of tree height and crown radius per location and scenario, including mean, median, standard deviation (SD), and interquartile range (IQR).
Summary of tree count (and percentage of total trees) for existing, replacement, and additional trees per size class for each future scenario (S1–4). “Existing trees” are trees that were not removed since baseline and remain in future scenarios. “Replacement trees” are new trees provided to future scenarios in place of aged-out trees. “Additional trees” are new trees that are provided to the future models to achieve scenario and location-based tree counts. Size class was based on species-specific mature tree height: > 15 m (large); 10-15 m (medium); and < 10 m (small). CA (climate-adapted); CoV (City of Vancouver).
Summary of tree count (and percentage of total trees) for existing, replacement, and additional trees per location for each future scenario (S1–4). “Existing trees” are trees that were not removed since baseline and remain in future scenarios. “Replacement trees” are new trees provided to future scenarios in place of aged-out trees. “Additional trees” are new trees that are provided to the future models to achieve scenario and location-based tree counts. CA (climate-adapted); CoV (City of Vancouver).
Maps of the Southeast False Creek Community Public Realm development in Vancouver, Canada for 2006 (prior development; orthophoto imagery retrieved from City of Vancouver Open Data Portal 2010) and 2015 (post development; orthophoto imagery retrieved from City of Vancouver Open Data Portal 2020).
Boxplot of tree height (m) for baseline and future scenarios (S1, S2/S3, S4) per planting location (park/civic, private, street). Within boxes, solid black lines denote median values, while large solid black dots indicate mean values. Boxes extend from the 25th to the 75th percentile of each group’s distribution of values (i.e., interquartile range [IQR]); vertical lines extend to the “minimum” (i.e., 25th percentile – [1.5 × IQR]) and “maximum” (i.e., 75th percentile + [1.5 × IQR]); small black dots denote observations outside the range of adjacent values (i.e., outliers). Mean, median, and IQR values are outlined in Table S7.
- © 2023 International Society of Arboriculture
Literature Cited
Jump to section
Related Articles
Cited By...
- No citing articles found.