Trees are some of the most prominent natural features in towns and cities from both visual and functional perspectives. The urban forest is a key type of green infrastructure system [1
] and a consequential component of other urban ecosystems and landscapes [2
]. Comprised of diverse tree species and vegetation structures, the urban forest includes individual trees, assemblages of trees in parks, groves, and extensively forested natural areas, which are distributed across public and private properties and along streets, waterfronts, railways, and riverbanks [5
The ecological functions and services of urban forests have been investigated extensively in recent decades [7
]. Benefits include the ability of trees to reduce greenhouse gases through carbon storage [12
], decrease stormwater runoff through interception and absorption of rainwater [15
], and mitigate the urban heat island effect through reductions in surface and air temperatures at a local scale [16
]. However, knowledge about the relationship between urban trees and human health is still developing. The academic literature on the linkages between nature and human health has grown rapidly using various specifications of nature, such as urban greening, green space, open space, parks, therapeutic landscapes, and restorative settings. As the evidence base has expanded, reviews have consolidated knowledge of associated health outcomes, but many have focused broadly on nature [17
], green space [23
], and greenness [22
More information about specific qualities of urban tree conditions and exposures are needed in order to help guide and inform planning, design, and implementation decisions. Local governments and other organizations show increased interest in promoting and enhancing community-based nature as a social determinant of health [30
]. From a practical standpoint, effective implementation requires better articulation of specific elements of nature and how they may influence health outcomes. Policy, professional staffing, and budgets are often allocated less to generalities of nature and more specifically to departmental administrations addressing parks, trees, vegetation in rights-of-way, natural areas or landscapes associated with development.
While other studies have focused on nature, green space and greenness, no comprehensive review has assessed the full range of evidence on the human health responses associated specifically with trees in urban areas. A review by Salmond et al. focused on street trees and health [32
] but did not address other forms and configurations of the urban forest. To address this gap in the literature, this scoping review synthesizes empirical findings about how urban trees and forest experiences can impact human health. Scoping reviews are used to assess “the extent, range, and nature of research activity in a topic area” [33
] (p. 371) and are used for ‘reconnaissance’ to clarify working definitions and conceptual boundaries of a topic or field, especially for a large and diverse body of literature that is not amenable to a systematic review [34
]. We conducted a literature screening across varied disciplines (including epidemiology, medicine, environmental and atmospheric sciences, psychology, and other social sciences), sources and study methods, then summarized results in a conceptual framework that can inform future research questions and methods.
This scoping review also has practical implications. In the United States, urban tree canopy cover is estimated to be declining at a rate of roughly four million trees per year due to rapid urbanization and tree diseases and pests [36
]; other countries may be experiencing similar losses. Understanding how urban trees are associated with human health can inform how health professionals, urban foresters, town and city planners, and urban designers can maximize the public health benefits generated by urban trees by supporting better tree policy, planning, and management.
2.1. Search and Inclusion Strategies
The review was conducted in accordance with PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [37
]. Published findings of quantified relationships between urban trees and human health were collected by searching Embase, Ovid Medline, PsycINFO, and PubMed. Keywords used included terms related to trees, land cover, urban settings, public health and known health outcomes (Table 1
The initial search produced a collection of 3358 peer-reviewed articles; the number was reduced to 1663 after duplicates and non-relevant publications were removed (Figure 1
). The resulting article collection was limited to English language and peer-reviewed journal articles; no other limits were applied.
Search results were qualitatively screened independently by two authors based on titles, abstracts, keywords and study approaches. Articles were retained based on the following inclusion criteria: (a) original research article; (b) use of quantitative or mixed-methods to quantify the associations between urban trees and human health; (c) with “human health” referring broadly to illness, disease, symptoms, and wellness pathways such as improved physical and mental health conditions, healthy behaviors, and known environmental health risks (e.g., air quality, heat, ultraviolet radiation (UVR) exposure, and noise); (d) tree exposure measures (i.e., subjective and objective) or an explicit reference to trees in the findings; (e) with “trees” expressed as streetscapes, single trees, canopy cover, trees in a park, forested natural areas, and photos or videos of trees and landscapes. Note that studies using measures of vegetation cover (such as normalized difference vegetation index (NDVI)) were only included if results specified tree-associated outcomes; (f) with “urban” expressed as a study conducted in urban settings and/or comprised of urban populations; and g) all potential human beneficiary populations (i.e., age, gender, cultural background, and socio-economic status).
Studies that implied health outcomes but did not specifically include evaluation of health consequences were excluded. For example, studies on the relationship between urban trees and air quality that measured or modeled air pollution levels but did not include measures of human response or explicitly call out the health implications of the pollution levels, were excluded. After all articles were independently screened, the lists were compared and disagreements about inclusion or exclusion were resolved through discussion and consensus by three authors. The qualitative screening reduced the article count to 243. As expected for scoping reviews, we conducted multiple structured searches for relevant literature [35
]. The screening revealed notable omission of articles from the environmental sciences. The database that supports the Green Cities: Good Health web site at the University of Washington was subsequently searched, adding another 91 articles. After again screening for duplicates and content relevance, the collection totaled 215 articles, all published prior to 1 March, 2018.
2.2. Quality Assessment
We then conducted a science quality assessment using a modified version of the Effective Public Health Practice Project Quality Assessment Tool [38
], a public health research assessment tool for quantitative studies. Collected articles originated from multiple disciplines, including public health, social sciences, neuroscience, and environmental sciences, and encompassed many types of study designs, ranging from experimental to cross-sectional studies. We therefore modified the tool with the aim of broadening its applicability and focused on five components: (a) study design, (b) sample selection, (c) confounds, (d) data collection method, and (e) analyses. Each paper received a score of either: 1 (strong), 2 (moderate), or 3 (weak) for each component, and a global rating: strong (no weak ratings), moderate (one weak rating), or weak (two or more weak ratings).
The quality assessment was conducted independently by two authors and disagreements about scores were resolved through discussion and consensus. Inter-rater reliability for the global ratings was assessed using Cohen’s Kappa and was found to be satisfactory (κ = 0.929, p < 0.0001). Only studies that received a strong global rating (i.e., no weak component ratings) were retained. Sixteen articles were excluded, leaving 199 articles—for a total of 201 studies (as two papers reported multiple studies)—for subsequent analysis.
2.3. Thematic Analysis
To synthesize the resulting diverse literature into a coherent narrative, we adopted the conceptual framework published by Markevych et al. [39
], after considering multiple existing nature and health frameworks. The intentionally comprehensive framework by Markevych et al. was developed during a transdisciplinary expert workshop, with a particular focus on potential underlying biopsychosocial pathways linking green space to health. The domains reflect, but also expand on, prior empirical foundations. Each of the three domains imply ways to modify environments to support adaptation and promote health. The first, Reducing Harm, considers the role of vegetation in mitigating the conditions that can compromise health, and includes concerns of exposure to air pollution, noise, and heat. The second, Restoring Capacities, describes how nature experiences are a resource that promotes improved psychological and physiological functioning, including cognitive attention restoration, and stress recovery. The last domain, Building Capacities, describes nature experience pathways that facilitate multiple conditions of wellness for both individuals and communities, such as encouraging physical activity and providing settings for social cohesion. Sorting of articles into these domains was conducted by one author and confirmed by two others.
3. Results—Study Attributes
3.1. Study Locations
Of the 201 studies, 39% were based in North America, with 67 studies undertaken in the United States, 9 in Canada, and 1 in Mexico. About one third were conducted in Asia, including Japan (17% of the entire collection), South Korea (7%), China (4%), and Taiwan (3%). Another 25 %of studies were based in Europe including Spain, the United Kingdom, Sweden, Germany, Portugal, and France. The remaining studies were based in Australia (3%), and South America (1%). No study was based in Africa, and one study did not specify a location.
3.2. Study Participants
The full range of the human life span was represented, as 13% of studies focused on young adults and 13% on children. Adults were the primary age group studied (71% of studies), with 3% focusing on older adults. Studies of all-male participants (8%) were more frequent than all-female (3%). Forest bathing experiments were prominent among studies based in Asia, and commonly included young adult-male samples. Forty percent of studies involved participants with pre-existing medical conditions. Controlling for socio-economic factors was common among cross-sectional studies. However, few studies included a detailed analysis of how urban trees impact specific sub-populations (e.g., race/ethnicity, socio-economic status, or other factors of health risk).
3.3. Outcomes Measures
Various human response measures were used in the studies (with some designs using more than one): (a) psychological tests, including questionnaires and cognitive assessments of stress, job satisfaction, and mental acuity (29%); (b) physiological measures such as heart rate, cortisol, adrenalin, and glucose levels (27%); (c) self-reported symptoms of illness and allergies (14%); (d) modelling of human health impacts related to heat and air quality (each 5%) using primary and secondary data; (e) actual air quality using primary and secondary data (45%); (f) hospitalization and medical records (7%); (g) medication usage based on drug prescription/sales data (3%); and (h) neurological measures such as functional Magnetic Resonance Imaging (fMRI) scans (3%). Additional measures, each representing less than 2% of the collection were: (i) syndromic surveillance records such as ambulance calls and emergency room visits; (j) therapy effectiveness, such as Cognitive Behavioral Therapy; and (k) measures of participant-experienced ultraviolet radiation.
3.4. Tree Exposures
The 201 studies characterized exposure to urban trees and forests in various ways, with some studies using multiple independent variables. Tree exposure was measured both objectively (e.g., using satellite imagery, other remote sensing, and geographic information systems), and subjectively (e.g., surveys). Tree presence was also characterized based on quantity (e.g., density, proximity), as well as quality (e.g., well/not maintained). Specific tree measures included: (a) experience or a visit within a forest or woodland (e.g., walking, climbing, social activity, therapy) (27%); (b) canopy cover (16%); (c) individual/clusters of trees (e.g., street trees, schoolyard trees) (14%); (d) associated measures such as pollen, moss, and tree loss to emerald ash borer (EAB) (18%); (e) viewing images/simulations of trees/landscapes (8%); (f) forest/woodland/land cover using satellite sensors (such as NDVI), LiDAR and other remote sensing technologies (7%); (g) experiencing trees in a park, open space or natural area (4%); and (h) view of trees/forest through a window (1%).
3.5. Study Designs
A variety of study designs were observed (Table 2
). Experimental studies (28% of the study collection) used both objective and self-report measures, and involved forest visits, tree views, or viewing simulations (e.g., videos or images of trees) as response stimulus or interventions. Sample sizes ranged from 8 to 625 participants. Within subjects, crossover studies of forest versus built environments were the most commonly used design among experimental studies. Natural experiments (13%) used situational exposures and changes, such as loss of tree canopy from the EAB, to analyze changes in health outcomes. We classified observational studies into two groups: (a) longitudinal cohort or case-control (6%), and (b) cross-sectional studies (34%). Potential confounders addressed in such studies included data about prior conditions (such as asthma hospitalizations or chronic disease morbidity), demographic data derived from local or national census programs (such as income or poverty level, ethnicity, and population density) and physical environment attributes (such as renter occupied housing or land use type). The remaining studies utilized modelling (12%), or time series designs (7%). Modelling studies were limited to those that explicitly present environmental data or monitoring in association with health outcomes, often using regulatory or public health benchmarks, such as air particle concentrations or thermal comfort guidelines.
This scoping review reveals a large and rapidly growing body of research on urban trees and human health that is characterized by diverse study designs, pathways to health, and health outcomes. Organized under three overarching domains and subsequent research themes (or subdomains), our review provides readers with the opportunity to rapidly access research that is most relevant to their purposes (e.g., active living, weight status, or mental health). It also provides a transdisciplinary foundation for future research to help build more consistent and defensible knowledge that can inform approaches to improving community health, urban greening interventions, nature-based therapy, and clinical treatment.
As a high-level synthesis of the extent and diversity of this body of literature, Figure 2
presents an illustrative summary of the multiple relationships between urban trees and health, as well as the growth in range and volume of urban tree and health research over the past few decades (Supplementary Material Table S1
: Citations and References for Table 2
). The flows passing through each column illustrates the connections between tree settings (first column: Tree Setting), the biopsychosocial pathways identified by Markevych et al. (second column: Domain) [39
], the subdomains that we interpreted based on the study results (third column: Research Theme), and the publication period by decade (fourth column: Year).
The studies examined exposure to multiple types of tree settings or influences (first column), including: individual trees, trees as pollen sources, trees in a park, canopy or land cover, forest immersion, and images or simulations of trees. ‘Other’ refers to VOCs, moss on trees, and tree loss due to EAB infestation. Across all three domains (second column), studies have incorporated the full range of tree settings. That being said, some types of tree settings appear more frequently in certain subdomains than in others. For example, forest immersion is associated predominantly with the Restoring Capacities domain, where a majority of the experimental studies involved visits to forested areas. Meanwhile, canopy/land cover—an approach to tree measurement that is often employed in epidemiological studies at larger geographical scales—was used more often in studies that fall under the Reducing Harm and Building Capacities domains.
Most of the 201 studies are classified within the Reducing Harm domain (82 studies), followed by Restoring Capacities (63), and Building Capacities (56), each containing multiple subdomains (third column). It should be noted that the volume of studies that represent any particular domain or subdomain does not equate to current strength of evidence. As shown in Table 2
, study findings can vary in terms of positive, negative, or mixed and/or statistically insignificant outcomes. For themes with a greater number of studies (e.g., Active Living), there might be a greater mix of positive, negative, and mixed findings, which may be partly due to the diversity of research questions and methods used by researchers. We have observed studies focusing on different aspects of active living (e.g., walking, cycling, and levels of play among children), for example, or using different measures of physical activity (e.g., self-reported or measured data) and tree exposure (e.g., perceived or measured). This diversity has contributed to many unique insights, but also makes it challenging to aggregate overall findings into conclusive statements about strength of evidence.
Finally, Figure 2
also depicts the growth in research from the 1980s to the present time (fourth column). Three studies were published between 1980 and 1989, and five studies were published between 1990 and 1999. Almost all studies, or 96%, were published since 2000, with many published in 2010 or later (71% of 201 studies). Interest in some research topics has expanded faster than others over the same period. For example, research on stress reduction expanded from 7 studies published between 2000 and 2009 to 19 studies in 2010 to 2018, while clinical outcomes studies only increased from 3 to 5 during the same period.
5.1. Review Limitations
The purpose of this scoping review is to convey the full range of relationships between urban trees and human health based on existing quantitative research. It may not capture all current research as we limited the review to quantitative studies and English language publications. Future reviews that include qualitative studies and those in other languages could potentially expand understanding of health outcomes and generate further insights about causal relationships.
As in every review, it is possible that additional articles could have been retrieved by extension of search terms. Nevertheless, our search terms had a broad scope and it is unlikely the outcome of our search would have changed substantially by adding any additional terms.
We conducted a quality assessment on each article and included studies that have diverse yet robust methodological approaches in this scoping review. However, we did not assess the strength of evidence for each research theme as this is beyond the purpose of a typical scoping review, and more appropriate for systematic reviews with a smaller body of literature focused on more directly aligned research questions [33
5.2. Research Needs
We found that the research on urban trees and health has evolved considerably from the 1980s to the present day, building on the momentum generated by promising findings from earlier studies (e.g., [112
]), improvements in technology and data quality, and growing public interest and funding opportunities. This body of research covers multiple disciplines including public health, environmental health, urban planning, and urban forestry. Within this diversity, researchers have employed a diverse range of research questions, hypotheses, and methods that make it challenging to compare results. The population and geographic scales of studies were also diverse and not readily categorized.
More consistent research design and methods across studies, and replication, would be beneficial to enable cross-comparisons, including meta-analyses to generate more robust and conclusive evidence about phenomena and causality [222
]. The development of shared research protocols or frameworks could also help build consistency across different types of research studies. For example, standardizing definitions in three key areas: (a) urban trees and forests, (b) health pathways and outcomes, and (c) site-specific characteristics, would facilitate better comparison across studies.
While studies were conducted in cities around the world, they were largely focused in more industrialized, temperate climate regions, with a majority located in North America. The generalizability of research findings could be enhanced by repeating investigations across multiple climate zones, seasons, and settings, all of which can have variable influences on the impact of trees on people’s health—a recommendation that echoes Zeng and Dong who called for “complementary studies with longer periods, larger-scale surveys and during other seasons” [223
] (p. 107).
Additionally, the experimental and quasi-experimental studies collected for this review had small sample sizes. Out of the 57 experimental studies, over two-thirds of them had sample sizes of less than 50 participants, and most recruited only all-male subjects and young adults. Future urban tree and health research can be strengthened through larger sample sizes as well as better representation of residents in a particular location, with greater diversity across different ages, gender, and socio-economic conditions. Additional studies concerning other cultural or ethnographic settings or groups would confirm the salience of findings for more varied community or national populations.
We observed that studies typically investigated short-term benefits such as improved blood pressure, stress reduction, and cognitive restoration. However, the durability and time-to-reverse of these benefits are less clear and would require longitudinal, time series, and other methods to assess sustained health outcomes. Future urban tree and health studies could also investigate how the duration and intermittence of people’s exposure patterns to trees can impact health outcomes. Other studies could explore what the consequences of short-term alleviation of symptoms following tree exposure might mean for patients with chronic diseases (such as improved blood pressure and cardiovascular function), and whether these short-term benefits could lead to reduced risk of developing chronic diseases later in life.
Given the multiple potential mediating influences on people’s health, future research could investigate how aspects of peoples’ nature experiences, such as the health of trees, can influence other social determinants of health. For example, Jones found a negative, persistent relationship between EAB-infected trees and how people allocate leisure time in nature, which has consequences for physical activity [194
]. Improving the understanding of the long-term effects of trees on people’s health would help strengthen the calls for investment in planting and nurturing trees.
Finally, future research could help improve understanding about the importance of site-specific characteristics on influencing health benefits. For example, Gilchrist et al. found that a different vegetation structure (e.g., the smooth ground layer, shrub layer, and canopy layer) can variably influence outcomes and suggested that the relationship of buildings to the landscape be considered at the outset of planning and design [130
]. More detailed measures and reporting of study settings or exposure conditions, in addition to beneficiaries’ traits and temporal-spatial conditions, would help to better understand nature ‘dosage’ and response [156
]. Standardized collection and reporting of functional form—such as the tree species, age, height, canopy density, physical location in relationship to nearby buildings, upkeep of the neighborhood, and density and size of tree stands—would also help inform urban forest planning and management to promote positive human health effects.
While extensive discussions are underway about the implications of climate change on ecological systems and human health (e.g., [224
]), we rarely encountered references to climate change in the 201 studies. Additional cross-disciplinary research is needed to better understand the risks of climate change on both human health and urban forests, including an exploration of how rapid environmental changes will affect tree health, and the consequent cascading impacts on human health, such as psychophysiological stress and heat-related illnesses and deaths. Such risks also highlight the importance of proactive urban forest management practices to secure and expand the ecological and health benefits of trees in the context of climate change mitigation and adaptation [228
Our findings support the growing public recognition of urban trees as an essential component of health-supportive environments (e.g., [229
]). It is important, however, not to overstate the current evidence. For instance, the effects of trees vary by person and may not always be beneficial, such as the potential for tree pollen to exacerbate allergy conditions. Additionally, the benefits of trees are affected by the health status of trees and forests. For instance, Donovan et al. found that EAB-infested trees have been associated with adverse health outcomes [213
]. Salmond et al. found that VOCs are emitted when trees are under stress (e.g., due to drought, heat, and pests) [32
]. There is variable response to tree shading within streetscape conditions [231
] and across winter and summer seasons [232
]. These site-specific considerations suggest that integrated and proactive design and management of urban trees could double as an experimental approach to human health research and minimize potential adverse impacts.
One observation of note is the prevalence of articles emerging from the PubMed search. Reactions to pollen, such as allergenicity and asthma, were prominent. Fewer articles highlighting salutogenic benefits of urban trees appeared from the PubMed search, perhaps leading to less awareness among public health and medical professionals about the diversity of tree-based health determinants.
Implementing trees as a health intervention in a community is a long-term, even multi-decade, investment. Urban forestry and health professionals could work together to better integrate human health outcomes into urban forestry best practices more directly by actively translating the full scope of science into practice [233
]. This could involve increased collaboration between health and environmental professionals in developing evidence-based resources such as tree planting guidelines that support positive human health outcomes, while considering site-specific characteristics and a range of population needs (e.g., to support active living across all ages). Greater collaboration between health and environmental professionals in the design process could also achieve the goals of co-designing for co-benefits. For example, trees that are planted with the primary goal of improving stormwater management could also be configured to optimize a range of additional positive health outcomes such as stress reduction and social cohesion [234
Overall, we have found that exposure to trees is associated with multiple health benefits. Underlying this relationship is the importance of access. Studies have found that there are often disparities in distribution of trees in urban areas with greater tree density being found in neighborhoods having higher household incomes (e.g., [235
]), which may in turn exacerbate existing socio-demographic health inequities. For example, people who may not have sufficient resources to operate air conditioning in their homes may also live in neighborhoods that lack the cooling benefits of urban trees, thereby compounding their vulnerability to extreme heat events [236
]. Adopting a health equity lens in the planning and management of urban forests can ensure a more equitable distribution of trees across towns and cities and provide residents with access to the health benefits of trees.
Identifying who is vulnerable to different health outcomes and where they live and work can also inform the development of more targeted strategies or interventions (e.g., urban greening and nature-based therapy) to maximize health benefits. When community members are involved in the development of these urban greening programs, additional benefits can be gained, such as increased civic engagement and social interaction [237
]. Most towns and cities face many competing funding priorities. Our review suggests that urban trees may be a low-cost policy intervention that addresses multiple environmental and human health co-benefits. Investing in the proactive planning and management of urban trees can pay human well-being and economic dividends.