Nature-based interventions for vulnerable youth: a scoping review

ABSTRACT Nature-based interventions hold promise for vulnerable youth experiencing mental, emotional, developmental, behavioral, or social difficulties. This scoping review examined wilderness therapy, animal assisted therapy, care farming, and gardening and horticultural therapy programs to raise awareness and guide future development of research and treatment options. Studies included in this review were identified through a systematic search of the literature informed by a scoping review framework. Studies were examined by design, sample, intervention, and key findings. The majority of studies were quantitative using repeated measures designs and were conducted primarily in the United States. Sample sizes were generally small. Interventions were residential and community based with varying degrees of duration. Outcomes were largely positive across a wide range of psychosocial and behavioral measures and often maintained post-treatment. We emphasize the importance of robust empirical designs, comprehensive description of the interventions and surrounding therapies, and identification of target groups.


Introduction
Contact with nature and the utilization of natural environments have been recognized to contribute to improving and promoting human health and well-being across the life span (Finlay et al. 2015;Cox et al. 2017;Frumkin et al. 2017;Aerts et al. 2018;Vanaken and Danckaerts 2018;Engemann et al. 2019;Mygind et al. 2019;Richardson et al. 2021). Researchers have investigated the relationships between natural elements, green spaces, landscapes, and outdoor activities and identified a diverse range of positive outcomes that include benefits for physical health, mental health, cognitive development, and social interactions (e.g. Abraham et al. 2010;Annerstedt and Währborg 2011;Bratman et al. 2012;Cox et al. 2017;Vanaken and Danckaerts 2018;Richardson et al. 2021). The increasing number of meta-analyses, scoping reviews, systematic reviews, and other research summaries in recent years has helped to synthesize the body of literature across outcome variables or specific populations as well as identify areas for future research (e.g. Frumkin et al. 2017;Vanaken and Danckaerts 2018;Mygind et al. 2019;Shanahan et al. 2019;Richardson et al. 2021). As the evidence base continues to grow, so has the need to identify consistencies in research findings, methodological approaches, and gaps in knowledge (Frumkin et al. 2017;Aerts et al. 2018;Vanaken and Danckaerts 2018;Richardson et al. 2021).
In this paper, we focus on nature-based interventions, sometimes collectively denoted as green care activities, for vulnerable adolescents and young adults experiencing mental, emotional, developmental, behavioral, or social difficulties. Nature-based interventions and green care activities consist of a diverse range of programs and services that utilize plants, animals, and/or landscapes to create therapeutic and treatment interventions designed to address health and care needs of general populations as well as specific population groups (Sempik and Bragg 2013). A broad range of organizations have developed and employed a variety of nature-based services and therapeutic interventions to promote health and well-being in specific populations. Similarly, the diversity of target groups and the variety of behavioral, clinical and non-clinical problems is extensive. While many of the studies deal with children and young people with a mental health diagnosis, there are studies that target educational and social benefits. The latter are excluded from the present research, which focuses on therapeutic benefits rather than skill building exercises.
Increasingly, concerns have been raised about high levels of adolescents and young adults struggling with mental health problems, adjustment difficulties, problem behavior, disengagement, or disaffection (e.g. Olfson et al. 2015;Coley et al. 2018;Keyes et al. 2019;Murray et al. 2019; Mojtabai and Olfson 2020;Cybulski et al. 2021). The severity of needs, gaps in existing services, and service provision challenges have generated interest in effective treatment approaches (McGorry et al. 2013;Vyas et al. 2015;De Vries and Wolbink 2018), which include nature-based interventions. The interventions are diverse not only in their targeted groups but vary also greatly in their design and setting (Annerstedt and Währborg 2011;Moeller et al. 2018;Shanahan et al. 2019). Based on the literature, the main types of nature-based interventions can be categorized broadly into wilderness therapy (Wilson and Lipsey 2000), animal-assisted therapy (Lentini and Knox 2015;May et al. 2016;Hoagwood et al. 2017), care or social farming (Murray et al. 2019), and gardening and horticultural interventions (Sempik et al. 2014;Park et al. 2016).
Although these programs are receiving attention in the academic literature, only a small number of evidence reviews have focused on nature-based interventions solely targeting vulnerable or atrisk youth populations (e.g. Wilson and Lipsey 2000;Lentini and Knox 2015). More often, youthfocused interventions are addressed as a specific subgroup in evidence reviews on nature-based interventions across age groups (e.g. Murray et al. 2019;Shanahan et al. 2019) or based on specific conditions, such as attention deficit hyperactivity disorder (Faber Taylor et al. 2001;Faber Taylor and Kuo 2009). Furthermore, the studies can be found in a variety of disciplines, thereby limiting awareness by the larger body of literature to the community of scholars and practitioners (Harper 2017). The growing concern for adolescent and young adult populations point to the need to examine nature-based interventions and services across intervention types. Mapping the existing literature helps guide further development and application of interventions for this population and can point to areas of future research needs.

The present study
The current evidence review fills a void by summarizing the literature on nature-based interventions for a youth population that is considered vulnerable or at-risk of significant mental, emotional, developmental, behavioral, or social difficulties, or placed in out-of-home institutional care. Our specific objective is to present a summary of methodological designs, samples, interventions characteristics, and key findings of the studies on wilderness-therapy, animal-assisted therapy, care farming, and horticultural-based interventions for adolescents and young adults. We permit both quantitative and qualitative research methodologies. While the strength of quantitative methods is the empirical support for theoretical advances, qualitative studies situate the interventions in their environment which can provide a rich contextual understanding of their impacts (Deighton et al. 2010). By examining studies across different types of interventions, our study summarizes the current knowledge on interventions and points to ways in which these interventions can be beneficial for vulnerable youth, and provides direction to researchers and practitioners who seek to advance the evidence base on this topic.

Materials and methods
Our study was informed by a scoping review framework identified as the most suitable approach to determine the extent and nature of the broad and diverse literature on the topic and to identify key areas for future research and scholarly engagement by identifying gaps in the literature (Arksey and O'Malley 2005;Levac et al. 2010).

Study eligibility criteria
Our review focused on studies targeting children, adolescents, and young adults between the ages of 10 and 24 years and who could be considered vulnerable and at-risk for poor psychosocial outcomes due to mental health diagnoses, inpatient hospitalizations, involvement in child welfare and juvenile justice systems, substance abuse, family issues, trauma, school and community difficulties, and other emotional, developmental, or behavioral concerns (Sanders and Munford 2014). We considered studies investigating a wide range of interventions aiming to utilize nature to improve and promote quality of life, human health, and wellbeing of eligible populations. With no restrictions on modes of delivery, intervention type or durations, the review included the breath and diversity of nature-based interventions. The review considered studies presenting primary data (original research) regardless of methodology, study design, publication date, or geography. Evidencesyntheses (literature reviews, meta-analyses, systematic reviews, etc.) were also eligible. If a primary data study was also included in an evidence-syntheses study, we indicated the overlap in the table that lists the evidence-synthesis studies (Table 5). Only peer-reviewed studies published in scholarly journals in English were considered. Studies without outcomes measures related to broadly defined quality of life, human health, and wellbeing categories were excluded from this review. This included studies focused primarily on educational or vocational activities, including academic instruction and learning. Likewise, studies that focused on everyday life activities or casual encounters with nature (e.g. private gardening, animal-assisted recreation, or outdoor recreational activities, such as walking, running, or cycling in nature) were excluded.

Literature search process
A comprehensive search strategy was developed and searches for relevant literature in both generic and subject-specific bibliographic databases were conducted, including Web of Science Core Collection, Scopus, CAB Abstracts and Global Health (via Web of Science), CINAHL Plus (via EBSCOhost), ERIC (via EBSCOhost), PsycINFO (via EBSCOhost), and SocINDEX (via EBSCOhost). We searched MEDLINE via the Scopus interface. Search terms were derived from initial scoping searches and reviewing the literature. The search strategy intentionally considered a broad range of nature-based interventions and was piloted to check the appropriateness of selected databases and search terms. The search string used in all databases was as follows: ("green care" OR "care farm*" OR "social farm*" OR "community farm*" OR "therap* farm*" OR "prison farm*" OR "social horticult*" OR "sensory garden*" OR "healing garden*" OR "rehabilitation garden*" OR "school garden*" OR "community garden*" OR "horticultural therap*" OR "therap* horticult*" OR "garden therap*" OR "therap* garden*" OR "forest* therap*" OR ecotherap* OR "eco therap*" OR "nature therap*" OR "nature assisted therap*" OR "nature intervention*" OR "animal assisted therap*" OR "animal assisted activit*" OR "wilderness therap*" OR "nature rehabilitation" OR "nature based rehabilitation") AND (child* OR adolesc* OR youth* OR young* OR juvenile* OR teen*).
Database searches were conducted on 17 December 2018. The search was repeated on 8 July 2020 to identify studies published since the original search date. Search results were de-duplicated to remove redundant citations identified from multiple sources. All records were reviewed at the title and abstract level for studies that potentially met the a priori inclusion criteria. In the next step, fulltexts of all records were obtained and assessed by two independent reviewers for eligibility. Potential discrepancies in screening were resolved by consensus. Following full-article screening, standardized descriptive data such as bibliographic information and descriptive study metadata, including sample populations, study type, intervention characteristics, outcome measures, and narratively summarized key findings were documented for each study. The final analysis consisted of tabulating and grouping the studies by intervention types. Because of the wide range of interventions and the variety of methods and study designs considered, study validity assessment (critical appraisal) was outside the scope of the review. Figure 1 summarizes the study selection process and indicates the number of articles excluded at each phase of screening.

Studies on wilderness-therapy interventions
Wilderness therapy programs, also referred to as outdoor behavioral healthcare, utilize the natural environment, generally in remote areas, as the setting for an intensive therapeutic process under the direction of skilled leaders. These programs consist primarily of residential programs ranging in length from a few weeks to several months or longer but also can include shorter intensive periods of three to 10 days that are incorporated into in-patient hospital, outpatient mental health, traditional residential treatment, or specialized treatment programs. Components include learning outdoor skills, hiking extensive distances, overnight camping, being away in remote areas, and individual and group therapy sessions. There may also be a family component, in which parents participate in therapy while their child is in care separately or as part of the wilderness therapy program. Additionally, parents may participate in outdoor activities with their child for a short period of time. Youth are placed in small groups with 6-8 other youth, with group process, trust, and cooperation essential components. Programs serve youth with severe mental health, behavioral, and social difficulties, including substance dependence or abuse. Prior to placement, youth typically have participated in other types of treatment modalities such as outpatient or inpatient mental health services which were unsuccessful in remedying the underlying issues (Russell 2001(Russell , 2003Bettmann et al. 2016;Fernee et al. 2017).
A total of 35 studies of wilderness therapy were identified which included 31 original studies and four evidence-synthesis reviews. Of the 31 original studies, 18 used quantitative methods, one used mixed methods, and 12 used qualitative methods (Tables 1 and Table 5). Samples ranged from 14 to 816 in the original quantitative studies and from four to 148 in the original qualitative studies. The sample of the original mixed methods study was 32 adolescents. In addition to the original intervention studies, four evidence-synthesis studies met inclusion criteria and consisted of two meta-analyses (Bettmann et al. 2016;Gillis et al. 2016), one scoping review (Harper 2017), and one qualitative review (Fernee et al. 2017). The two meta-analyses included 2,399 and 2,667 participants. The review of qualitative studies examined a total of 102 adolescents.
Ages of youth in the treatment programs ranged from 12 to 34 years of age. In the 31 original studies, data were obtained from youth in treatment programs (N = 29), parents or caregivers (N = 9), and/or staff members (N = 3) ( Table 1). Most of the studies examined youth in programs that served males and females, but five studies examined programs exclusively for males (Lambie et al. 2000;Russell 2000;Gillespie and Allen-Craig 2009;Somervell and Lambie 2009;Margalit and Ben-Ari 2014) and two studies examined programs exclusively for females (Caulkins et al. 2006;Pryor et al. 2006). Almost all programs indicated they served youth with severe mental health, behavioral, and social difficulties, including substance abuse and dependence or focused on specific groups such as youth who have committed sex offenses or delinquent acts, or experienced trauma.
The majority of the 31 original intervention studies examined stand-alone wilderness therapy programs (N = 18) that lasted between three and 22 weeks (Table 1). Other studies examined hiking, camping, or other wilderness components that were integrated into inpatient hospital, specialized mental health care, or drug treatment programs (N = 5, Berman and Anton 1988;Pryor et al. 2006 Lambie et al. 2000;Somervell and Lambie 2009;Conlon et al. 2018). These experiences lasted a total of four to 20 days over an extended period of time.
The 18 original quantitative (Table 1) and one original mixed methods studies (Gabrielsen et al. 2019a) used well-established measures and repeated measures designs to assess psychological, social, and behavioral functioning from pre to post treatment, as well as family functioning (N = 5, Bandoroff and Scherer 1994;Russell 2000Russell , 2005Harper et al. 2007;Johnson et al. 2020), attachment to parents (N = 2, Bettmann and Tucker 2011;Bettmann et al. 2017) and body mass index and weight (N = 1, Tucker et al. 2016). Measures such as the Youth Outcome Questionnaire or Youth Outcome Questionnaire-Self Report (Burlingame et al. 2004) have indicators of clinical significance that allow for differentiating clinical and non-clinical samples. Thus, clinical as well as statistical significance was often reported.
Overall, positive change was indicated across studies on a broad range of measures from pre to post treatment that included symptom distress, self-esteem, self-efficacy, locus of control, problem behaviors, substance use, social interaction, school attendance, recidivism, and other psychosocial and well-being indicators, often with medium to large effect sizes (Table 1). Additionally, positive change was maintained upon completion of treatment to later follow up periods ranging from six weeks to 18 months later (N = 12, Bandoroff and Scherer 1994;Russell 2003;Clark et al. 2004;Harper et al. 2007;Bettmann et al. 2013;Hoag et al. 2013;Margalit and Ben-Ari 2014;Paquette and Vitaro 2014;Combs et al. 2016;Roberts et al. 2017;Gabrielsen et al. 2019a;Johnson et al. 2020). Change was often noted by t-tests with more recent use of regression analysis and hierarchical linear modeling (Paquette and Vitaro 2014;Combs et al. 2016;Bettmann et al. 2017;Roberts et al. 2017;Johnson et al. 2020). Some studies included analyses by gender, age, or diagnosis with mixed program impact findings based on subgroup (Berman and Anton 1988;Russell 2003;Harper et al. 2007;Bettmann and Tucker 2011;Combs et al. 2016). Findings were also mixed on measures of family functioning or attachment to parents (Harper et al. 2007;Bettmann and Tucker 2011).
The 12 qualitative studies (Table 1) all utilized interviews with participants with the exception of one study that analyzed open-ended survey responses (Harper et al. 2019) and one study that interviewed parents only (Liermann and Norton 2016). Additionally, five studies used participant observation, focus groups, and document analysis along with interviews (Lambie et al. 2000;Russell 2000;Russell and Phillips-Miller 2002;Caulkins et al. 2006;Somervell and Lambie 2009). These methods were used to gain greater insights into the treatment process, including beneficial components, allowing youth a greater voice in outcomes assessment, and gaining multiple perspectives from youth, parents, and staff. Four studies included interviews with parents (Lambie et al. 2000;Russell 2000Russell , 2005Liermann and Norton 2016) and three studies included interviews with staff (Russell 2000;Caulkins et al. 2006;McIver et al. 2018).
Most interviews were done immediately at post-treatment with five studies conducting interviews at three to 24 months post-treatment (Russell 2000(Russell , 2005Pryor et al. 2006;Liermann and Norton 2016;Fernee et al. 2020). The studies examined impacts from participation in the wilderness therapy program as a whole, although one study specifically examined the family therapy component (Liermann and Norton 2016) and one study examined the backpacking component (Caulkins et al. 2006).
Confirming the quantitative studies, the qualitative research noted positive impacts on a range of psychological, social, and behavioral outcomes, as well as relationships with parents and peers. Most gains were maintained post treatment. Going beyond the insights gained from quantitative studies, two qualitative studies, and the qualitative component of the mixed methods study (Gabrielsen et al. 2019a) noted transition difficulties after treatment and youth needing more time to process the changes that occurred through treatment (Russell 2000(Russell , 2005. Also unique to qualitative research, most of these studies included discussion of key components of wilderness therapy that contributed to positive outcomes, such as the role of nature, peer and staff relationships, being in a new environment, physical rigors, and other challenges.

Studies on animal-assisted interventions
Animal-assisted interventions represent a broad category of structured interventions that include or incorporate animals in health, education and human services with the goal of reaching therapeutic gains in humans (IAHAIO 2018). Animal-assistant interventions are aimed at improving behavioral, social, emotional, cognitive or physical functioning and typically include defined goals and measured outcomes (Maber-Aleksandrowicz et al. 2016). Of the 37 studies on animal-assisted therapy, 29 are original intervention studies with 15 quantitative, five mixed-methods, and nine qualitative designs (Table 2). An additional eight studies present evidence-synthesis studies, including four systematic reviews (Maber-Aleksandrowicz et al. 2016;Hoagwood et al. 2017;Jones et al. 2019;White et al. 2020), two literature reviews (Lentini and Knox 2015;May et al. 2016), one meta-analysis (Wilkie et al. 2016) and one systematic map (McDaniel Peters and Wood 2017) ( Table 5).
The sample ranged from 29 to 138 participants in the 15 original quantitative studies (Table  2); the one meta-analysis examined 377 children and adolescents. The sample of the original mixed methods studies ranged from seven to 28 study participants and the sample size of the qualitative studies ranged from five to 80 children and adolescents. Ages of study participants ranged from two to 22 years of age. Most of the 37 studies on animal-assisted therapy examined male and female participants; four studies focused on males (Mallon 1994;Conniff et al. 2005;Williams and Metz 2014;Boshoff et al. 2015;Hemingway et al. 2015) and three studies focused on female children and adolescents (Conniff et al. 2005;Carlsson et al. 2015;Carlsson 2018;Naste et al. 2018). Overall, the majority of this literature examined youth with mental health difficulties, including several studies that focus on populations with specific conditions, such as autism, post-traumatic stress syndrome, and attention deficit hyperactivity disorder. Many of the 37 studies targeted specific samples, such as children and adolescents in foster care, adolescents living in residential care, incarcerated adolescents, and hospitalized children and adolescents.
Key findings of the 37 studies addressed psychological, social, and behavioral outcomes. Studies that examined psychological outcomes addressed these at different levels. The literature indicated the associations of animal-assisted interventions with global psychological measures, especially stronger psychological well-being, better internal regulation and cognitive functioning, higher emotional states, fewer emotional problems, and lower levels of post-traumatic stress disorders. With regard to individual differences measures, animal-assisted interventions were associated with several personality measures, including higher levels of self-image, self-control, trust, selfconfidence, self-esteem and self-efficacy, as well as measures of life satisfaction, calmness, and feeling of safety. Additionally, the studies documented the association of animal-assisted interventions with lower levels of mental health disorders, such as irritability, self-stigmatization, anxiety, and depression, as well as better management of emotions, especially anger and fear.
Social outcome measures that were found to be related to animal-assisted interventions included global measures of better overall social functioning and higher levels of social cognition. In addition, the research documented the association of animal-assisted interventions with interpersonal skills in social interaction, including empathy, respect, humor, and patience in treating others. Studies also reported higher engagement, bonding, attachment formation, helping others, and steps toward taking leadership. A number of studies pointed to communication-related outcomes, such as use of communication strategies and better coping with teasing and bullying.
Behavior-related outcome measures that were found to be related to animal-assisted interventions included global constructs of a lower number of negative, disruptive, and difficult behaviors, lesser behavioral dysregulation and hyperactivity. A positive association of animal-assisted interventions and attendance of school and treatment sessions was also reported.

Studies on care farming interventions
Care farming interventions, also described in the literature as social farming or green care farming interventions, are provided primarily through agricultural farms and landscapes. They utilize structured programs of farming-related activities to deliver health, social, and educational services to adolescents and young adults (Sempik and  Four studies in the final dataset investigate adolescent-focused care farming interventions based on our inclusion criteria. Three studies are original research located in Europe (Kogstad et al. 2014;Schreuder et al. 2014;Leck et al. 2015 , Table 3). Of those three studies, two studies employed a qualitative, semi-structured interview methodology to retrospectively analyze experiences and observations and semi-structured interviews (Kogstad et al. 2014;Schreuder et al. 2014). The third study used a mixed-method design included 216 baseline and 137 follow-up survey responses as well as 33 semi-structured interviews with a subsample of survey respondents (Leck et al. 2015). Adolescents struggling in mainstream education comprised the principal subpopulation (N = 30). In addition to the three original intervention studies, the fourth study in this sample was a systematic review, which assessed studies based on a logic model that examined the studies' settings, intervention components, mechanisms linked to theoretical concepts, proximal outcomes, and outcomes (Murray et al. 2019, Table 5). Four studies on disaffected adolescents were included in the systematic review, including the above three studies.
The two qualitative studies had nine and 11 study participants (Kogstad et al. 2014;Schreuder et al. 2014), the mixed methods study had 30 study participants (Leck et al. 2015), and the systematic review included studies with a total of 112 children and adolescents (Murray et al. 2019). The care farming interventions with residential programs varied in length from about three months to over a year, where this information is provided.
Key findings across the three original intervention studies point to three main components of successful care farming. The first component is the farm setting, the farmer, and the farming environment that is conducive to recovery. The second effective component is the variety of tasks and the environmental engagement that allow for a gradual increase in adolescents' self-efficacy. Experiences with farm animals provide comfort for adolescents lacking trust in people and life in nature to recover a positive sense of self. The role of routine in farming provides structure to the recovery of adolescents. Third, successful interventions provide opportunity for personal development and the development of social interactions to improve physical and mental health and wellbeing. The systematic review pointed specifically to the findings that care farming increased the quality of life and decreased depression and anxiety for disaffected youth (Murray et al. 2019). The three original studies and the single evidence-synthesis study indicate weak associations between mechanism and outcomes and point to the need for robust analysis of the pathways through which care farming interventions affect the proposed outcomes.

Studies on gardening and horticulture-based interventions
Gardening and horticulture-based interventions examine whether programs that teach gardening and horticulture skills, for example, how to grow and maintain a community garden, are associated with youth's mental, psychological, and behavioral health. We identified six studies, including five original intervention studies (Table 4) and one literature review ( Table 5). Three of the original studies used mixed-method approaches (Twill et al. 2011;Sonti et al. 2016;Chiumento et al. 2018) and two used qualitative designs (Allen et al. 2008;Delia and Krasny 2018). The sample sizes in the mixed-methods studies ranged from nine to 50 study participants. The samples in these studies included males and females, where reported. The ages ranged from six to 25 years, with four of the studies focusing on specific age groups, including age ranges of up to seven years. The single evidence-synthesis study was a literature review that summarized 133 studies, but did not list sample sizes of the individual studies; demographic information is provided in summary statistics (Park et al. 2016, Table 5). All interventions were based on community gardening, including specializations such as urban farming (N = 2, Allen et al. 2008;Delia and Krasny 2018), school gardening (N = 1, Chiumento et al. 2018), neighborhood beautification (N = 1, Allen et al. 2008), and gardening in juvenile rehabilitation (N = 1, Twill et al. 2011). The literature review study further identified interventions that included indoor planting, craft-making, and flower arrangements (Park et al. 2016). The interventions ranged from three to nine months, typically spanning the growing season; intervention hours were not reported.
The three mixed-methods studies provided mixed findings about the role of horticulture-based interventions for youth mental health (Twill et al. 2011;Sonti et al. 2016;Chiumento et al. 2018). For example, one set of findings did not support a relationship between the intervention and selfconcept, emotional, and behavioral management (Twill et al. 2011). Another set of findings pointed to positive associations with food, health, and environmental behaviors (short-term) and communication and decision-making skills (longer-term) (Sonti et al. 2016). One study did not provide statistical analysis (Chiumento et al. 2018). The two qualitative studies emphasized the association of horticulture-based interventions with social and leadership skills (Allen et al. 2008;Delia and Krasny 2018), thus providing additional perspective to the empirical findings. Finally, the literature review indicated that the relationship of intervention and outcome is dependent on the type of mental disorder of the targeted youth (Park et al. 2016).

Discussion
This study contributes a summary of the literature for the specific population of vulnerable youth experiencing mental, emotional, developmental, behavioral, or social difficulties across four types of nature-related interventions: wilderness-therapy interventions, animal-assisted interventions, carefarming interventions, and gardening and horticulture-based interventions. The literature includes quantitative, qualitative and mixed-methods study designs, with the majority being quantitative studies. Commonly used methodological approaches are repeated measures, control or comparison  Caulkins et al. (2006), Russell (2000), Russell (2005), and Russell and Phillips-Miller (2002 Leck et al. (2015), Kogstad et al. (2014), and Schreuder et al. (2014) groups, and post-treatment assessment. The majority of sample sizes is small with only 13 original studies including more than 100 study participants. Interventions vary in their focus, ranging from broadly designed, multi-faceted tasks in care farming and horticulture-based interventions to highly specific activities in animal-assisted interventions. The key findings reflect the nature of the interventions, in particular, the most robust findings can be associated with the more narrowly defined animal-assisted interventions. Outcomes were largely positive across a wide range of psychosocial and behavioral measures and often maintained post-treatment. The psychosocial and behavioral measures that the nature-based intervention contributed to the most include mental health measures, such as hyperactivity, anxiety, depression etc., social measures, such as relationship building, interpersonal skills, etc., and measures of personal development, such as self-concept, meaningfulness etc. This scoping review identifies several avenues for future research. A first avenue for future research that emerged from the analysis of the study designs is the need for a greater emphasis on the use of methodologically robust empirical designs, to achieve a higher degree of external validity. The majority of studies in this scoping review are based on pre-post evaluation-type repeated measures. The approach is often justified with the special nature of the intervention which makes a comparison group difficult to obtain, a challenge that is particularly evident in wilderness-therapy interventions (Russell 2003;Clark et al. 2004;Roberts et al. 2017). To some extent, the animalassisted interventions literature can serve as a role model here, with its larger number of statistically robust research designs that often include larger sample sizes and randomized approaches with control groups. Alternatively, newer statistical approaches, such as intent-to-treat designs (Gupta 2011), or staggered research designs, such the step-wedge designs (Hemming et al. 2015), deserve greater attention in this literature.
A second avenue for future research comes from the analysis of the interventions covered in this review and points to a continued need for evaluating youth-based interventions. In particular, knowing of the large number of interventions targeting adolescents and young adults with mental health difficulties, future research should examine how well the literature reflects the breadth of the field. This analysis could be beneficial for identifying additional research needs and initiate a discussion of best practices in areas that are better understood. Ideally, this research should be approached from an international perspective, to increase the understanding of effective interventions from a cultural lens (e.g. Levinger et al. 2021). A second, intervention-related observation is the often limited amount of information on the actual intervention in the research studies. On the one end are the highly-focused studies that list the exact number of hours of each intervention within a therapy plan, while on the other end of the spectrum are studies that provide a list of interventions but lack detailed information that would facilitate a better understanding of which components were particularly effective (e.g. Caulkins et al. 2006). A third, related observation is the challenges with establishing causality and demonstrating which part of an intervention has beneficial effects for which group of users. Although beyond the scope of this review study, our evidence synthesis suggests a greater need for critical engagement with the reported benefits that takes into account issues, such as self-selection bias or negative outcomes and challenges, that are typically not reported in original studies. For this reason, it is difficult to assess the factors for successful interventions and groups (e.g. with more severe clinical depression or disabilities) for whom naturebased therapy is not a viable option.
A third avenue for future research emerged from the analysis of the study participant samples in this scoping review. Future research should identify the subgroups more clearly for whom the treatment is most effective, i.e. by age group, gender, or diagnosis (Berman and Anton 1988;Gillis et al. 2016). In particular, more understanding is needed how well the treatment modalities work for the variety and severity of psychosocial, behavioral, clinical, and non-clinical issues experienced by youth in the short and long-term. For example, some of the wilderness therapy studies have examined target groups, but typically with mixed results and not enough consistency across the studies to make strong recommendations (Berman and Anton 1988;Harper et al. 2007;Bettmann and Tucker 2011;Combs et al. 2016). A related topic for future research is disparities in naturebased interventions due to a program's funding structure. Whether self-paid, government funded, or a combination of both, the financial foundation of an intervention can determine participant composition and quality of an intervention (Frankford 2007). Information on the funding structure and its implications would provide additional transparency in this literature and also will help eliminate biases in who participates in the different programs (Bettmann et al. 2016).
A fourth avenue for future research emerged from the analysis of key findings for this scoping review. Few studies acknowledge the fact that nature-based interventions are generally used after other treatment modalities have been tried or can be part of an integrative treatment model (e.g. Compitus 2019). Yet the literature does not control for the impacts of this prior treatment. Furthermore, few studies have included discussion of aftercare plans. Future research should examine the best placement of nature-based intervention within a holistic assessment of treatment models. In a related perspective, many nature-based interventions, especially care farming, have an integrative function which aims to reconnecting youth with the community. This aspect could be more widely investigated in the literature.
Several limitations should also be noted for this study. First, we excluded reports and other gray literature, to keep the focus on the peer-reviewed studies. Second, the context of most of the studies is often highly location specific, which limits their generalizability. For this reason, we describe rather than measure differences and similarities in the literature. Third, we excluded the miscellaneous categories of nature-based interventions from this review, such adventure therapy, green exercise activities, challenge programs, therapeutic camping and other outdoor experiential programs that may include youth participating for recreational rather than therapeutic reasons (Wilson and Lipsey 2000;Gillis et al. 2008). Difficult to categorize, these interventions should be examined in a separate study that focuses on this diverse array of approaches to working with youth. Fourth, our understanding is also limited by the publication of findings from only a small percentage of all possible programs that exist. Increased attention to nature-based interventions by scholars across disciplines can increase the potential for collaborative research efforts and help build capacity of residential and community-based programs to implement research studies, disseminate findings, and add to the evidence base (Cunningham et al. 2015;Thompson et al. 2017;Wainberg et al. 2017).