The systematic search conducted by Marselle et al. (2019) and previously by Lovell et al. (2014) was replicated for this study using the Web of Science, to capture more recent studies conducted following the Marselle et al. (2019) review. From this, it was determined that 28 recent studies met the inclusion criteria for full-text review. Data extraction methods used by Marselle et al. (2019) have been replicated and built upon. In doing this, together with determining the author disciplines as with earlier studies, changes in disciplinarity and use of variables in recent works could be compared.

Using the Web of Science, journal articles which had published original work between the years 2017–2021 were searched for. The search terms used and process of identifying papers appropriate for full-text review are summarized in Table 1 and Figure 2. Although the year 2017 was included in the review by Marselle et al. (2019), this year was included in the present study in order to capture any potential studies that were missed in that review. The search terms (Table 1) and inclusion and exclusion criteria were adapted from Marselle et al. (2019), with the aim of being able to build upon and draw comparisons with the evidence base they obtained in their review.

The following inclusion criteria were used during the systematic search process:

Studies were excluded if they did not directly assess (i) biodiversity and (ii) mental health or mental wellbeing related outcome measures. Studies assessing preference, or the amount of greenspace without specifically measuring its biodiversity were excluded. Studies not reporting primary research such as review papers, were also excluded. In contrast to Marselle et al. (2019), studies assessing physiological stress measures were included in this search. It is acknowledged that including studies written only in English may have reduced the full scope of research that has been conducted on this topic.

This study builds upon the work of Marselle et al. (2019). Using the 24 studies they reviewed and the additional 28 recent studies found, the authorship of each of these studies has been reviewed. A novel approach was adopted to determine author disciplines for this research. To do this, Google Scholar author searches were used to determine author affiliations, “key words” and the topics on which they had previously written. Author “key words” were determined to be reflective of the focus discipline of researchers, as the particular school or organization that they belonged to did not always reflect their specific research interests or expertise. The discipline of the author was defined as the research specialization as self-described by the author on their Google Scholar profile. Where this information was not available, the school and home institution to which the author belonged was recorded from the relevant journal article webpage. This enabled the disciplinary framing of each study to be determined (acknowledging that this is a novel and rudimentary approach). For each of the 52 studies included in this review, each author was recorded and included in the analysis.

Based on potential sources of inconsistency as described in previous works, key elements of study designs that may be contributing to inconsistent findings were extracted from each study:

As seen in Marselle et al. (2019), biodiversity variables were assigned to one of three levels: (i) ecosystem/habitat, which includes variables such as landscape heterogeneity and structural diversity, (ii) species communities, which includes variables such as species richness, perceived species richness and the abundance of a specific taxonomic group, and (iii) single species level. Mental health and wellbeing variables were recorded as described in each study. Mental wellbeing variables were separated out and included attention restoration, recovery from stress, emotion/mood, and quality of life/life satisfaction, while “mental health” as a variable was recorded individually. In this study, a correlational type of study refers to an observational study that explores relationships between two or more variables, and may have obtained data using cross-sectional, longitudinal, cohort or retrospective record methods MacDonald et al. (2015). A mechanistic study refers to a study that directly explores the mechanisms of action of a particular experimental intervention.

A standardized spreadsheet was used to record all data, and a process of narrative synthesis, as developed by Popay et al. (2006), was used to interpret and compare the data from the two time periods. This type of synthesis has been used in previous reviews, including Marselle et al. (2019), and is a form of vote-counting synthesis that is useful when dealing with variables and data that are heterogenous, and where other forms of quantitative or statistical analyses are not suitable. This involves extracting key themes from the literature in order to compare and highlight points of difference.

Similar results emerge for disciplinarity from the two time periods, with environmental psychology continuing to be the most prominent author discipline. There has been a slight increase in interdisciplinary work in recent years, shown by a greater number of studies including three disciplinary perspectives (Table 2). The majority of studies, however, continue to include collaboration from just two disciplines. This is not necessarily surprising considering the two fields that this research combines. Again, unsurprisingly, the most prominent collaboration occurs between the disciplines of environmental psychology, and ecology or biological sciences. These results do, however, show that there is quite a diversity of disciplines that have contributed to the research so far. In recent works there has also been a greater contribution from the perspectives of landscape architecture, horticulture, health sciences and environmental economics. This may have implications for subsequent study designs and the application of findings.

Attention Restoration Theory (ART) and Stress Recovery Theory (SRT) are the most prominent guiding theories for studies from both time periods (Figure 3). These theories are similarly based on the notion that natural elements within landscapes provide the opportunity for mental relief from the many things that demand attention in everyday life and enable recovery from stress or the use of active attention through passive engagement with the environment (Kaplan and Kaplan, 1989).

The use of these theories reflects the short-term nature of most of the studies, which examine what may be referred to as “momentary wellbeing” (Fisher et al., 2021). Another theory that has been used in both early and recent works is the Biophilia Hypothesis (Figure 3), which suggests that humans have an innate tendency to seek connections with nature and other forms of life (Kellert, 1995). In recent work, new theories have been used such as Cultural Ecosystem Services, which describes the non-material benefits offered by natural environments, and which has importance for environmental economics and decision making (Milcu et al., 2013).

The most surprising finding here is that the majority of studies in both earlier and recent years do not refer to a guiding theory at all (Figure 3). This is reflected in the type of correlational study design used in most of the research as opposed to experimental, mechanistic studies. This may also have flow on effects in subsequent aspects of study designs and interpretation of findings. There is no notable correlation between author disciplines and the use of guiding theoretical frameworks. There are many studies with the perspective of environmental psychology without being guided by ART or other theory (e.g., Wheeler et al., 2015; Schebella et al., 2019), and conversely, there are numerous studies without this psychological perspective but which utilize a guiding psychological theory (e.g., Hoyle et al., 2017; Meyer-Grandbastien et al., 2020).

There is a significant variation in the biodiversity and health and wellbeing variables used across the 52 studies reviewed (Figures 4, 5). So far, there are no two studies which used the same study design or variables. Furthermore, it is important to note that many studies utilize multiple variables of both biodiversity and mental health.

The relative frequency with which particular metrics of biodiversity have been utilized are summarized in Figure 4. Comparing the two sets of studies [Early (E) and Recent (R)], it can be seen that species richness, measuring biodiversity at a species communities’ level, is the most prominent metric used to evaluate biodiversity in the studies in both time periods, however, the variables and instruments used across studies are highly heterogeneous. The ecosystem/habitat level of biodiversity is also studied frequently across both time periods. In recent works, there is a greater representation of perceived species richness as a biodiversity variable. As reflected in the results for guiding theory (Figure 3), attention restoration continues to be the most prominent variable by which mental health is measured. This is followed by the use of the broader “mental health” variable, for both time periods. As with the biodiversity variables, many studies measured multiple different outcomes of mental health and/or wellbeing.

There is such a diversity of both mental health and biodiversity indicators utilized across studies that it remains difficult to compare or draw any correlation between discipline and use of variables. In recent work we do see a move to diversify the range of biodiversity variables tested, moving beyond just simply vegetation indices to include other taxa such as birds and insects (Wood et al., 2018; Cameron et al., 2020). We also see a large number of studies that include perceived biodiversity metrics, which have been shown to be of great importance to obtaining mental health and wellbeing benefits from nature (White et al., 2017; Nghiem et al., 2021).

Throughout the two time periods, correlational study designs continue to pervade, as reflected by the frequent use of ecosystem/habitat level of biodiversity, and broader “mental health” as a variable (Figures 4, 5). There have yet only been five experimental studies conducted within the recent body of literature, and only one experimental study conducted in early literature. Within recent literature, newly emerging experimental study designs include those with the planting of specific experimental plots (Lindemann-Matthies and Matthies, 2018; Southon et al., 2018), which have been conducted from horticultural and/or landscape architectural perspective. Other emerging methodologies include measuring physiological stress responses in a quasi-experimental environment in situ (Lindemann-Matthies and Matthies, 2018; Hussain et al., 2019) or in response to the presentation of photographs or short videos (Schebella et al., 2020).

It is clear that researchers are only just beginning to understand the nuances of relationships between biodiversity and mental health and psychological wellbeing, and the variation in methodologies and results reflects the growing interest in this area, and the many perspectives from which people are trying to understand and test these relationships. Although there is yet little consistency in terms of disciplinarity and methodologies, there are some key elements of study designs that can be informed by the unique perspective of particular disciplines. For example, the use of physiological measurements as informed by health sciences indicates a growing interest in understanding the underlying mechanisms at play, and the use of experimental planting plots from the perspective of landscape architecture or horticulture indicates an interest in using these findings to apply to landscape design. The importance of contribution from perspectives of psychology and ecology is clear for this type of research, however, the contribution of additional disciplines would be of great benefit for both understanding the underlying mechanisms of these relationships and applying them to decision-making or design.

To assist with bringing in a wider range of perspectives, we have created a visualization of how different disciplines might contribute to the body of research at various scales (Figure 6). At the top of this inverted triangle there are correlational studies, which so far have tended to work at large geographic scales, obtaining for example, census data for mental health variables and satellite derived biodiversity data. These operate at the larger ecosystem/habitat scales of biodiversity. As we move to the bottom of the triangle, there are the more experimental, mechanistic studies, which study biodiversity at the community or single species level, and which measure mental health at physiological or attention restoration level. These tend to be on much smaller geographic scales, in for example a specific greenspace or planting plot, with an accordingly small sample size. These are the findings that would be more specific than those at the top of the triangle, as indicated by the arrows on the right-hand side.

Beside the triangle on the left-hand side, some of the key research disciplines which may make the greatest impact at various scales are displayed. Environmental psychology and ecology are displayed across the whole length of the triangle as these will be of greatest importance at every level of this research. Collaboration between psychology and ecology could greatly benefit from additional perspectives, for different outcomes and applications of findings and to develop a more mechanistic understanding of these relationships. In addition to the importance of foundations in psychology and ecology, perspectives such as public health and environmental economics could assist in framing the research from the perspective of decision making. For example, it could be important from the perspective of environmental economics and decision making to consider the concept of Cultural Ecosystem Services (including psychological restoration, improved physiological health, better social relations and spiritual development) (Kosanic and Petzold, 2020). A challenge exists in ascertaining whether cultural ecosystem benefits are sensitive to variations in biodiversity (King et al., 2017). Cultural Ecosystem Services could be an important concept for guiding decision making based on findings from research in this topic and warrants further investigation, as so far only one study has framed their research and findings in the context of this concept. Toward the mechanistic end of the research, perspectives of landscape architecture, horticulture, medical and health sciences may contribute important elements for experimental study designs, such as the use of physiological measurements in response to biodiversity, or the establishment of specific planting plots in which the ecology is well-documented. From a medical perspective for example, demonstrating stronger mechanistic links between biodiversity and mental health and wellbeing variables could provide important evidence to support healthcare approaches that formalize the role of nature, through concepts such as “green prescription” (Ulmer et al., 2016).

It is important to note that the categories shown in Figure 6 are not clear-cut, and that important interactions would occur between various aspects of studies at different scales. However, overall, it will be more consistent work in these lower, more mechanistic areas of the inverted triangle, that will help to understand the underlying mechanisms and relationships on local scales that can account for variables such as sociodemographic factors and the importance of place-based data (Harvey et al., 2020). Mixed methods approaches that combine perspectives from multiple disciplines, with clear theoretical foundations and use of robust experimental study designs would enable a clearer progression of findings in this field. It is clear that a longer period of time, including the use of longitudinal research may be needed for these relationships to be understood, particularly in how biodiverse environments might contribute to mental health over time, for example beginning in childhood (Harvey et al., 2020). There is also a need to understand the full range of potential mediators and moderators of these relationships, such as biodiversity knowledge, connectedness to nature and socioeconomic factors, as these have been shown to influence the way benefits are obtained from biodiverse environments (Schebella et al., 2020). For example, psychological benefits have been reported as being greatest for those with lower socio-economic status, lower education level or low biodiversity knowledge in multiple studies (Hoyle et al., 2017; Coldwell and Evans, 2018; Marselle et al., 2020). Further investigation of the relative benefits of biodiversity for a greater diversity of populations and socioeconomic factors could be of great importance in the context of environmental equity. Similarly, it would be valuable to see a greater representation of non-western populations such as those represented in Fisher et al. (2021), Chiang et al. (2017), Chang et al. (2016), and Nghiem et al. (2021). We acknowledge, however, that the inclusion of English language only studies in our review may have precluded representation of such populations within the body of literature.

In order to progress toward transdisciplinary research within this field, there are several theoretical frameworks that future researchers might consider as a means of framing studies in a more holistic manner. The Actor Network Theory for example, provides a means by which researchers may frame relationships between biodiversity and health in the context of social-ecological system resilience (Horgan and Dimitrijević, 2018), which may be of great importance for applying findings to urban planning. Given the importance of perceived biodiversity as a metric conferring health and wellbeing benefits, we believe that tools such as the “Place Standard,” which invite community participation in assessing greenspace qualities, may provide important information for decision-makers in understanding how communities relate to biodiversity through concepts such as place attachment and sense of place (Hasler, 2018; Colley and Craig, 2019). These concepts, which describe cognitive-emotional connections between people and their valued places have been demonstrated to mediate the wellbeing benefits obtained from greenspace (Scannell and Gifford, 2017; Basu et al., 2020; Han et al., 2021) and could therefore be important for future research to consider. However, by drawing upon a common set of metrics used to represent biodiversity, it will become easier to identify consistent pathways through which biodiversity contributes to human mental health, regardless of the frameworks being used. The metrics identified by this study make an excellent starting point for future studies looking to quantify biodiversity, particularly when they are combined with the conceptual framework we present in Figure 6.

One limitation of our study is the novel approach used to determine authorship, which may not have captured the true research specialization or interests of each author. However, our method does provide a useful indication of the disciplinary framing of the paper, and this brings a unique perspective that hasn’t previously been examined. The 3 years gap between the two time periods studied is also a potential limitation, as a longer period of time may be required in order to draw meaningful conclusions as to whether research in this field is progressing toward a cohesive body of evidence. This may have also largely contributed to the lack of differences found between time periods for the variables examined. Furthermore, we acknowledge that the wide variation in terminology used by different disciplines contributing to this body of research may have precluded several studies from being included in our analysis. For example, a mechanistic study of frontal alpha asymmetry as a direct neurological response to urban green spaces was not included in our study as they used a contemplative landscape model to represent variation between greenspaces, rather than a direct measure of real or perceived biodiversity (e.g., Olszewska-Guizzo et al., 2020). Bringing together these diverse fields and framings is one of the key challenges that need to be addressed. The simplest solution is to ensure that biodiversity is explicitly included in the title, abstract or keywords of studies that include it as a consideration; and that a simplified search term such as “human health” or “wellbeing” is included in the keywords for studies with more specific measures, as demonstrated by Olszewska-Guizzo et al. (2020). This will be particularly important for studies that employ additional frameworks and tools such as the Actor Network Theory or Place Standard mentioned above, as these approaches do not always explicitly consider biodiversity, and therefore could potentially be overlooked in the absence of an additional signal.

The range of contributing disciplines and use of variables to the body of research reviewed in this study reflects a growing interest in this topic, and the complexity of relationships between biodiversity and mental health and wellbeing. The variation in uses of guiding theory and variables also reflects the many different ways by which researchers are trying to understand and test these relationships. The use of mixed methods approaches within recent works demonstrates a growing interest in understanding the underlying mechanisms (physiology) and how these relationships can inform greenspace design (experimental planting plots). This study has shown that while there is still great inconsistency and lack of coherence in study designs even in recent works, it is clear that different disciplines have unique perspectives to offer, and that continued interdisciplinary collaboration within mechanistic studies could contribute to a cohesive body of evidence, to inform strategies or policies that aim for win-win scenarios for both biodiversity conservation and mental health and wellbeing.