Research on the benefits of nature to people: How much overlap is there in citations and terms for ‘nature’ across disciplines?

1. Research on the diverse benefits of nature to people is characterised by a broad range of disciplines involved, encompassing a variety of approaches, methods and terminologies. While a diversity of approaches is valuable, it can lead to dif - ficulties in integrating and sharing findings, and could form a barrier to effective knowledge exchange, hindering the development and applications of research outputs. 2. As a starting point for this scoping review, we chose four broad research areas (medicine, psychology, education and environment), selected to represent dis - parate approaches to research on the benefits of nature to people, within and across which to explore overlap in citations and terms used to describe nature. 3. We conducted expert consultation and a snowball-based approach to source publications, resulting in a sample of 210 papers spanning multiple disciplines within each of our four research areas. For each paper, we recorded the discipline of the journal in which it was published (publishing discipline), the discipline of its first author (first-author discipline), the number of times journals of each disci - pline were cited in their bibliographies (cited discipline) and the term(s) used in the paper's title or abstract to


| INTRODUC TI ON
Research into the health and wellbeing benefits of nature is characterised by the involvement of a wide range of disciplines, terminologies and methods.Several existing systematic reviews have attempted to capture this range (e.g.Browning & Rigolon, 2019;Collins et al., 2020;Kosanic & Petzold, 2020;Lovell et al., 2014Lovell et al., , 2015;;Reyes-Riveros et al., 2021;Wolf et al., 2020), but they are often hindered by the diversity of terms and approaches used across disciplines, making it difficult to bring together all relevant research through standard systematic search approaches.This lack of integration between disciplines could prove a significant barrier to the development of the field since findings are not efficiently disseminated, collaborations are not supported and techniques are not shared.At a basic level, if research from different disciplines uses different terms to refer to nature and its benefits, searching the published literature to find the most up-to-date work, either to inform new research or shape policy, becomes a challenge.This hinders not just research in this area but also its applications in the real world.
Comparing the findings of two well executed flagship systematic reviews illustrates the problem well.'A systematic review of the health and wellbeing benefits of biodiverse environments' by Lovell et al. (2014) used a narrative synthesis approach to map the interdisciplinary field of inquiry and to synthesise the current evidence base on whether biodiverse environments are health promoting.
'Benefits of nature contact for children' by Chawla (2015) also used a narrative approach to characterise research from the 1970s to 2015, mapping changes in approaches and focus on different dimensions of children's wellbeing.Both reviews state that they took a narrative synthesis approach in order to account for the heterogeneity of the literature.Despite this, no papers cited in Lovell et al. include any aspect of learning, nature connection or creative play in the range of benefits measured, while these types of benefits are a key theme within the Chawla review.This is understandable given that Chawla (2015) focused on benefits specific to children, which are more likely to include an education or school focus, while Lovell et al. (2014) set out to characterise evidence on the health and wellbeing benefits of nature, which may not extend to education directly.However, the fact that natural environments can affect the instance of creative play among children and promote children's learning (Kuo et al., 2019) suggests that there may be parallel benefits to explore in adults-research that, if it exists and if the definition of wellbeing were sufficiently broad, would ideally also be synthesised in the Lovell et al. review.This suggests both that research stemming from an educational background might exclude research with adults as a study group and that research stemming from a public health or clinical discipline might not include relevant findings from the education literature.
Characterising research on people and nature as broad is not new.Several reviews exist on the topic (e.g.Lovell et al., 2014), as do several reviews of reviews (e.g.Hartig et al., 2014).Hartig et al. (2014) identified three central problems with their reviews of this research.First, that search strategies often fail to capture the full range of 'natural environments' because of the need to use multiple terms for 'nature' in the search string and the resulting high likelihood of missing relevant work.They also highlighted that part of this issue lies with the challenge of defining 'nature'.For example, in some research, the term 'natural environment', meaning one with an absence of human presence or interference, is adequate and appropriate.However, in other papers, the 'nature' of interest is part of the built environment (e.g.view of trees through a window, urban parks), so other terms, such as 'urban nature', 'green space' or 'nature experience' are more appropriate.The plethora of terms used, therefore, often have distinct meanings, suggesting that all disciplines using the same term is not constructive or feasible.
Secondly, Hartig et al. (2014) highlighted that disciplines with an individual-based approach (e.g.psychology, education) often ignore research from disciplines with a population-based approach and public health papers, indicating that research from these disciplines may be particularly prone to being overlooked by search strings.
6.The wide range of disciplines cited is encouraging, since this indicates that diverse research areas are generally aware of each other's work.However, to avoid unnecessary expansion of nature terms and support searchability, we propose four key terms for nature: ('outdoor learning' OR 'outdoor education'), ('nature' OR 'natural'), ('green space' OR 'greenspace') and ('biodiversity' or 'trees'), which could be used across disciplines.We particularly propose that at least one of these be included in every paper, and all four should be included in review search strings.This is likely to result in a better understanding of the valuable, disparate contributions made by different disciplines to this expanding and important topic.
health, integration, knowledge exchange, literature review, nature benefits, relational values, scoping review, snowball, wellbeing (e.g.public health).Again, the challenge is how to capture and exchange knowledge across such different approaches, since both perspectives, at the individual and population level, offer important insights on the benefits that nature can offer.Third, Hartig et al. highlighted the failure of reviews to successfully address the variety of health outcomes explored.This is partly because outcomes are sensitive to contact with nature through different pathways and over different timescales, so reviews that pool results, in an attempt to make up for a lack of multiple studies following consistent methods, may mischaracterise the evidence base.In short, studies that aim to create an all-encompassing search or review approach that suits many disciplines might choose to focus only on the more straightforward commonalities, which means that the approach might suffer from being too reductionist in either the categories it chooses to study or those it finds.For example, in a systematic review on the health outcomes of having urban trees, Wolf et al. pooled outcomes into three broad groups to account for the wide range of outcomes recorded: reducing harm (e.g.UV exposure, pollen, air pollution), restoring capacities (e.g.attentional restoration) and building capacities (e.g.birth outcomes, active living) (Wolf et al., 2020).While this approach was valuable for bringing out consensus findings, it does mean that more specific effects are likely to have been obscured.
Therefore, the fact that such a broad range of research exists, spanning multiple keywords, measuring multiple outcomes and adopting a range of approaches, is a challenge that has already been well characterised.However, having such a broad evidence base is not necessarily a problem if the involved disciplines read and cite each other, and understand each other's language and approaches.
In this paper, we take four broad research areas as our starting point (medicine, psychology, education and environment), identified by the authors as being particularly distinct and disparate in their approaches to human-nature research.Within these research areas, we quantify the overlap between disciplines (specifically, the disciplines of the journals in which this research is published ('publishing discipline') and the disciplines of its first authors ('firstauthor discipline')) with regard to the number of times journals of each discipline are cited in bibliographies ('cited discipline') and the terminology used to describe nature in paper titles and abstracts ('nature terms').By doing so, we aim to identify areas of commonality and suggest pragmatic approaches to better integrate disciplines, fostering efficient collaboration and knowledge exchange.Specifically, we address the following key questions: 1. Within our starting research areas of medicine, psychology, education and environment, which disciplines are carrying out research on the benefits of nature to people? 2. What disciplines are being cited by this research, and does this vary between publishing or first-author disciplines?
3. What terms are used to refer to nature, and does this vary between publishing or first-author disciplines?

| Summary
In a conventional literature review, the first stage is often to design a search string that best captures the keywords of the research topic of interest.This search string is then used to generate a list of all relevant published papers.However, the topic 'people and nature' is so broad, encompassing disciplines that use vastly different approaches, perspectives, methods and terminologies, that designing a search string broad enough to capture all relevant papers generates an unmanageably large, and majority irrelevant, dataset of papers.If the search string is made more specific, the list of results is then too narrow to capture the full range of relevant research.To circumvent this issue, from our knowledge of the literature, we identified four broad, disparate research areas, each spanning multiple disciplines, that we used as a framework to draw up a list of benchmark papers and experts, whom we then consulted to generate a starting list of relevant papers.We then applied a snowball-based approach (Moher et al., 2009) to this starting list, in which we added papers to our list from the bibliographies of this starting list (see Section 2.4), resulting in a final sample that captured a snapshot of relevant research across these four disparate disciplines but likely not included every relevant published paper (Figure 1).

| Selection of benchmark papers
We began by identifying four disparate research areas that captured a broad snapshot of research into the benefits of nature to people: medicine, psychology, education and environment (Table 1).
We defined 'nature' as any aspect of the natural world as opposed to the built environment, but we were not prescriptive about how or whether this was quantitatively measured.We chose these four areas a priori because our knowledge of the literature indicated that research into people and nature in each of these areas adopts a distinct perspective and often uses different terminology to explore their respective research questions.For example, research in psychology often uses 'attentional restoration theory' as a framework, using language such as 'restorative settings' and 'attentional fatigue', while a superficially similar paper in the medicine research area might take a physiological approach, talking about 'stress', 'cortisol levels' and 'green space'.
Using these four broad research areas as our initial framework, we identified a list of 10 'benchmark' papers (Table 1) published between 1984 and 2017 (Medicine: 3, Psychology: 2, Education: 2, Environment: 3) and a list of 21 experts (Medicine: 2, Psychology: 8, Education: 3, Environment: 8), ensuring all four research areas were represented in each list.
We selected our list of 10 benchmark papers based on our own knowledge of the literature, with the intention of capturing the disparate approaches and terminologies used by the four research areas.Each paper was chosen because it is regularly cited in the introductions of research into the benefits of nature to people.
Given that this list was subject to our own biases and knowledge gaps, we then sent it to our 21 experts in an effort to expand the scope of the list.
The experts included authors of influential and well-cited research studies in the field of people and nature, including the authors of some papers in our benchmark list, and academics whom we knew to have expertise in this area.Again, this list was drawn up by ourselves to capture the disparate approaches across our four research areas.All experts had a PhD in a relevant subject and were currently practicing academics in the field of human-nature research.

| Expert consultation
We emailed our list of benchmark papers to all 21 experts in July 2020, asking them to suggest other papers that could be added to this list (Figure 1; see Appendix S1 for the full email request).We gave them the 10 benchmark papers to enable them to appreciate the full scope of the work we wanted to capture and to encourage them to think of relevant papers beyond their own research area.
Nine experts responded in July and August 2020 (Medicine: 2, Psychology: 3, Education: 2, Environment: 2), yielding a total of 47 additional papers (Medicine: 13, Psychology: 11, Education: 10, Environment: 13).Only two of these 47 additional papers were recommended by more than one expert.Combining these 47 with our 10 benchmark papers gave us a starting list of 57 papers spread across the four research areas, on which we carried out a snowballbased approach to achieve our full sample (Figure 1).From this point on, we no longer used our four starting research areas as a classification basis, since these were designed solely as a starting point from which to expand our search.

| Snowball-based approach
We exported the bibliographies of all 57 papers and identified all citations that were relevant to our research area.We classed as relevant any paper that explored the relationship between some aspect of nature and some benefit to humans, inclusive of quantitative, qualitative and descriptive research and synthesis reviews.We defined 'nature' as any aspect of the natural world as opposed to the built environment, but we were not prescriptive about how or whether this was quantitatively measured.Our definition therefore encompassed crude aspects of surroundings (e.g.'natural views' (artificial or real) or 'green space'), measures of 'greenness' (e.g. using NDVI or land area), individual aspects of the natural environment (e.g.trees or tree cover) and more specific measures of biodiversity (e.g.plant or butterfly species richness).We defined 'benefits to humans' as benefits to physical health, mental health, wellbeing, behaviour, psychology, education and nature connection.
We then randomly selected up to three relevant citations from each paper to add to our sample.To ensure selection was random, we used Microsoft Excel to generate three random numbers F I G U R E 1 The stages followed to source our sample of 210 papers.White boxes show the distribution of papers at each stage across the four relevant research areas.
between one and the total number of citations for each paper.If a randomly selected citation was either not relevant (according to our above inclusion/exclusion criteria), already present in our sample or unavailable in English, we continued randomly selecting citations until we had up to three new, relevant citations from the bibliography.Following this approach, we rejected 400 papers on the basis of our inclusion/exclusion criteria and 101 duplicates (Table S1).This approach resulted in a total sample of 210 papers (Figure 1, Table S2).

| Data gathering
For each paper, we recorded the discipline of the journal in which it was published (publishing discipline), the discipline of its first author (first-author discipline), the number of times journals of each discipline were cited in their bibliographies (cited discipline) and the term(s) used in the paper's title or abstract to describe the aspect of nature being explored (nature term).Publishing disciplines and cited disciplines were decided from a combination of a journal's title and scope (title and summary for grey literature, or title and blurb for books).First-author discipline was decided by looking at a combination of the author's affiliation and list of publications.Nature terms were selected from a paper's title (n = 194), only referring to the paper's abstract if no nature term was present in the title (n = 16).All data gathering was carried out by KH in September-November 2022 to ensure consistency.
Our aim in choosing the disciplines for each category (publishing discipline, first-author discipline and cited discipline) was to provide as precise a classification of the discipline of a journal or first author as possible, that is more precise than the four broad research areas used as our starting point.Inevitably, this meant that some disciplines were more precise than others.For example, the category 'Science' (within publishing discipline and cited discipline) is the most precise classification possible for journals such as Science and Nature, since these publish research across the sciences, whereas 'Psychology' (e.g. for journals such as Frontiers in Psychology) is more precise again than 'Biology' (e.g. for journals such as Biology Letters).These classifications are therefore not mutually exclusive categories.Rather, they represent an attempt to classify the journals in which this research is being published to the most precise degree possible based on the journal scopes.Within the cited disciplines, the category 'Other' was created to accommodate nine publications in law, philosophy and current affairs, since these disciplines do not fit into any other discipline but are not sufficiently connected with human-nature research to merit their own categories.For full details of the inclusion criteria for each discipline, see Table 2.  TA B L E 2 Inclusion criteria and example journals for each discipline within the publishing discipline, first-author discipline and cited discipline.(Neuwirth & Brewer, 2022), ggsignif (Ahlmann-Eltze & Patil, 2021), gg-plot2 (Wickham, 2016) and cowplot (Wilke, 2020) for data wrangling, exploration and visualisation.Exploration followed Zuur et al. (2010), in which the authors set out a data exploration protocol that reduces the chances of type one and type two errors, including through early identification of outliers and checking for heterogeneity of variance and zero inflation.We fitted multivariate generalised linear models (mGLMs) using mvabund (Wang et al., 2022) and chi-square tests of independence using stats (R Core Team, 2022).Unless otherwise stated, we fitted models to negative binomial distributions, including publishing discipline and first-author discipline as fixed effects.

| Cited disciplines
We used mGLMs (Warton et al., 2012;Warton & Hui, 2017) to analyse communities of cited disciplines, followed by univariate analyses if publishing discipline or first-author discipline was significant (p = 0.05).
We validated mGLMs by plotting Dunn-Smyth residuals against fitted values and covariates and verifying no patterns were present (Wang et al., 2012(Wang et al., , 2022)).We determined the significance of fixed effects using likelihood ratio tests (LRTs) and by bootstrapping probability integral transform residuals using 10,000 resampling iterations (Warton et al., 2017).If either fixed effect was significant (p < 0.05), we ran univariate analyses.We adjusted univariate p values to correct for multiple testing using a step-down resampling algorithm (Wang et al., 2012), but otherwise our statistical approach remained unchanged from the multivariate parent models.

| Nature terms
Chi-square tests of independence were used to test for differences in the proportion of papers using nature terms in their title or abstract from different categories between publishing disciplines and between papers with first authors from different disciplines.

| First-author disciplines
Based on the first-authors' affiliations and list of published papers, the total number of first-author disciplines within our sample of

| Cited disciplines
Overall, among the bibliographies of our sample of 210 papers,

| Nature terms
A total of 103 different terms for nature were used in the titles or abstracts of our sample of 210 papers (Figure S2), which we lumped into 12 broad categories: biodiversity, forest bathing, forest school, green, natural, nature, outdoor, restorative, trees, urban, other and multiple (Table 3; see Table S3 for a full list of terms in each category).The majority of papers used a nature term in their title  The frequency of papers using nature terms from different categories was significantly different between publishing disciplines (χ 2 = 227.81,df = 99, p < 0.0001; Figure 3a) and between papers with first authors from different disciplines (χ 2 = 236.48,df = 77, p < 0.0001; Figure 3b).

| DISCUSS ION
We found that the communities of disciplines cited differed between publishing disciplines and first-author disciplines.The communities cited by psychology, education and public health papers were particularly distinct, each citing a larger proportion of papers within their own disciplines.Medicine was the most consistently cited discipline across publishing and first-author disciplines.This difference in the focus of citations between disciplines is not surprising given the need for research to be informed by and grounded in similar work and approaches.However, given the multidisciplinary characteristics of research focused on the benefits of nature to humans and the wide range of potential outcomes from this work, this bias in some disciplines could limit the transfer of ideas and wider applications of research findings.In contrast, it is encouraging to see the generally wide range of disciplines being cited across publishing and first-author disciplines, since this suggests that disparate research areas are generally aware of each other's work, although more integration is possible.
Similar results were found in the communities of disciplines cited based on both publishing and first-author disciplines.Specifically, both education-and psychology-based papers differed in their cited communities from environment-and public health-based papers when analysed based on both publishing discipline and first-author discipline.This consistency in findings suggests that authors are generally publishing in journals of their own discipline.While this might be effective in disseminating findings to peers within disciplines, this approach is likely to be less effective in communicating findings across disciplinary boundaries-a particular problem for an inherently multidisciplinary field such as human-nature research.To address this, we suggest authors aim to publish in a wider range of journal disciplines, beyond those closely aligned with their own area.
This might encourage the use of nature terms beyond those usually used in the research discipline, making research more easily findable.
The types of nature terms used in paper titles and abstracts were also significantly different between different publishing and first-author disciplines.Part of this is likely due to the logical consistency seen within disciplines.For example, 'green' terms were favoured by public health-authored papers (48.3% of papers), 'outdoor' terms by education-authored papers (42.9%), 'urban' terms by urban planning-authored papers (38.5% of papers), 'nature' terms by psychology-authored papers (34.8%) and 'forest bathing' terms by medicine-authored papers (32.3%).This makes sense given the different approaches and focuses taken by these disciplines.For example, public health papers are often focused on 'public green space' (e.g.Benton et al., 2018) or take a measure of 'residential greenness' (e.g.Markevych et al., 2014); education papers often explore the effects of 'outdoor learning' (e.g.Christie et al., 2016) or 'outdoor education' programmes (e.g.Quibell et al., 2017); urban planning papers often focus on the function of 'urban trees' (e.g.Donovan et al., 2011) or the effects of the 'built environment' TA B L E 3 Categories of nature terms used in paper titles and abstracts, along with the inclusion criteria and the number of terms for each.
See Table S3 for a full list of terms in each category.(e.g.Helbich et al., 2016); psychology papers often talk about 'nature restoration' (e.g.White et al., 2013) in the context of attentional restoration theory; and medicine papers are the most common home for studies examining the clinical effects (e.g.T killer cell activity or cortisol levels) of 'forest bathing' (e.g.Li et al., 2008).
This suggests that disciplines are using distinct terms to refer to the aspect of nature relevant to their own discipline with a good degree of consistency and logic.Given the wide range of pathways through  3).
which the people-nature relationship can form and manifest, consistent use of nature terms within disciplines is helpful, since it is likely to clarify the distinct contribution made by each discipline to understanding this multifaceted relationship.
However, within certain disciplines, a large range of nature terms were used.This was particularly the case for psychology .This suggests that there is progress to be made to achieve full consistency within each discipline or that some disciplines are covering a wider range of research topics than others.While some degree of variation in nature terms within a discipline is expected (e.g. it is reasonable to expect medicine papers to investigate the clinical effects of 'green exercise', e.g.Duncan et al., 2014 and'forest bathing', e.g. Li, 2010), the wide range of categories used by disciplines such as psychology and public health could be acting as a barrier to effective knowledge exchange between disciplines, adding noise and inefficiency to literature searches that are intended to inform research and underpin systematic reviews.
As Hartig et al. (2014) identified, the problem of defining 'nature', and the variety of ways in which the nature-human pathway can act, has led to a plethora of nature terms in regular use in the literature (Hartig et al., 2014).To some degree, this is necessary in order to describe accurately and precisely the various aspects of nature (e.g.urban nature such as allotments vs. natural landscapes such as forests) and the various pathways of interaction with people (e.g.attention restoration vs. facilitation of active lifestyles).
However, this can cause problems for defining efficient search strings and retrieving all published work on a topic, whether to inform new research, draw policy conclusions or support literature reviews (Hartig et al., 2014).
Similar pairwise groupings between disciplines were present in both sets of results: communities of cited disciplines and nature terms used.This speaks to consistent, greater similarities and cross-over between some discipline groupings than between others.
Specifically, education, psychology, environment and public health emerged as consistently different from each other, while medicine and public health appeared consistently similar, as did environment and urban planning.The four consistently different disciplines we have identified in our results (education, psychology, environment and public health) align closely with the four broad research areas we used for the basis of this study (medicine, psychology, education, environment), from which we built our benchmark list and expanded our search.Our results therefore provide validation of our methods and of our identification of these broad research areas as using distinct frameworks and terminologies to frame and explore the human-nature relationship.We suggest that it is the boundaries between these four research areas in particular that could be hindering better integration of human-nature research.While efficient communication within these areas makes intuitive sense, we propose that facilitating better knowledge exchange between these four research areas would significantly improve the development of human-nature research and the consequent communication and application of findings.

| Limitations
There are important limitations to our approach, and therefore caveats to our results, that merit discussion.First, and most fundamentally, by choosing our four research areas as the starting point for this scoping review (medicine, psychology, education and environment), we limited the disciplines that were included in our results.
Our process of expert consultation was an attempt to lessen these biases, since we reached out to experts in disparate areas (e.g.education and medicine) to ask about potentially overlooked papers or disciplines.However, we drew up this list of experts ourselves, and as such, it was subject to the same biases that were inherent in our starting research areas.Ideally, the starting research areas and list of experts would have been determined by a pilot study and/or scoping interviews.However, given limited resources, we opted instead to accept that there are inherent biases in our approach and results and to caveat them here.
Another limitation to consider is that we chose to limit the scope of our review to the benefits of nature to humans rather than to encompass the full scope of interactions between humans and nature.
Different results may be found via a larger, more inclusive review.
The fact that the term 'ecosystem services' did not appear in our results as a key term used to describe nature, despite the fact that a lot of research in this area focuses on cultural or relational ecosystem services, is likely to be a result of this or our first limitation and indicates the need for further study.
Third, we chose the discipline of the first author of papers as a measure of the disciplines in which this research is being conducted.
It is often the case that the first author is a student or early-career researcher whose disciplinary affiliation is incipient and in flux, while a paper's last author is likely to have a more concrete disciplinary alignment.However, we chose this approach because it is often the case that several authors more senior than a paper's first author have significant input into a paper's approach, particularly in inherently cross-disciplinary research.The first-author's affiliation is therefore often a reflection of the affiliations of more senior authors, given the frequency of the supervisory nature of this relationship.Given logistical constraints, we did not have sufficient resources to carry out the same approach for both first-and last-author disciplines.We therefore opted for this measure of first-author discipline as an imperfect but workable approach, but nonetheless, we would like to highlight this caveat here.
All statistical analyses were performed in R version 4.1.3(R Core Team, 2022) within R Studio version Build 461 (RStudio Team, 2022).We used tidyr(Wickham & Girlich, 2022), RColorBrewer TA B L E 1 The four starting research areas, along with the topics that each includes.
1603 different journals or other sources (e.g.grey literature, books) were cited, with the 65 most commonly cited journals each being cited at least 20 times (Figure S1).Based on journal title and scope (title and summary for grey literature, or title and blurb for books), the total number of cited disciplines within the bibliographies of our sample of 210 papers was 12: biology (n = 133), education (n = 112), engineering (n = 9), environment (n = 263), geography (n = 47), medicine (n = 427), psychology (n = 220), public health (n = 108), science (n = 55), social science (n = 182), urban planning (n = 38) and other (n = 9).Both publishing discipline (LRT = 300.7,p < 0.0001; Figure 2a) and first-author discipline (LRT = 201.6,p < 0.0001; Figure 2b) differed significantly in the communities of disciplines cited.Specifically, the communities of cited disciplines in papers published in education journals differed from those published in environment, public health and urban planning journals, and papers published in psychology journals differed from those published in environment and public health journals (p < 0.05 for post-hoc comparisons).For first-author discipline, the communities of cited disciplines in papers with a first author in education differed from those with a first author in environment, medicine and public health; those with a first author in psychology differed from those with a first author in environment, medicine and public health; and those with a first author in public health differed from those with a first author in environment (p < 0.05 for post-hoc comparisons).Univariate analyses indicated that different cited communities in education (LRT = 53.91,p = 0.0005), environment (LRT = 78.57,p = 0.0001), geography (LRT = 46.87,p = 0.0009), medicine (LRT = 47.49,p = 0.0009), psychology (LRT = 65.01,p = 0.0001), public health (LRT = 61.86,p = 0.0002) and urban planning (LRT = 58.21,p = 0.0003) papers were the primary drivers of these differences.For first-author discipline, univariate analyses indicated that different cited communities in papers with a first author in education (LRT = 53.91,p = 0.0002), environment (LRT = 78.57,p = 0.0001), geography (LRT = 46.87,p = 0.0009), medicine (LRT = 47.49,p = 0.0009), psychology (LRT = 65.01,p = 0.0001), public health (LRT = 61.86,p = 0.0001) and urban planning (LRT = 58.21,p = 0.0001) were the primary drivers of differences.
(n = 164), with only 16 needing us to refer to the abstract to find a nature term (e.g.Title: 'View through a window may influence recovery from surgery').

F
Communities of cited disciplines by (a) publishing discipline and (b) first-author discipline.Each bar is a separate (a) publishing discipline or (b) first-author discipline, arranged alphabetically.Bars are coloured by the discipline of the cited journals.Error bars indicate the minimum and maximum number of citations in the dataset.Significant differences between pairs of publishing and first-author disciplines are indicated by square brackets.*p < 0.05, **p < 0.01.Not all cited disciplines are represented in the lists of publishing or firstauthor disciplines.

F
Percentage of papers using different categories of nature terms in the paper titles or abstracts, split by (a) publishing discipline and (b) first-author discipline.Each bar is a separate (a) publishing discipline or (b) first-author discipline, arranged alphabetically.Bars are coloured by category of nature term (see Table

(
11 categories by publishing discipline, 9 by first-author discipline), public health (10 categories by publishing discipline, 10 by first-author discipline), environment (7 categories by publishing discipline, 10 by first-author discipline), medicine (9 categories by publishing discipline, 7 by first-author discipline) and urban planning (9 categories by publishing discipline, 6 by first-author discipline).By contrast, education papers showed greater consistency in the nature terms used in their titles and abstracts (only 4 categories by publishing discipline, 5 by first-author discipline)

Research area Included topics Benchmark papers
MedicinePublic health, physiology, physical activity, mental health, physical health Maas J, Verheij RA, et al. 'Green space, urbanity, and health: how strong is the relation?'Journal of Epidemiology and Community Health.2006.60(7):587-92 Ulrich RS. 'View through a window may influence recovery from surgery.' Science.1984.224(4647):420-1 Mitchell R, Popham F. 'Effect of exposure to natural environment on health inequalities: an observational population study.'Lancet.2008.Casares-Porcel M, et al. 'Assessing allergenicity in urban parks: A nature-based solution to reduce the impact on public health.'Environmental Research.2017.155:219-227 Dallimer M, Irvine KN, et al. 'Biodiversity and the Feel-Good Factor: Understanding Associations between Self-Reported Human Well-being and Species Richness.'BioScience.2012.62(1):47-55 Fuller RA, Irvine KN, et al. 'Psychological benefits of greenspace increase with biodiversity.'Biology Letters.2007.3(4):390-4 BiodiversityTerms that require measurement of a quantifiable aspect of biodiversity (e.g.species richness, vegetation) 11Forest bathing Terms that refer to the Japanese practice of shinrin-yoku (the Japanese term for the practice of forest bathing)