Main

Multiple anthropogenic drivers are causing biodiversity loss worldwide1. Such biodiversity declines have profound consequences for ecosystem functioning and, consequently, the goods and services that underpin human health and well-being2,3. For instance, it is now widely accepted that interacting with nature (for example, in urban parks, forests and coastal areas) leads to stress relief, enhanced mood, improved cognitive ability and social cohesion, amongst an array of other benefits4,5. Such evidence is accumulating from across the world, including from low-, middle- and high-income countries6. When these individual-level gains in health and well-being are scaled up to entire populations, they can equate to substantial cost savings for the public health sector. This is pertinent in locations where the prevalence of mental ill health is particularly high (for example, in Europe7) and given that human well-being is a predictor of both life expectancy and mortality8,9.

Despite abundant research demonstrating that interactions with nature benefit human well-being, we still lack conclusive empirical evidence regarding the role of biodiversity specifically. Biodiversity is the living component of nature, incorporating “the diversity within species, between species and of ecosystems”10. Many existing studies use proxy measures of nature, such as remotely sensed ‘green space’11,12, revealing correlative associations with well-being across large-scale cross-sections of the human population. However, these macroscale approaches treat green spaces as homogeneous entities, overlooking the fact that people’s relationships with biodiversity are both contextually and culturally specific13. The context relates to the physical and ecological place-based characteristics of a green space, which can be highly variable within and between ecosystems of the same type14. How a person responds to a green space will also be influenced by personal, societal and cultural associations, as well as previous experiences13,15. Understanding how people experience biodiversity is therefore key to successfully managing biodiversity to facilitate human well-being, incorporating it into sustainable land-use planning initiatives16, nature-based solutions and social (‘green’) prescribing interventions17.

Following the ‘biopsychosocial–spiritual’ model of health, which is adapted from medicine18,19, human well-being is thought to comprise five separate domains: physical (the body and how someone feels physically), emotional (positive and negative mood), cognitive (state of mind), social (perceived connections with others) and spiritual (relationships with one’s self or something greater than one’s self). People’s multisensory experiences of biodiversity may elicit both positive and negative responses in one or more of these well-being domains20. For instance, hearing the song of a male European robin (Erithacus rubecula) might prompt a positive emotional response (for example, joy), while the stinging hairs of a common nettle (Urtica dioica) may provoke a negative physical response (for example, physical discomfort). Each species, however, may support multiple traits, potentially with independent impacts (for example, robins sing and have plumage that is red in colour, each trait potentially stimulating a different type of well-being). Studies of biodiversity–well-being relationships must thus consider the ecological community that makes up a green space in its entirety, moving beyond just single traits (for example, flower colour21) and/or taxonomic groups (for example, birds22,23,24) and towards combinations of traits across multiple taxa simultaneously. The subsequent compound effects on multiple types of human well-being would then better reflect the real-world experience of interacting with biodiversity in a particular place.

Ecologists have traditionally examined how the biotic and abiotic environment influences species traits (‘response traits’)25. Some, but not all, of these traits also function to supply ecosystem services (‘effect traits’) (for example, proboscis length of insect pollinators and pollination efficiency)26. Species’ traits that directly elicit human well-being responses could therefore be considered effect traits. The functional effect of a species on an ecosystem is thought to be proportional to its contribution to total biomass across the ecological community (that is, the effect traits of dominant species drive ecosystem function, the so-called mass ratio hypothesis27). Species may also occupy functionally distinct niches, using available resources in a complementary way (niche complementarity hypothesis28). Across species, there may also be overlap in effect traits that deliver multiple functions (multifunctionality), within and across different ecosystem service classes29,30,31,32. Ecological communities can subsequently be examined for redundancy (where species delivering the same functions as others become functionally redundant/exchangeable) and complementarity (optimal combinations of species that deliver the maximum services). Such an approach is useful when designing cost-efficient conservation actions and allocating resources to support the delivery of specific ecosystem services. While well established for regulating and provisioning ecosystem services, only a handful of studies have examined effect traits for cultural ecosystem services (the identities, capabilities and experiences that people actively create and express through interactions with ecosystems33), and those studies have either been restricted to a single taxon (birds) or have not measured well-being as an outcome22,24.

Here we demonstrate a novel analytical approach through which the linkage between species’ effect traits in an ecological community and human well-being can be examined at a granular level. We asked two questions. (1) Which species’ effect traits relate to each type of human well-being? (2) To what extent are species, and the effect traits they exhibit, redundant or complementary in the delivery of human well-being? We held a series of participatory workshops, one per season (winter, spring, summer and autumn), with a diverse cross-section of the public (n = 194). During each workshop, participants were taken to the same two British forests. We then documented how the species’ traits identified by participants elicited self-reported positive and negative responses across the five well-being domains (physical, emotional, cognitive, social and spiritual; Table 1). We also identified ‘global’ well-being34, recognizing that these multiple domains are interdependent in contributing to how one feels overall (akin to the idea of ‘whole-person health’, an overall sense of health/wellness35,36). We used the words of participants when documenting incidences of species’ effect traits eliciting well-being. For example, one row of data is formed when a participant describes a negative physical response (allergy) to the behaviour (blossoming) of an elder (Sambucus nigra): “some fluffy stuff on it which set off my hay fever in the spring so I don’t like those”.

Table 1 Definitions of the self-reported human well-being domains
Full size table

Our study centred on forest ecosystems, which declined in areal extent by 31.6% globally between 1990 and 2015 due to deforestation, fragmentation and other pressures37. Today, forests cover approximately one-third of global terrestrial surface area and support ecosystem services valued at ~9% of global gross domestic product38. Including cultural service benefits within such assessments remains a challenge, particularly given the diverse ways in which people relate to nature, yet this is crucial for creating conservation policies that are inclusive of the people they seek to benefit3.

Results

Identifying species’ effect traits

Participants articulated 102 unique effect traits (Table 2) across 403 species (taxonomic kingdoms: animals, fungi and plants) eliciting a well-being response (n = 1,815 unique effect trait–well-being combinations). Of these effect traits, colours (for example, pink, gold and silver), and behaviours (for example, hopping, decaying and elusive) had the greatest variety (29.4% each), followed by sounds (for example, creaking, chirping and screaming; 19.6%), with a smaller number of textures (for example, smooth, spongy and prickly; 14.7%) and smells (for example, damp, pine and sweet; 6.9%) mentioned. However, sounds most frequently stimulated well-being responses (40.4%), above behaviours (26.5%), colours (23.7%), textures (7.3%) and smells (2.1%; Fig. 1a). This reveals the relative importance of forest sounds for well-being over the other effect trait types. It is possible that sounds could be more conspicuous than other effect traits for species that are difficult to encounter directly in forest vegetation. Additionally, it highlights the role of species’ behaviours, which have received very little research attention in relation to well-being.

Table 2 Definitions of trait types and example effect traits
Full size table
Fig. 1: The contribution of species’ effect traits to different types of human well-being.
figure 1

Effect traits eliciting a well-being response, with n = 1,815 unique effect trait–well-being combinations. a, The number of unique effect trait–well-being combinations broken down by effect trait type and well-being type. b, The shape of the species–effect trait relationship for each type of well-being. For the line colour code, see c. c, Ordination based on non-metric multidimensional scaling (NMDS) of a Bray–Curtis dissimilarity matrix. The positions of points (effect trait–well-being combinations, shaped by trait type) represent dissimilarity in the number of incidences that effect traits elicited different types of well-being. The labelled effect traits are indicators of each well-being type (Supplementary Table 1). Large circles represent mean centroids for each well-being type, with horizontal and vertical error bars showing 95% confidence intervals. A low level of stress (<0.05) indicated excellent fit. Note, no incidences meant it was not possible to create centroids for negative cognitive, social or global well-being.

Full size image

Redundancy and complementarity

Over 85% of effect trait–well-being combinations were positive, spanning physical, emotional, cognitive, social, spiritual and global well-being, but particularly spiritual well-being (Fig. 1a). Indeed, there were comparatively few negative effect trait–well-being combinations (Fig. 1a). Moreover, a high level of redundancy (plateauing lines) was reached after relatively few species for negative types of well-being (Fig. 1b). This suggests that a small number of species were sufficient to deliver negative well-being, with little additional impact arising from greater numbers of species in the ecological community. This plateau also implies that all the species and effect traits that elicit negative well-being were documented through our methodology. In contrast, the inclines for positive emotional and spiritual well-being imply there are still more effect traits and species to be captured. Some ‘keystone’ tree species supported a disproportionate number of unique effect traits, particularly silver birch (Betula pendula), horse chestnut (Aesculus hippocastanum) and English oak (Quercus robur; Extended Data Fig. 1). However, as each additional species brings with it additional effect traits, this suggests that maintaining diversity in forest ecosystems is beneficial for human well-being (Fig. 1b).

By visualizing the data, we found that some effect traits were similar in the frequency with which they elicited different types of well-being, resulting in clusters (for example, sounds in Fig. 1c). These patterns were explained mostly by the species exhibiting the effect trait (23.1%), the type of trait (colour, texture, sound, smell and behaviour; 16.2%) and taxonomic kingdom (animal, plant and fungi) of the species (10.0%), thereby triangulating our understanding of how people relate to forest biodiversity13 (Extended Data Fig. 2). Furthermore, the sets of effect traits that linked to each type of well-being were significantly different (P = 0.001 for each pair; Fig. 1c). We quantified this dissimilarity using Sørensen’s index (>0.5, pink shades in Fig. 2). These differences support the niche complementarity hypothesis28, whereby specific effect traits deliver largely unique types of well-being. Such detail could be used to improve the design of nature-based public health interventions, by managing ecosystems for the species that exhibit particular effect traits (for example, alterations to the biodiversity in a particular place where people interact with nature, such as planting regimes in public parks designed to enhance cognitive restoration39).

Fig. 2: The (dis)similarity of effect traits and species contributing to different types of human well-being.
figure 2

Sørensen’s similarity index for sets of effect traits (lower left triangle) and species (upper right triangle) contributing to different types of well-being. Similarity on a scale from 0 (similar, green) to 1 (dissimilar, pink).

Full size image

However, each species may comprise multiple effect traits. When we calculated the Sørensen’s index between the identities of species for each type of well-being, there were high levels of similarity (<0.5, green shades in Fig. 2). Some species therefore influence multiple types of well-being. For example, the tawny owl (Strix aluco), whose “calling” and “communicating” sounds alongside its “using trees” behaviour elicited three different positive types of well-being (physical, cognitive and spiritual). In some cases, species caused both positive and negative types of well-being: the colours (“black”, “pink” and “red”) of bramble plants (Rubus fruticosus) linked to multiple positive well-being types (physical, emotional and social), while its “prickly” texture generated negative well-being (emotional). In one instance, the “tweeting” effect trait of passerine bird species was an indicator of both positive and negative spiritual well-being (Supplementary Table 1). By contrast, Sørensen’s index for both the effect traits and species that elicited negative physical well-being were largely dissimilar from all other types of well-being (0.71–1 and 0.84–0.99, respectively). One inference that could be drawn from this finding is that such species and their associated effect traits could be removed from forests to improve human well-being. However, this would have potentially profound adverse consequences for biodiversity conservation and the functioning of ecosystems, given that these species and their effect traits could be influencing the delivery of multiple ecosystem services across different classes (for example, provisioning and regulating) that were not examined in this study.

Discussion

Our approach, working across an ecological community, exposed granular levels of detail on how species functionally deliver well-being benefits. Effect trait–well-being incidences depend on the identity of the species and taxon supporting each effect trait, and it is therefore not possible to disaggregate effect traits from the species that host them when it comes to determining whether, and how, human well-being is delivered. Moreover, our findings show that numerous effect traits from across the ecological community can elicit a multitude of well-being types, as well as global well-being, illustrating the true complexity of the biodiversity experience. It is possible that multiple effect traits may also interact, resulting in additive or multiplicative impacts on human well-being (for example, the cumulative effect of smells alongside sounds from one or more species), which warrants future investigation. Such intricacies could be further detailed by measuring the strength of an effect trait (for example, light to dark red, thus accounting for phenotypic variation within species), as well as by measuring levels of human well-being. This identification of thresholds could better inform public health recommendations (for example, ref. 40) and align our study with those examining how differing levels of, and interactions between, multiple effect traits modulate levels of regulating and provisioning ecosystem service benefits26,31.

Our participatory methodology identified the multitude of ways in which people experience biodiversity and positive/negative well-being in particular places. However, our study participants also related to species’ effect traits encountered in the forests through their past experiences with the same/similar species’ effect traits outside of the workshops in other locations (for example, memories of childhood, at home or on holiday). This emphasizes the need for researchers to incorporate such pluralism into ecosystem services assessments3,41, as well as the need for inclusive land-use planning initiatives, nature-based solutions and green prescriptions.

Our approach is a step change in how biodiversity has been considered in biodiversity–health/well-being research so far, moving away from a focus on a limited set of taxa (for example, birds), biodiversity metrics (for example, species richness) or specific types of trait (for example, colour or sound)21,42. The multiplicity of species’ effect traits, and their influence on well-being, captured by sampling participants from a diverse range of socio-demographic/economic backgrounds, emphasized the variation in how people experience forest biodiversity. The rich variation in species’ effect traits was also augmented by holding our participatory workshops across the course of a year, ensuring that any influence of seasonal variation in the conspicuousness of species and, therefore, effect traits was covered by the study design (for example, the colour blue of bluebells Hyacinthoides non-scripta and the winter plumage of birds). These approaches are likely to reveal a further array of effect traits that influence people’s well-being in different ways when applied to other ecosystems (for example, a prevalence of negative physical well-being from allergenic tree or grass pollen in urban ecosystems, particularly in summer). This opens avenues for further research into how biodiversity–well-being linkages could be affected by climatic variability influencing ecological phenology and processes29,43.

Forest restoration is one of twelve targets for maintaining Life on Land (Sustainable Development Goal 15 (ref. 44)) and has become a policy focus globally. A surfeit of regional, national and international initiatives have been devised and implemented to retain, restore and create forests, pledging to plant billions of trees worldwide45,46. Yet these interventions could have low success rates and/or fail to meet anticipated outcomes44. While these initiatives often seek to provide the so-called ‘triple wins’ for climate change, biodiversity and human well-being47,48, many neglect to consider their social and cultural impacts46,49,50. Indeed, without support from those who live in and interact with the landscape, it will be more difficult for restoration and conservation initiatives to succeed. Our findings highlight that biodiversity will not be beneficial for everyone in the same way, which needs to be accounted for in forest restoration policies if they are to deliver both equitable and socially just outcomes. These potential trade-offs between conservation and societal goals can be better informed by granular levels of detail about which species’ effect traits benefit people, as well as the other ecosystem functions and services that they support.

We found that compared with other forest taxa (for example, insects and birds), participants described trees as having a disproportionately large number of unique effect traits that stimulated well-being responses. This is likely attributable to the year-round visibility of trees, with effect trait diversity enhanced by seasonal changes and longevity13. This has important implications for the conservation of forests and trees, particularly those that are old growth. Moreover, tree species are likely to comprise the dominant biomass of such ecosystems, therefore supporting the mass ratio hypothesis27. Species-rich boreal and temperate forests also support high levels of provisioning ecosystem services, with no single species able to deliver them all51. When combined with our study, this demonstrates the multifunctionality of forests and trees, critical to reinforcing national and global policy initiatives to conserve, enhance and restore forests and trees for both people and nature.

Managing the biodiversity within ecosystems to select for species’ effect traits that benefit human well-being has potentially important implications for conservation. For instance, removing the species and effect traits that elicit negative physical well-being may have knock-on negative implications for the ecosystem (for example, disrupting symbiotic relationships or trophic interactions). Furthermore, species with substantial aesthetic value or prominent cultural meaning may be less ecologically or evolutionarily distinct, non-native or not of conservation interest52. In practice, trade-offs may have to be made. If we are to manage ecosystems to promote well-being, extreme care needs to be taken to ensure that there are no unintended adverse consequences for biodiversity conservation and the functioning of ecosystems, given that species’ effect traits can be operating across multiple other classes of ecosystem service.

One potential limitation of our study is that many mental processes are unconscious, meaning that people may sometimes mistakenly attribute cause and effect, relying on existing beliefs or expectations that may be biased or contain errors of judgement53,54. Such phenomena could mean that there were inaccuracies in the way our participants articulated species’ effect traits eliciting well-being responses. Nonetheless, a growing body of literature has demonstrated links between biodiversity and improved objective measures of health and well-being (for example, visiting a sensory garden led to reductions in physiological stress, measured via salivary cortisol55, and a higher density of urban street trees was linked to reduced antidepressant prescription rates56).

Here we have detailed an empirical approach revealing how the biodiversity in a particular place underpins human well-being, which can help to inform how ecological communities could be managed to promote different well-being outcomes. Our approach can be operationalized to create better-tailored public health interventions or architectural/landscape designs (for example, maximizing the likelihood of interactions with certain species), while reducing health inequalities and promoting socially inclusive natural environments41. From a conservation standpoint, it further illustrates the functional consequences of biodiversity loss for human well-being, while raising debate about the consequences of manipulating ecological communities for human benefit. However, our approach can be harnessed to optimize conservation solutions, such as ecological restoration57, biodiversity net gain58 and systematic conservation planning59, for both social and ecological outcomes. We therefore provide a novel grounding for advancing our understanding of, and integration between, the fields of environmental psychology, functional ecology and their wider cultural dimensions (for example, as cultural ecosystem services). Such interdisciplinary thinking is pivotal if we are to effectively move towards a more sustainable and equitable society in the face of global environmental change.

Methods

Study system

This study centred on forests, the focus of several global policy initiatives to conserve and enhance forest carbon in the face of climate change, including the Reducing Emissions from Deforestation and Forest Degradation60, the Sustainable Development Goals61, and the United Nations Decade on Ecosystem Restoration38. Forest restoration and creation (tree-planting) schemes are rapidly gaining traction, but overlook the social and political implications these policies have on people47, despite requiring public support to succeed46. In the United Kingdom, there are 3.2 Mha of forests and woodlands, split between broadleaf (49%) and coniferous (51%) habitat62. They are generally publicly accessible, and are amongst the most frequently visited outdoor spaces in the country63. We took participants from across the country to two forests, geographically located in a central region of Great Britain, to ensure encounters with a diversity of traits: Sherwood Forest (an ancient woodland) and Clumber Park (a managed mixed-deciduous and coniferous plantation forest). These forests were selected as their objective physical and biological characteristics varied and they were not ‘local’ to any of the participants. This was a purposeful decision to maximize the variety of place-based characteristics (species and traits) within and across the two ecosystems, and to minimize the potential influence that previous experience of the forests might have had on the participants’ well-being responses to the objective qualities of the place.

Workshop participants

Participants (n = 194) were recruited via a social research agency between February and October 2019. Individuals were selected across gender (male = 92, female = 102), ethnicity (white = 146, other = 48), age (18–29 years = 60, 30–59 years = 68, 60+ years = 66), region of residence (Scotland = 11, Wales = 10, England = 173), social grade (AB = 56, C1 = 58, C2 = 42, DE = 38), and urban and rural living (urban = 153, rural = 41). The social grades are defined as follows: AB, higher and intermediate managerial, administrative, and professional occupations; C1, supervisory, clerical and junior managerial, administrative, and professional occupations), C2 (skilled manual occupations) and DE (semi-skilled and unskilled manual occupations and unemployed). All participants had to have been living in Great Britain for at least 5 years, irrespective of their nationality. Our approach meant we captured the diversity of the British public, including sectors of society that are often underrepresented in research (for example, elderly, ethnic minorities and lower income earners)64. To encourage workshop attendance and promote inclusivity, participants were incentivized by travel reimbursement and financial remuneration to cover their time with us.

Participants were split over four weekend-long workshops (n = 46–50 per workshop) across the year (winter = February, spring = May, summer = July, autumn = October), with activities designed to stimulate discussion about forest biodiversity (but not well-being specifically). Participants took part in a 1-hour scavenger hunt in situ, in which they were asked to record what they noticed in terms of, for example, smells, colours, textures and sounds. We asked participants to focus on biotic attributes (for example, biodiversity) rather than anthropogenic (for example, pathways). While participants undertook these activities alone, they were then divided into small groups to discuss their impressions of forest biodiversity together. We encouraged participants to expand on why they noticed certain traits. The conversation topics raised by participants expanded upon their experiences outside of the workshops (for example, memories of childhood, at home or on holiday), thus widening the diversity of species referenced. On the second day, participants undertook multiple image-based Q methodology activities13,15 designed to stimulate further discussion about species traits and preferences ex situ, followed by discussions about different trait types.

Qualitative analyses

All audio-recorded activities were transcribed and imported into NVivo (Version 12, QSR International). We then coded where participants had discussed specific trait types (for example, texture, colour, behaviour, smells or sounds; Table 2). We also coded where participants self-reported positive or negative sentiments, as well as discussed benefits/disbenefits to their well-being following the biopsychosocial–spiritual model of health (Table 1). This model accounts for five different domains of human well-being (physical, emotional, cognitive, social and spiritual), which can be both positive and negative (Supplementary Table 2)18,19,34. References to ‘global’ well-being were also reported (sense of overall health/well-being, or lack thereof)35,65 and classed as an additional code34.

Identifying species’ effect traits

For each of the trait types, we extracted all relevant references from NVivo. For each reference, we then identified the species to which the participant was referring (for example, the specific species shown in the Q-methodology pictures13 or a species named by the participant). We then identified the particular effect trait (for example, “slimy”) mentioned in relation to each species, using the participants’ own words. We disregarded references to abiotic factors such as running water, rain or wind, but retained statements when abiotic factors were related to living things (for example, “wind in the trees”). Traits were aggregated when the terms had alternative endings (for example, the sound “screaming” contained “scream”, “screams” and “screaming”) or were synonymous (for example, the texture “gnarly” contained “gnarly”, “gnarled” and “twisted”) to create a standardized final list of traits. This final list was agreed upon by all co-authors. Using the well-being codes, we were able to show whether traits had been linked to a particular valence (positive or negative) and type of well-being (physical, emotional, cognitive, social, spiritual or global). We then cross-referenced when traits and species were spoken about in relation to valence and well-being codes, creating a data matrix of binary responses (1 or 0 for each well-being type) for each mention of a species’ effect trait. If two participants made different comments about the same species’ effect trait eliciting the same type of well-being, this would aggregate to two incidences. For analyses, the matrix was formatted to display species–well-being combinations as rows, with the corresponding effect traits as column headers, populated by values that represented the cumulative number of incidences across participants.

Species mentioned by participants that did not occur in British forests were removed from the data (for example, locusts Schistocerca gregaria and monkey puzzle tree Araucaria araucana). When participants only described particular phenological elements (for example, acorns), we made inferences about the associated species name (for example, acorns were listed as English oak Quercus robur). When participants made general references to a collective group of organisms, we consulted reputable sources (Supplementary Table 4) to derive a list of species with that trait (for example, participants noticed the trait “spots” on “birds”, from which we generated a list of 13 species of British forest birds that had spots). When deriving this extended list, we excluded species that were rare occurrences, accidental records, passage or scarce visitors. We did not generate lists of species for traits that were too generic across an entire taxonomic group (for example, trees that were “green”).

Statistical analyses

All statistics were conducted in R (version 4.3.0, https://www.r-project.org/). To quantify the variety of effect traits for each trait type, we summed the number of unique effect traits for each trait type, then calculated proportions using the total number of unique effect traits. To investigate the shape of the species–trait relationship, we plotted accumulation curves of trait and species richness for each type of well-being (function ‘accumcomp’ in package BiodiversityR66). To visually explore the association between traits and different types of well-being, we used NMDS (function ‘metaMDS’ in package vegan67). An NMDS is an iterative ordination analysis that uses rank orders and can be applied to a variety of data types68. In our case, it enabled a visual interpretation of the relative number of incidences of effect traits linked to different types of well-being in two-dimensional space. We did not transform the data before analysis (recommended for non-ecological data)69, but calculated a matrix of Bray–Curtis dissimilarity coefficients to input into the NMDS. A measure of ‘stress’ was used to determine how well the points in the NMDS are represented across two-dimensional space, determining model fit (the stress in our model was <0.05, indicating very good representation70). We also plotted the species (and taxonomic kingdom) that supported these effect traits, in relation to each well-being type, and examined the approximate directional relationship between the well-being types and the species’ effect traits in k-dimensional space by overlaying the well-being types as vector arrows (function ‘envfit’ in the package vegan67 with 999 permutations; Extended Data Fig. 2).

We used permutational multivariate analysis (ADONIS, function ‘adonis’ in package BiodiversityR66) to investigate predictors (species, taxonomic kingdom and trait type) of the visualized trait–well-being patterns (Fig. 1c). Next, we tested whether differences in the visualized patterns of effect traits between each type of well-being were significantly different, conducting a pairwise permutational multivariate analysis (PERMANOVA, function ‘pairwise.perm.manova’ in the package RVAideMemoire71 with 999 permutations). To quantify the extent of any overlap, we calculated Sørensen’s similarity index72 (function ‘vegdist’ in package vegan67) for the effect traits, as well as species that elicited each pair of well-being types. This index produces continuous values that range from 1 (highly dissimilar) to 0 (very similar).

We identified which effect traits contributed to the dissimilarities identified. We carried out an indicator analysis (function ‘indicators’ in the package Indicspecies73) to determine which species’ effect traits were significantly associated with each type of well-being. This function produces an indicator value that can range between 0 and 1, where 1 represents a circumstance where all mentions of the effect trait are in relation to this well-being type only, and mentions of the effect trait are in every elicitation of this well-being type.

Ethics

Workshop participants provided written informed consent before data collection. Ethics approval was provided by the School of Anthropology and Conservation Research Ethics Committee, University of Kent (ref: 009-ST-19).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.