Exploring Shared Public Perspectives on Biodiversity Attributes

13 1. Researchers, practitioners and policy-makers have widely documented the multifarious 14 ways that nature influences human wellbeing. However, we still have only a limited 15 understanding of how the public interact with, respond to and talk about attributes of 16 biodiversity. 17 2. We used image-based Q methodology to explore the shared and contrasting perspectives 18 people hold for biodiversity. This approach is a powerful way of allowing people to 19 articulate what is, or is not, important to them, free from constraints associated with 20 statement-based stimuli. 21 3. We used British woodlands as a study system, as they are accessible and well-visited by 22 the public. The elements of biodiversity incorporated in the Q methodology represented 23 vertebrates, invertebrates, trees, and understorey plants and fungi. 24 4. The shared public perspectives varied, and the reasons underpinning those perspectives 25 were rich and diverse. People articulated reasons related to an array of biodiversity 26 attributes (e.g. functions, behaviours, colours, smells, shapes). Many of the perspectives 27 transcended specific species or taxonomic groups. 28 5. Even though woodlands were used as a study system, people reference perceptions and 29 experiences external to this habitat (e.g. within their gardens) and associated with their 30 everyday lives. Cultural influences and memories linked to particular people and places 31 were also prominent. 6. Few of the shared perspectives map onto the objective measures and dimensions that 33 researchers use to describe and categorise biodiversity (e.g. rarity, ecosystem service 34 provision).

The Q set needs to comprise an assortment of heterogeneous items that reflect variety in the 152 subject matter and are communicable (Watts & Stenner, 2005). In this case, the research team 153 colours, morphologies, textures, sounds, smells, behaviours), functions (e.g. food provision, 156 pollination) and cultural significance (e.g. folklore, popular media, symbolic). Due to the high 157 levels of biodiversity in woodland, we created four broad Q sets to make the preference task 158 manageable: vertebrates (n = 32 images), invertebrates (n = 43), trees (n = 32), and understorey 159 plants and fungi (n = 32). The invertebrate Q set was larger to account for the greater diversity of 160 species. The multiple Q sets also allowed us to compare and contrast shared preferences for 161 attributes (i.e. traits, functions and cultural significance) across the broad taxonomic groupings. 162

163
The Q set images were all illustrations from identification guides, presented against a white 164 background to minimise the influence of context and artistic style (Fig.1). They were presented 165 on A5 cards, each having a unique number within the Q set (i.e. 1 to 32, or 1 to 43). Throughout 166 the study, the researchers referred to image numbers rather than species names. Participants 167 could then discuss images without needing to identify or name the species, facilitating a focus on 168 attributes. 169 170

Data collection 171
Data collection in Q methodology, called the Q sort, is performed in two stages: a sorting task, 172 followed by a post-sort interview (Watts & Stenner, 2012). Each participant was asked to 8 complete a Q sort for all four of the Q sets, the order of which was randomised to limit the 174 possibility for bias due to fatigue. For each Q sort, participants were given ten minutes to rank the 175 Q set images, guided by the following instructions: "We ask that you choose the pictures that 176 represent what you would most want to encounter or come across in woodlands in England,177 Scotland Participants recorded their preference rankings on an answer sheet representing a quasi-normal 189 distribution grid (Brown, 1980) (Fig. 2). The image number of the species that the participant 190 would most prefer to encounter was written in the top square box of the grid, through to least 191 prefer to encounter in the bottom square box. Participants were told that each image number 192 should only appear once throughout the grid, and that only a single image number should be 193 placed in each of the square boxes (known as a forced-choice distribution). The single rectangular 194 box in the middle of the distribution is therefore for the images where preferences are weakest 195 (Watts & Stenner, 2012 All Q sort answer sheets were filled in by participants, so there was no attrition. However, some 221 participants incorrectly completed their answer sheets by, for example, using a single image 222 number in more than one box in the grid. These answer sheets were excluded from the analyses, 223 resulting in a total of 158 (vertebrates, n = 42; invertebrates, n = 36; trees, n = 38; understorey 224 plants and fungi, n = 42) taken forward for analysis. The four Q sets were analysed independently 225 of each other, providing separate sets of results. 226 227 For each Q set, quantitative multivariate data reduction techniques were used to identify shared 228 perspectives from across the individual views expressed (Zabala & Pascual, 2016). The ranking 229 data from the Q sort answer sheets were intercorrelated and factor analysed using the package 230 qmethod (Zabala, 2014) in R (v3.6.0) (R Core Team, 2019). Principal component analysis (PCA) 231 and varimax rotation (Ramlo, 2016) reduced the multivariate dataset to dimensions, known as 232 'factors' in Q methodology (Watts & Stenner, 2012). We applied the criteria that factors needed to 233 have an eigenvalue greater than 1, and that two or more participants must load onto a factor to 234 represent a shared perspective (Sandbrook et al., 2013). Some participants did not load on any 235 factor, and no participant loaded on more than one factor, signifying that the views captured by 236 each factor were distinct from those captured by other factors (Watts & Stenner, 2012). 237 238 Factors are interpreted by studying factor arrays (a hypothetical Q sort for each factor, formed by 239 calculating the scores for each image; Zabala 2014) (Tables S1 to S4) alongside the corresponding 240 transcriptions. Once the minimum quantitative criteria for factors were met (i.e. eigenvalue >1, two 241 or more participant on each factor), the transcripts of all participants loading onto each factor were 242 extracted and coded using NVivo (Version 12) (QSR Intl Pty Ltd., 2018). This qualitative data was 243 evaluated alongside the scores for each image in that factor array (Tables S1 to S4) to identify 244 shared perspectives, an iterative process of factor reduction until the factors could be interpreted 245 as qualitatively different from one another. Defining factors in a purely statistical way can result in 246 multiple factors that are very similar in terms of participant perspectives that underpin them. Too 247 many factors meant that similar perspectives appeared on more than one factor, while too few 248 meant that nuances were lost. While strength of preference is reflected by those images placed in 249 boxes at the extreme ends of the forced-choice distribution, the location of all images can be useful 250 in interpreting the factor array (Watts & Stenner, 2012). Quantitative analysis provides factors that 251 group participants who ranked images in a similar way, while qualitative analysis allowed us to 252 determine the shared perspectives on biodiversity attributes rather than species per se. As well as 253 elucidating which attributes the participants focused on, we explored how they related to those 254 attributes, which could be positively or negatively. Our approach to coding and qualitative analysis 255 follows the logic of Interpretative Phenomenological Analysis (IPA). It assumes in this context that 256 encounters with nature are a shared phenomenological experience that can be understood in 257 analytical terms through the building up of codes inductively, set within a recognition that the data 258 collected nonetheless reflects and follows a general schedule of questions and stimuli created by 259 the project team. Factors were named according to the shared perspectives associated with the 260 factors, in an analogous way to Q sorts that use statements (e.g. Guenat Sandbrook et al., 2013). Given that the focus of this study was biodiversity attributes, the 262 names thus reflect the shared perspectives for the attributes being discussed. For instance, for 263 factor 1 for Q set 'Trees', discussions centred on memories in participants' childhoods, so the factor 264 was named "childhood memories" (Table 1) (Table 1). Not only did the number of factors extracted differ across image sets, 271 but attributes that defined the factors also differed (Table 1). Our findings showed that while there 272 was commonality in how people interact with biodiversity and its attributes, how and why people 273 related to those attributes varied. Participants discussed the images in a multitude of ways, 274 drawing on regular encounters, memories, associations, responses to particular behaviours and 275 other sensory information (e.g. sounds, smells). The mechanisms by which participants 276 expressed their perspectives revealed a wide use of allegory beyond generic terms, such as 'nice' 277 and 'horrible'. Below, we discuss the results for each of the four image Q sets, before comparing 278 and contrasting shared preferences for attributes across the broad taxonomic groupings.   For the vertebrate Q set, 30 participants loaded on five factors (Table 1). There was a general 291 preference for encountering mammals and larger birds (e.g. tawny owl, goshawk), rather than 292 songbirds. No factor reflected shared positive perspectives for amphibians or reptiles. 293 Paradoxically, familiarity with the more common animals, especially those that participants 294 encountered in their everyday lives (e.g. brown rats, grey squirrels, blackbirds) made them both 295 the most, and least, preferred to encounter. 296 297

Appearance and behaviour in the wild 298
This factor related to shared perspectives about behaviour, especially movement, coupled with 299 appreciation for aesthetics, but not directly linked to specific attributes. These attributes were 300 viewed positively for larger birds and mammals, but negatively for reptiles, rodents and bats 301 (Table S1). Example observations include: "… the owl. So beautiful. It's so graceful…I love their 302 faces, their feathers, they're just gorgeous" and from another participant: "It's a nice activity to go 303 and see the deer. They're quite big so they're more of an animal you can interact with. The baby 304 ones look a bit like Bambi". Some people had encountered the vertebrates as roadkill or in 305 zoological collections, and expressed the desire for, and the importance of, encounters with living 306 animals in the wild. Two comments illustrate this: "I picked the badger because I've never seen 307 one alive…I also put deer because I've never seen one in the wild" and: "if you manage to see a 308 deer in the wild it feels like, 'I was lucky today, I saw a deer'". Some behaviours were appealing, 309 such as "squirrels…scampering up the trees and things, quite interesting to watch", while others 310 were not, for example: "any mouse, anything that scutters quickly, moves quickly, I don't like any 311 of those" or: "I'm scared of snakes. I don't really get how they move. They kind of confuse me. 312 They creep up on you". Preference for encountering was also linked to how elusive an animal 313 was perceived to be (e.g. deer, owls, badgers, dormice, treecreepers). Interestingly, negative 314 perspectives, such as: "I'm not really a fan of the slimy snake" stemmed from perceptions of 315 appearance and behaviour and were not borne of experience. 316 317

Encounters 318
Perspectives underpinning this factor were related to participants' imagined and previous 319 encounters. Preferences were for encountering mammals and birds perceived as "beautiful" and 320 "smart", and framed using favourable traits, such as the "majestic owl", "harmless deer" and 321 While some welcomed an encounter with such mammals (e.g. "I really like foxes. We feed them. 361 They come into our garden. They're a delight, a pleasure") others did not. One participant who 362 loaded negatively on this factor disliked foxes because they encountered them regularly in their 363 urban environment. Furthermore, this same participant drew a distinction between foxes that lived 364 in different habitats: "urban foxes. They're not proper like bushy majestic foxes. They're just like 365 little crap ones". 366 367

Captivating 368
In this factor, participants again expressed a preference for encountering mammals, but here they 369 expressed a disinterest in smaller birds. Woodland mammals, especially the pine marten, and 370 larger birds, were described as expressing behaviours that would be exciting to watch, especially 371 if for the first time. Illustrative comments include: "Foxes are really cute and really smart and just 372 loads of fun to watch" or: "It's just a cool looking critter. I've never seen one before" and: "It would 373 be quite interesting to watch. I've seen so many squirrels so something new would be good". Four factors were extracted for the invertebrate image set, onto which 31 participants loaded 384 (Table 1). In three of these four factors, the bumblebee and butterflies were preferable to 385 encounter, but for different reasons (Table S2). How humans interact with, benefit from, and are 386 affected by invertebrate behaviour formed the basis of discussion. Lockwood (2013) notes that 387 insects both frighten and mesmerise humans, and our paradoxical relationship with them is 388 steeped in evolutionary history. Discussion around colour was common to all four of the 389 invertebrate factors, especially in relation to butterflies, ladybirds, and bumblebees. Colour can 390 be used as a defence against predation (aposematism), which humans also notice. While several 391 of the invertebrates in the Q set had black and yellow striations, participants focused primarily on 392 other attributes that are more culturally ingrained. For instance, people described bees favourably 393 due to their ability to deliver valuable ecosystem services, whereas wasps were stereotyped as 394 dangerous (Sumner et al., 2018). 395 396

Purpose, function and benefit to humans 397
The common thread in this factor was the purpose and function of invertebrates, to both the 398 environment and to humans. A particular focus was pollination. The bumblebee was most 399 preferred (Table S2) and was frequently associated with anxieties about environmental change: 400 "They're productive, they're endangered…and they're a good insect to have in the world", and: 401 "bees have been in decline so when you start seeing them you feel…a bit hopeful". Pollination 402 was only attributed to bees, despite other pollinators in the image set (e.g. hoverflies, butterflies, 403 wasps). Bumblebees were also considered important as some believed that they produce honey, 404 despite that only being produced by honeybees. Worms, ladybirds and spiders were perceived

Harmless 420
Participants expressed preferences to encounter invertebrates that they perceived as harmless 421 (Table S2). Attributes such as movement, colour (e.g. "I am amazed by butterflies…I love the 422 beauty of them, the colour"), and positive childhood memories (e.g. "I always relate [butterflies] 423 to childhood, to going into fields and seeing them all flying around") were viewed positively. 424 However, the shared perspective was the perception that encounters would be "safe", "harmless", 425 "gentle", and "calming". One participant described their selection as: "some sort of ladybird…it's 426 a very safe, gentle creature to come across". Least preferred to encounter were attributes 427 considered to be harmful (e.g. flying, biting, moving), as in: "the beetles and wasp…because I 428  (Table S2), regarded as: "shiny so it seems slimy and venomous, even 433 though it's not but it seems it", demonstrating that knowledge of attributes (e.g. harmless) does 434 not necessarily counteract judgements based on visual cues. 435 436

Encounter of spiders 437
These participants form a separate factor due to a common disdain for encountering spiders, 438 irrespective of the species (Table S2). Participants shared an aversion to a range of spider 439 attributes including: "lots of legs, they move very fast and I just imagine them crawling on me" 440 and: "the way they move and look and everything". They also alluded to cobwebs: "I just really 441 don't like the idea of the webs…I find them sort of dirty, just dusty really". Spiders were described 442 as being "unpredictable", "repulsive", and something to "fear". This is similar to previous findings 443 20 suggesting that movement and appearance are the attributes mainly influencing the fear of 444 spiders (Lindner et al., 2019). 445 446

Curiosity 447
These participants preferred to encounter "unique" invertebrates that piqued their curiosity as 448 opposed to "mundane" ones regularly encountered. Each individual associated their curiosity with 449 childhood memories as exemplified by: "when I was a kid you'd hear the noise and you'd try to 450 track them down in the grass and then try to catch them" and: "It reminds me of my childhood 451 when I was grabbing at stuff". Curiosity related to different attributes, but most often movement, 452 for instance: "it amazes me how they can just move all these legs" and: "snails just crawl up the 453 wall". Interestingly, this was the only time that the attributes at different stages of a butterfly's life 454 cycle were mentioned: "when they're caterpillars they can be destructive…but when they're 455 butterflies they're really pretty". 456 457

Trees 458
Thirty participants loaded onto five factors. The English oak was prominent in the discussions 459 (Table S3) and celebrated for being "majestic", "quintessential" and "classic". Indeed, it is known 460 to symbolise traits such as longevity, cohesiveness and robustness in the Northern Hemisphere 461 (Leroy et al., 2020). While this reverence for oak trees was common, different factor arrays 462 represent diverse perspectives, and is an example of the plurality of attention paid to the different 463 attributes of the same species. Trees can be long lived and individuals are static. They display 464 annual cycles that are predictable across seasons affecting appearance and behaviour (Zhao et  Participants spoke about a variety of attributes, linking them to positive childhood memories 472 (Table S3). There was a sense of trees being part of a wider socialisation and learning process: 473 "the oak tree…that's the first tree as a child you learn the name of, it's easy to spell, it's the first 474 one you pick up on". Most memories were interactive, including imagination: "I like acorns…when 475 I was little I can remember taking them out and pretending to be teacups", play: "just childhood 476 memories, playing conkers" and food: "when we were kids we used to pick elderberries". Trees 477 considered "climbable" were also important: "I like the big slow-growing leafy trees…that evoke 478 childhood memories". This behaviour in humans is thought to be linked to the rewards associated 479 with arboreal foraging (Kraft et al., 2014). Additionally, individual attributes of trees (e.g. "sticky 480 buds", "conkers", "pretty flowers and berries") had childhood associations. Active interactions with 481 plants as a child has shown to be associated with positive values about trees as an adult (Lohr & 482 Pearson-Mims, 2005). 483 484

Size/shape and resource provision 485
The size, shape and provision of resources associated with longer lived trees were the common 486 perspectives in this factor. Preference was to encounter deciduous rather than coniferous trees 487 (Table S3), except for the yew which one participant described as: "ancient and old…quite a 488 poetic sort of tree, evocative". The oak was the symbolic focus of such perspectives, described 489 in terms of being a: "big, majestic tree. Slow growing…it's like the king of the woods" and: "it 490 seems such a symbol of history". The physical qualities of trees are part of how people construct 491 meaning and understanding of the natural environment, with non-utilitarian perspectives 492 reflecting that trees and woodlands resonate with people culturally at personal and community 493 levels (Henwood & Pidgeon, 2001). Shape was also considered important in terms of a tree being 494 suitable to climb: "they've got character, you can climb them, so it's more than just a tree standing 495 there", whereas evergreens were considered: "boring to look at…very kind of straight…not very 496 branching", as well as: "the prickly ones…you can't climb them". There is evidence that 497 preference for deciduous trees over conifers is not only linked to size and shape of a tree, but 498 also the ratio of the crown size compared to trunk height (Gerstenberg & Hofmann, 2016). Mature 499 22 trees were also associated with provision of resources that benefit both humans (e.g. "eating 500 chestnuts", "made elderflower cordial", "conkers for games", "cones for decoration"), as well as 501 wildlife: "they shelter a lot, they feed a lot of things" and: "it can feed animals and birds". 502 expectations of what a woodland should look like, for example a mix in height, shape, colour and 506 seasonal variation (Table S3). Perspectives around seasonality included: "holly…you'd like to see 507 something evergreen, and the berries and the white flowers that come out before…it is one of the 508 nicest trees in the woodland in the winter time" and: "I like conker trees, sycamore trees, things 509 that drop stuff… it reminds you of the time of the year". Larger trees were again preferable to 510 encounter, illustrated by comments such as "I prefer the bigger, more sort of standout'ish trees" 511 and: "they've obviously taken a long time to grow that tall, are quite majestic. Like a proper tree". 512 This perspective was complemented by trees that were "uninteresting" or "weedy" being the least 513 preferred to encounter: "I don't know what it is but it looks just insipid and useless…it just looks 514 like nothing in particular" and: "branches coming out from the floor…makes it less inviting to go 515 into that part of the woods". Humans have been shown to prefer forests with mixed composition 516 and tree heights over uniformity (Filyushkina et al., 2017). Here, participants' preferences to 517 encounter were also linked to variety occurring due to seasonal change. 518 519

Flowers, berries, leaves and cones 520
Within this factor, preference was for encountering attributes associated with flowers, leaves, 521 berries, cones, without reference to the structure of the tree itself. The colours, shapes and 522 textures of flowers, leaves, berries and cones were appealing, exemplified by comments such as: 523 "I like pine trees because I love cones" and: "the colours and the ones that you see flowers and 524 blossom on, and fruit and berries and things". These perspectives were predominantly positive: 525 "seems brighter and a bigger variety than just all green all the time" but not always: "don't like the 526 horse chestnut because …there are sticky buds, they get all over my shoes and it drives me 527 crazy". Furthermore, some of these attributes were viewed positively as they were likely to attract 528 animals (e.g. birds) that participants wished to encounter. 529 530

General likes and dislikes 531
Participants expressed generalised views on the appeal of trees. However, when prompted to 532 elaborate, they were not able to verbalise their perspective in relation to specific attributes. 533 Statements such as: "I just liked it, I don't know why" and: "I just like it. The branches, the leaves, 534 really just catching to the eye", showed a general appreciation. Furthermore, comments such as: 535 "Just really not nice to look at really. Nothing about them" and: "Not as aesthetically pleasing but 536 quite nice", provide an indication of perspectives towards trees that rely on a generic 'feeling' 537 rather than relating to specifics. 538 539

Understorey plants and fungi 540
For the understorey plants and fungi, 38 participants loaded onto the three factors (Table 1). 541 There was a strong focus on visual attributes, especially colour, but perspectives included 542 attributes affecting other human senses, such as smell and taste. Each factor showed a 543 preference for encountering flowering plants over grasses and mushrooms, especially for flowers 544 perceived and described as 'colourful' (Table S4). The colour of objects in the natural environment 545 serve various functions, such as indicating that a fruit is ripe, for camouflage or for attracting 546 pollinators (Marshall, 2010). Colour is also linked to emotions in humans, linked to both single 547 colours and colour combinations (see Ou et al., 2004). Colour is not a physical property of an 548 object, but a psychological property that is unique to that observer (Palmer, 1999 There were shared preferences to encounter spring and summer flowering plants and not to 554 encounter fungi (Table S4). Many participants related positively to "colourful flowers" being 555 aesthetically pleasing, commenting: "it's always the pinks and purples that seem to jump out to 556 me" and: "they're just pretty aren't they". Preferences were also framed in terms of the properties 557 of colour, distinguishing between those considered "bright" and "vibrant", or "plain" and "boring", 558 rather than the colours themselves. Green and brown flora were described as lacking colour and 559 therefore considered less interesting: "they're just green. There is no attraction, nothing to them". 560 Fungi were the least preferred to encounter: "mushrooms that are kind of beiges and browns are 561 not very attractive". The fly agaric toadstool caught people's attention due to being bright red, as 562 well as its association with popular culture and mythologies (e.g. "fairies", "Disney", "Super 563 Mario"). Participants also believed that "drab" fungi looked dangerous yet, paradoxically, these 564 were mainly edible while the fly agaric is highly toxic. 565 566

Colourful and complexity 567
Within this factor, discussion was around both colour and complexity (Table S4). Bright, vibrant 568 colours made some flowers and mushrooms "eye-catching" and provided variety. In addition to 569 vibrant colours, participants noted the importance of shape: "mushrooms are just fantastic 570 shapes. Some are really gnarly and some are very smooth tops" and structure: "It was the most 571 complex… the bright colours and the fact that there is a lot more going on". Least preferable to 572 encounter were the "plain green" grasses and ferns that lacked structure and variety in colour, 573 described as: "nondescript…the leaves are not really an interesting shape, they're just like a child 574 draws a leaf" and: "disinteresting and more weed-like". These perspectives were linked to 575 expectations of the natural environment: "you want to see sort of how the plants are naturally 576 competing with each other", and the desire to see a variety of colours and shapes that would add 577 interest when experiencing woodland. 578 579

Appealing to the senses 580
Sensory interactions with species were the subject of this factor, particularly smell, texture and 581 edibility (Table S4). These were both positive: "if it smells of garlic then I think that's a good thing" 582 or: "you can eat it which is always a good thing about it", and negative: "doesn't feel nice to touch. 583 It's sticky". Reference to touch comprised intrigue as to how the organism would feel or react: "I 584 like that fern…when you touch it, it curls back up on itself", as well as avoidance of perceived 585 threats: "I think any fungi that grows off tree is dangerous, you shouldn't touch it". Participants 586 also spoke about the provisioning services of plants and fungi. "Blackberry picking" was frequently 587 articulated in terms of being a happy childhood activity. 588 589 3.5 Shared perspectives across Q sets 590 The four Q sets provided an insight into shared perspectives across a range of attributes. 591 However, the participants did not articulate or respond to the attributes uniformly across the four 592 broad taxonomic groups. Some overlap did exist, for example regarding how vertebrates and 593 trees characterise woodland, but the shared perspectives mainly focused on divergent 594 biodiversity attributes. For instance, the discussions for understorey plants and fungi were heavily 595 focused on colour, yet this attribute received little attention in other Q sets. Likewise, the 596 importance placed on the purpose and function of invertebrates was not apparent for the other Q 597 sets. The same attributes of biodiversity are therefore not consistently related to or talked about 598 by the public, across broad taxonomic groups. 599 600 While the Q sorts were a highly visual activity, the discussions captured a diversity of emotions, 601 anthropomorphisms and associations. Notable cross-cutting perspectives emerged from the 602 data. Participants spoke of associations that transcended physical properties of attributes, linking 603 perspectives with culture. These were in association with a variety of cultural influences, such as 604 literature: "reminds me of stories, like The Animals of Farthing Wood", cinema: "deer…they're just 605 really, really nice…I remember crying my eyes out when Bambi's Mum got shot", and gaming: 606 "my favourite just because it was channelling some Super Mario vibes". Notably, fairies were 607 strongly associated with woodland attributes: "I just imagine all the fairies that dance round them". 608

609
Memories and reminders featured strongly in a number of ways, for example: "I really like that 610 plant because it reminds me of things I did in childhood" or: "violets…They remind me of 611