Abstract

Policies prioritising the repurposing of abandoned land (‘brownfield sites’), in order to spare undeveloped sites, are widely perceived to limit damage to biodiversity. However, brownfield sites can be of significant ecological value, providing scarce habitats, including as a source of greenspace in urbanised areas, and promoting ecological richness at landscape scales. Therefore, policies to promote brownfield site repurposing might do more harm than good in terms of preserving biodiversity, calling into question what the public understand by, and value in terms of, brownfield sites.

To examine public opinions relating to the prioritisation of biodiversity in the selection of sites for development, we undertook a survey of 2,247 respondents in the UK.

We found that a stigma exists against brownfield sites. Their ecological value is underappreciated, but those who associated brownfield sites with wildlife-related keywords viewed them more positively. Less importance was placed on a site’s history of development (brownfield vs undeveloped) than on its actual value to wildlife when considering whether it should be repurposed, and to what use; repurposing to housing or renewable energy production were favoured when sites had little value to wildlife, regardless of whether they had previously been developed.

Overall, our results suggest that to best serve the preferences of local people, future planning decisions should consider the current biodiversity value of a site alongside its status as a brownfield site or otherwise. The existing stigma against brownfield sites could be alleviated by outreach and education about the value of such sites to nature.

Introduction

A key goal driving the redevelopment of ‘brownfield’ sites (here defined as abandoned land that has previously been developed), particularly for housing, is sustainability (Schulze Bäing & Wong, 2012). The repurposing of land already diverted from an assumed ‘natural’, or at least rural, state, is perceived to provide the social and economic benefit of reducing the blight of derelict land (Rizzo et al., 2015), whilst conferring the environmental benefit of preserving undeveloped land. For example, policies promoting brownfield regeneration may intend to relieve pressure to develop rural land (Dixon, 2006; Loures & Vaz, 2018), especially within ‘green belts’ immediately surrounding urban areas (Schulze Bäing, 2010). Besides housing, brownfield sites are often put forward for use in renewable energy production (Milbrandt et al., 2014) or can be brought into agricultural production (Hard et al., 2019). There are therefore multiple pressures for changes to land use with strong social and ostensibly, environmental, purposes. The potential benefits of bringing brownfield land back into use have been explored by policymakers internationally (for example, EU, Morar et al., 2021; USA, De Sousa et al., 2023; China, Gao, 2023). In the UK, the drivers for brownfield development are highlighted by changes to planning policy in 2020, promising ‘fast track’ planning permission for housing on brownfield sites (Town and Country Planning (General Permitted Development) (England) (Amendment) (No. 3) Order 2020).

Policies favouring brownfield development have inherent assumptions about the relative environmental value of different types of land-use. These are embedded into the planning system in the UK (Cox & Rodway-Dyer, 2023). However, the environmental goals of sustainable development can be at odds with brownfield regeneration (Dixon, 2006). Brownfield sites can have high ecological value, providing habitat for rare or threatened species (Eyre et al., 2003; Tropek et al., 2010; Kamp et al., 2015; Broughton et al., 2021), sometimes including water bodies (Preston et al., 2023), and the presence of such sites can enhance biodiversity at landscape scales (Macgregor et al., 2022). They can generate ecosystem services (Gardiner et al., 2013; Mathey et al., 2015), including cultural ecosystem services through provision of urban greenspace (Pueffel et al., 2018; De Valck et al., 2019; Masood & Russo, 2023). Concerns have therefore been raised that development on brownfield sites may have unintended negative consequences for biodiversity (Meehan et al., 2010; Fletcher et al., 2011; Broughton et al., 2021; Macgregor et al., 2022; Cox & Rodway-Dyer, 2023), and that rigid protection of green belts, in particular, can divert development onto brownfield sites that are ecologically more valuable (Amati & Yokohari, 2006). Given the biodiversity value of some brownfield sites, their protection and management as publicly accessible nature reserves could provide an alternative form of repurposing to redevelopment (Heatherington et al., 2019), which may be well received by local communities (Masood & Russo, 2023); the provision of public access in itself can improve locals’ perception of such sites (Hofmann et al., 2012). Fast track approval carries the implication for reducing possibilities for local objections to planning applications; but consultation with communities around developments that relate to them is itself a sustainability goal.

Given the ongoing international drive to prioritise development on brownfield sites, especially linked to renewable energy production (Milbrandt et al., 2014) and housing (Schulze Bäing & Wong, 2012), it is important to consider local people’s own prioritisation of different alternative land uses (Loures & Crawford, 2008), as well as the extent to which these are impacted by site type as a brownfield site or otherwise, and by site biodiversity value. Policies promoting the repurposing of brownfield sites often garner public support (Loures et al., 2016), probably due to a combination of idealised perceptions of rural land (Rust et al., 2021) and negative perceptions of brownfield sites and their characteristics (Mathey et al., 2018), although deliberate regeneration of brownfield sites into urban greenspace can also be viewed positively (Masood & Russo, 2023). Whilst the ecological value of brownfields is increasingly well understood by scientists, conservationists, and to some extent planners (Loures & Vaz, 2018; but also see Cox & Rodway-Dyer, 2023), it is unclear to what extent this knowledge has permeated among the general public, or whether increased awareness of brownfield site biodiversity would lead to more positive perceptions of such sites. In particular, it is unclear whether knowledge about the actual biodiversity richness of candidate sites for redevelopment would alter public perceptions about whether sites should be repurposed, and if so, into what use; and whether this knowledge of biodiversity richness carries greater or lesser weight in the public’s eyes than a site’s status as a brownfield site (that is, its history of development). Recent studies have employed questionnaire-based public surveys to understand public preferences for land uses including agriculture (Rust et al., 2021) and urban greenspace (Tomitaka et al., 2022; Masood & Russo, 2023), including in relation to biodiversity conservation (Fischer et al., 2020), but this approach has not yet been applied to brownfield sites, except with restricted locations or contexts (Loures et al., 2016; Masood & Russo, 2023).

In this study, we conducted a survey of a sample of 2,247 adults living in Great Britain (namely, England, Wales, and Scotland; but not Northern Ireland), and analysed responses in order to analyse participants’ opinions about brownfield sites and their repurposing, especially in the light of information about biodiversity value of such sites. Therefore, our study provides the first questionnaire-based assessment at a national (rather than regional or local) scale of public preferences for brownfield site use, incorporating a wide range of options for repurposing including use for nature, housing, industry and renewable energy. We hypothesised that: (i) awareness of the biodiversity value of brownfield sites would be associated with more positive opinions of such sites; (ii) both a site’s status as a brownfield site and its biodiversity value would influence people’s preferences for its repurposing. Under hypothesis ii, we expected redevelopment to be favoured on brownfield sites above undeveloped sites, but management as a nature reserve or remaining abandoned preferred by more people on sites which are richer in wildlife.

Methods

Survey questionnaire

In order to characterise public opinion at a national scale, we needed to access responses from a broad geographic area and across a wide section of the population. Thus, a questionnaire survey provided an efficient means to collect data from sufficient respondents to permit statistical analysis of the results (Rowley, 2014). Further research could adopt a more intensive approach, for example, based on interviews of a much smaller number of respondents in order to better understand their motivations; but at this stage, we sought an overview of public opinion, that is, an extensive approach in the terms of Sayer (2010). Data were gathered by means of an online survey, which was approved by the University of Hull Ethics Committee (reference no. FEC_2021_18). The survey was undertaken by YouGov in order to obtain a high number of responses in a short space of time, and all respondents consented to participation.

The survey was conducted in April 2021, using YouGov’s GB Omnibus survey type (YouGov, 2021). A total of 2,247 unique responses were obtained from adults living in Great Britain, distributed broadly across a range of socio-demographic variables including age, gender, social grade, region, and employment status (Table 1). Socio-demographic information was primarily drawn from data collected by YouGov about respondents during their sign-up process and were supplemented by four further questions covering local area settlement size, duration of occupancy in the settlement, duration of occupancy in the home, and affiliation with relevant stakeholder groups (for example, as employees, volunteers or supporters of conservation organizations).

The survey included 21 questions (Appendix 1, in Zenodo data archive). All questions were framed to prompt respondents to consider their opinions of brownfield or undeveloped sites ‘in their local area’. The majority of questions were either multiple choice (commonly presented on a Likert scale, where respondents rate their agreement/disagreement with a statement on a five-point scale (Likert, 1932)), or arrays in which respondents ranked multiple options from most to least preferred. Some questions allowed open-ended follow-ups for respondents to provide more information about why they had responded in a particular way. The first question was preceded by a clear definition of what was meant by a ‘brownfield site’ in the context of the survey, which read: ‘Brownfield sites are defined as “abandoned land that has previously been developed”. They can include former mining or quarrying works, disused industrial sites, and closed landfill sites. In the UK, brownfield sites are prioritised for redevelopment: the government set a policy target in 2008 that over 60 per cent of newbuild housing should be built on brownfield sites, and this target is currently being exceeded’. Respondents were asked questions relating to:

• Whether they associated brownfield sites with a range of keywords which were expected to have positive, negative or ambiguous connotations (including keywords associated with biodiversity).
• Whether their opinion about brownfield sites was positive or negative (and to explain their choice in their own words), and whether that opinion would change upon provision of information about wildlife on said sites.
• What their opinion was of photographs of semi-natural habitats on four unspecified sites in rural locations (Figure 1 (a-d)), and whether they thought each photograph showed a brownfield site or an undeveloped site (in fact, the photographs depicted two brownfield and two undeveloped sites, all of which are currently managed as nature reserves by the Yorkshire Wildlife Trust, but respondents were not informed of this; see below).
• Whether they agreed or disagreed with a range of key phrases, variously with positive or negative connotations, when shown a photograph of:
  • a brownfield site;
  • a nature reserve;
  • an arable farm; and
  • a renewable energy facility (solar farm) respectively (Figure 1 (e-h)).
• What their preferences (ranked from most preferred to least preferred) were for a variety of options for repurposing of sites, for each of four hypothetical sites which were respectively described as:
  • a brownfield site, with little value to wildlife;
  • a brownfield site, rich in wildlife;
  • a previously undeveloped site, with little value to wildlife; and
  • a previously undeveloped site, rich in wildlife’. A final open question allowed respondents to provide any other opinions or comments that they felt were relevant.

For the questions in which respondents were asked to give their opinion of photographs, we intended to show respondents four similar photographs of similar sites, in order that their opinion of the site was as likely to be influenced by any biases relating to brownfield sites as the actual character of the sites in question, or of the photographs of those sites. To this end, we selected four sites that were visually similar, all of which comprised areas of semi-natural habitat managed by the same nature conservation organisation. Two out of the four sites were brownfield sites (a former quarry and a former landfill site), whilst the other two had no prior history of industrial use. Photographs were taken of each site (Figure 1a–d) that were deliberately intended to be compositionally similar, with open areas in the foreground, some trees in the background, and an approximately equal proportion of the photograph consisting of sky. Not all photographs were taken at the same time of year, but all photographs were taken in relatively bright weather conditions. As far as possible, we avoided showing man-made features in any photograph as we felt respondents might naturally assume the site in question was a brownfield site based on this feature, without assessing the holistic character of the site; however, the photograph of site ‘Brownfield_1’ did include some farm buildings located outside the site boundary. Photographs shown to respondents in connection with their response to a selection of key phrases (Figure 1e–h) were purely illustrative, and intended to show features of a ‘typical’ site in each category, including: unmanaged semi-natural habitat and some man-made structures (brownfield site, Figure 1e), managed wetland habitat (nature reserve, Figure 1f), crops (arable farm, Figure 1g), and an array of solar panels (renewable energy facility, Figure 1h).

Table 1: Breakdown of survey respondents (out of 2247 total) across various socio-demographic variables

Note: Breaks may not total to 2247 in cases where respondents could belong to more than one category simultaneously, or where some respondents did not provide answers.

Statistical analysis

Statistical analyses were conducted in R version 4.0.3 (R Core Team, 2024), using a range of packages of which the most important were lme4 (Bates et al., 2015) for constructing and testing generalized linear mixed-effects models (GLMMs), and ggplot2 (Wickham, 2009) for data visualization. R scripts are archived on Zenodo (doi: 10.5281/zenodo.11313478).

To assess whether the two hypotheses described above are supported, we interrogated survey responses with a series of questions, combining responses to multiple survey questions where appropriate, as follows:

1. What keywords did respondents associate most with brownfield sites?
2. What opinion did respondents have of the idea of a brownfield site in their local area, did opinions differ between different socio-demographic categories, and did opinions correspond to the connotations of respondents’ chosen keywords?
3. Would provision of information about wildlife on a brownfield site affect respondents’ opinion of that site?
4. Did respondents give more positive opinions about photographs of undeveloped sites than photographs of brownfield sites?
5. Did respondents give more positive opinions about photographs of sites when they believed them to be undeveloped sites?
6. Did respondents agree with key phrases (describing potential benefits/disbenefits of sites) more or less when applied to brownfield sites than to other types of site?
7. Did respondents prefer certain options for repurposing sites to others, and were their preferences affected by (a) site type (‘brownfield’ vs ‘undeveloped’) and (b) wildlife value (‘little value to wildlife’ vs ‘rich in wildlife’)?

In some cases where questions in the original survey were answered on a Likert scale, responses were assessed qualitatively by comparing percentages of respondents giving each possible response (questions 1-3 above). However, in order to facilitate quantitative statistical testing of hypotheses, qualitative answers were simplified to binary formats wherever appropriate, representing either ‘agree/disagree’, or ‘agree/do not agree’ in cases where a neutral response was available (in the latter case, ‘neutral’ was considered to indicate an implicit lack of agreement). Additionally, we generated a ‘keyword score’ for each respondent, using respondents’ answers to a question in which they selected, from a list of provided keywords, which terms they associated with brownfield sites. Specifically, we assigned each keyword to have either clear positive connotations (n = 4, for example, ‘Birdsong’), clear negative connotations (n = 4, for example, ‘Pollution’), or ambiguous connotations (n = 3, for example, ‘Boundary fencing’). Positive and negative keywords were assigned weights of +1 and -1 respectively (ambiguous keywords, and additional keywords written in by respondents who had selected ‘other’, were not used in the generation of keyword scores). The keyword score for each respondent was the sum of the weights for all keywords they had selected, and therefore ranged from -4 (respondent selected all available negative and no positive keywords) to 4 (respondent selected all available positive and no negative keywords). We also counted the number of wildlife-related keywords selected by each respondent, out of four in total: (‘Birdsong’, ‘Bees and butterflies’, ‘Wild flowers’, and ‘Ponds/lakes’).

We used a mixture of generalised linear models (GLMs), generalised linear mixed-effects models (GLMMs) and simpler statistical tests to test the questions listed below. For all GLMs and GLMMs, model fit was validated by visual inspection of plotted model residuals.

To test whether respondents’ opinions of brownfield sites were related to a range of socio-demographic variables (question 2; gender, age, social grade, geographic/governmental region, employment status, children living at home, parent/guardianship status, dog/pet ownership, settlement size, duration of settlement tenure, home ownership status, and duration of home tenure), we constructed and tested GLMs with Gaussian error distributions, with each socio-demographic variable in turn as the independent variable and respondents’ opinions on a Likert scale (1–5, with respondents who answered ‘Don’t know’ excluded) as the dependent variable. The significance of each overall model was tested using an F-test, and where models were deemed significant, significance of between-category differences was tested using a post-hoc Tukey test.

To test whether respondents who indicated more positive opinions of brownfield sites also selected more positive keywords and more wildlife-related keywords (question 2), we constructed two GLMs with a Gaussian error distribution, with opinion (on a Likert scale) as the independent variable, and keyword score and number of wildlife-related keywords respectively as dependent variables. The significance of overall models and between-category differences was tested as above.

To investigate whether respondents’ opinions of sites in photographs were affected by actual or perceived site type (questions 4-5), we conducted Pearson’s chi-squared tests. To test the effect of actual site type, we compared the frequency of each response on the Likert scale between photographs of brownfield sites and photographs of undeveloped sites. To test the effect of perceived site type, we split the dataset to separate responses to photographs of brownfield sites and responses to photographs of undeveloped sites, and for each site type, compared the frequency of each response on the Likert scale between cases where the respondent had guessed the photograph depicted a brownfield site, and those where they guessed an undeveloped site was depicted. We additionally tested whether respondents were better than chance at identifying whether sites were brownfields or undeveloped from photographs by conducting exact binomial tests on the proportion of correct guesses; and whether the likelihood of a correct guess was affected by a respondent’s opinion of brownfield sites, by constructing a GLM with a binomial error distribution, with opinion as the independent variable and guess (correct/incorrect) as a binomial dependent variable. Model significance was tested using a Likelihood Ratio Test (LRT).

To test whether respondents associated various key phrases more with some site types than others (question 6), we constructed a GLMM for each key phrase separately. Models had a binomial error distribution, with site type and response (agree/do not agree) as the independent and dependent variable respectively, and respondent identifier as a random effect. In each case, overall model significance was tested using an LRT, and significance of between-category differences was tested using a post-hoc Tukey test.

To investigate whether respondents ranked various options for repurposing sites as more or less preferred (question 7), we constructed and tested a GLMM as above, with land use option (one of eight) and ranking (1-8) as the independent and dependent variable respectively, and respondent identifier as a random effect. To test whether these rankings were affected by site type (‘brownfield’ vs ‘undeveloped’) and site wildlife value (‘little value to wildlife’ vs ‘rich in wildlife’), we conducted a series of tests. For each of the eight repurposing options, we separately tested whether the option’s ranking was affected by site type and by site wildlife value using both a paired t-test and a paired Wilcoxon signed-rank test. We conducted both types of test because the t-tests allowed us to obtain (and compare) effect sizes, but the non-parametric Wilcoxon tests gave us greater certainty about significance (or otherwise) of effects, given that for some repurposing options, the distribution of rankings was non-normal (for example, ‘no change’ had a bimodal distribution of rankings, tending to be ranked as either one of the most preferred options, or one of the least preferred, with respondents rarely placing it as an intermediate option). Nevertheless, both tests produced the same qualitative (significant vs non-significant) outcome in all cases. Rankings were paired both within respondent identifier and within-category for the non-focal variable, such that we compared each respondent’s ranking of the same repurposing option for sites described differently for the focal variable only (for example, ‘a brownfield site, with little value to wildlife’ compared to ‘a brownfield site, rich in wildlife’, and so on).

Results

This section is structured around the seven questions/themes through which we analysed the results.

General views of respondents on brownfield sites (questions 1-3)

Overall, the most selected keywords were ‘Abandoned/derelict buildings’ (selected by 80.9 per cent of respondents), ‘Overgrown vegetation’ (74.6 per cent) and ‘Fly-tipping/litter’ (73.0 per cent), all of which were expected to carry negative connotations. A much higher proportion of respondents selected keywords with negative connotations than positive connotations, whilst ambiguous keywords that could be interpreted as either positive or negative were generally intermediate in frequency of selection (Fig. 2). Respondents who selected ‘other’ described a range of negative features (for example, ‘abandoned machinery and vehicles’, ‘broken ground’, ‘drug use paraphernalia’, and ‘hazardous material – chemicals’), and a smaller number of positive features, which were almost exclusively associated with biodiversity (for example, ‘bats’, ‘invertebrates’, and ‘trees’), and occasionally indicated a significant level of knowledge about natural history (for example, ‘often very rare flora – [especially] bryophytes’).

Overall, the majority of respondents felt either ‘Neutral’ (37.2 per cent of respondents) or ‘Fairly negative’ (35.3 per cent) about the idea of having a brownfield site in their local area (Fig. 3), compared to less than ten per cent who felt either fairly or very positive. When asked why, respondents who felt negatively about brownfield sites cited concerns around their appearance (for example, ‘eyesore’, ‘unsightly’, ‘ugly’; 43 per cent of the 1,094 respondents who felt ‘Fairly negative’ or ‘Very negative’), utility (for example, ‘waste of space’, ‘unused land’, ‘could be redeveloped’; 31 per cent), or facilitation of undesirable activities (for example, ‘crime’, ‘antisocial behaviour’; 18 per cent). Respondents who felt positively about brownfield sites were evenly split between those who cited the biodiversity value of such sites (for example, ‘provides natural habitat’, ‘wildlife’, ‘return to nature’; 27 per cent of the 192 respondents who felt ‘Fairly positive’ or ‘Very positive’) and those who instead viewed such sites as providing opportunities for future development (for example, ‘waste land put to good use’, ‘regeneration’, ‘improvement’; 25 per cent). Biodiversity value was also the most commonly cited factor amongst those who felt ‘Neutral’ (for example, ‘consideration for nature’, ‘wildlife’, ‘good for environment’, ‘nature takes over’; 14 per cent of 831 respondents).

There was some relationship between socio-demographic factors and these opinions, albeit with very small effect sizes. Specifically, those aged 55+ and retired people had slightly more negative opinions about brownfield sites than younger age groups and those in employment (effect of age: F = 4.37, P = 0.0016; effect of employment status: F = 3.19, P = 0.0071), and those in social grade ABC1 (‘middle class’, defined by the occupation of their household’s chief income earner being managerial, administrative or professional) had slightly more negative opinions than those in social grade C2DE (‘working class’, chief income earner’s occupation is skilled or unskilled manual or casual work; effect of social grade: F = 3.89, P = 0.0488). We found no significant effect of self-identifying with any of the key stakeholder groups investigated.

Respondents who had more positive opinions about brownfield sites also tended to select more wildlife-related keywords (F = 34.44, P < 0.0001), whereas respondents who had more negative opinions about brownfield sites selected more keywords with negative connotations (F = 77.37, P < 0.0001; Fig. 4).

The majority of respondents answered that provision of information about wildlife that may be living on a brownfield site would make them feel either ‘a little more positively’ (36.8 per cent of respondents) or ‘much more positively’ (19.2 per cent) about the site (Fig. 5). Respondents’ answers were significantly related to their originally stated opinion on the idea of a local brownfield site (Χ² = 131.92, d.f. = 10, P < 0.0001), with those who had previously expressed a negative opinion of brownfield sites being more likely to select ‘a little more positively’, and those who had previously expressed a positive opinion being more likely to select ‘much more positively’.

Opinions of respondents on photographs of brownfield and undeveloped sites (questions 4-5)

Contrary to our expectations, respondents gave significantly more positive opinions about photographs of brownfield sites than photographs of undeveloped sites (Fig. 6; Χ² = 160.32, d.f. = 5, P < 0.0001). However, this effect appeared to be driven by particularly high rates of positive responses to one of the two brownfield sites in particular, rather than necessarily an overall effect (Fig. 6b).

When responding to photographs of the two brownfield sites, respondents rated their opinion as ‘Very positive’ significantly more often for photographs that they guessed showed an undeveloped site, and as ‘Neutral’ or ‘Fairly negative’ more often for photographs that they guessed showed a brownfield site (Fig. 7). This held true regardless of whether the photograph actually showed a brownfield site (Χ² = 167.31, d.f. = 5, P < 0.0001) or an undeveloped site (Χ² = 289.19, d.f. = 5, P < 0.0001).

Overall, respondents performed marginally better than chance when asked to guess whether each of the four photographs depicted a brownfield site or an undeveloped site (Exact binomial test, n trials = 8988, n successes = 5005, P < 0.0001), but with a caveat. For all four sites, more than half of respondents identified the site as being undeveloped. For three out of four sites, including both brownfield sites, approximately 55-60 per cent of respondents guessed the site was undeveloped. The exception was site ‘Undeveloped_2’, which 79.0 per cent of respondents correctly identified as undeveloped. Excluding the latter site, respondents only guessed correctly in 47.9 per cent of cases, which was significantly worse than chance (Exact binomial test, n trials = 6741, n successes = 3229, P = 0.0006). Respondents’ positive or negative opinion of brownfield sites had no influence on their chance of identifying such sites correctly (Χ² = 2.592, P = 0.7626).

Opinions of respondents on benefits/disbenefits of various site types (question 6)

For all nine available key phrases (see Fig. 8 legend), site type significantly affected the proportion of respondents who agreed with the statement (Fig. 8). In comparison to arable farms, nature reserves, and renewable energy facilities, brownfield sites were considered by fewer respondents to contribute to the local economy, create jobs, and/or provide value to the local community; but by more respondents to enable anti-social behaviour and criminal activity, and create pollution. Brownfield sites were considered to fit into the landscape, improve biodiversity, and provide a place for rare species by more respondents than renewable energy facilities, but by fewer than nature reserves; in comparison to arable farms, brownfield sites were considered to fit into the landscape and improve biodiversity by fewer respondents, but to provide a place for rare species by an equal number.

Preferences of respondents for repurposing sites, given site type and wildlife value (question 7)

When considering hypothetical repurposing of sites, on average, respondents ranked nature reserves as their most preferred option (Fig. 9a), followed by all four renewable energy options (with ‘agriculture’ most preferred of these). Industrial use was ranked lowest, whilst ‘no change’ (that is, leaving a site in its current unmanaged condition) was ranked at the extremes of the scale by most respondents, with few placing it as an intermediate choice.

However, rankings of different repurposing options were affected by both site type (‘brownfield vs undeveloped’) and site wildlife value (‘little value to wildlife’ vs ‘rich in wildlife’). Perhaps surprisingly, respondents showed greater support for brownfield sites to be repurposed as nature reserves than undeveloped sites (Fig. 9b), and also ranked industrial uses and solar farms higher on brownfield sites than on undeveloped sites, whilst ranking ‘no change’ higher on undeveloped sites than on brownfield sites. Respondents ranked nature reserves and ‘no change’ much higher on sites rich in wildlife compared to those with little value to wildlife, whereas all six options that involved site development were ranked higher on sites with less value to wildlife. Effect sizes (that is, average differences between rankings within each comparison) were much larger for the effect of rich vs poor in wildlife (site wildlife value) than the effect of brownfield vs undeveloped site (site type), indicating that the former variable invoked stronger changes in preference than the latter.