Waste water reuse in France – social perception of an unfamiliar practice Open Access

Every second year, half the French departments set up restrictions on use due to water scarcity (Benoit et al. 2017). Irrigation is then limited or even prohibited for weeks. Waste Water Reuse (WWR) is thus seen as an interesting solution in these territories, especially as it may be considered as participating in the circular economy and contributing to a so-called virtuous circle (Maurer 2018). However, only 0.2% of treated urban wastewater effluents are reused annually in France. This figure is lower than the European average (2.4%), and a long way from the practices of Italy and Spain, which reuse between 8% and 12% of their treated effluents (Aquarec 2006).

This situation can be explained firstly by the very restrictive regulatory framework. In France, thanks to the 1992 French Water Law, treated wastewater can be utilized but only to irrigate lands, ‘provided that its characteristics and manner of use are compatible with the requirements for the protection of public health and the environment’ (decree of June 3, 1994). And the Order of June 25, 1994, defines four classes of water quality from A to D. Each class is associated with constraints on use, distance and terrain to be respected according to the crop, the area to be irrigated, and strict technical regulations for sprinkling irrigation – the dominant practice in France (see the following link for an in-depth presentation of the regulation: https://www.arpe-arb.org/files/20171130_2017Rutiliserleseauxtraites.pdf). A survey done by the public institution CEREMA (Centre for Studies and Expertise on Risks, the Environment, Mobility and Development) reports that the main treatment is ‘activated sludge’ which accounts for 73% of WWR cases (against 44% of all wastewater treatment plants in France), ahead of ponds 23% (against 28%), and membrane processes 3% (against 0.3%)’ (Franck-Néel 2020). CEREMA identified only 63 schemes in operation in 2017 in France, including 35 for agricultural purposes, concerning much less than 1‰ of the 1.5 million hectares irrigated annually in the country. Most projects, less than 100 ha, were initiated before the 2000s in rare areas with a really acute lack of water (i.e. islands, coastal zones, etc.), to satisfy some agricultural uses close to wastewater treatment plants.

But this poor practice of WWR is also due to the French situation, where water has not been scarce (thanks also to other solutions to fight against water scarcity, like dams, canals, etc.). This second key reason (both reasons underlined by Lazarova & Brissaud (2007) and Cerceau (2015)) may explain the gradual change that appeared after 2000, with 40% of the new projects considered as addressing quantitative or qualitative water issues at the catchment level. There is now an annual growth rate of 7% in the number of projects (Franck-Néel 2020). WWR is emerging in public and/or national French debates as a way to cope with repeated droughts, especially to supply water to territories not yet covered by irrigation networks, and thus to crops previously not irrigated (wine, lavender, olive trees). For instance, the ‘Assises de l'eau’, a national forum organized by the National Water Committee, proposed in their conclusions (16 July 2020) to triple the volume of treated WWR.

In this context, French Water Agencies are increasing pilot experiments, to explore the potentialities of this ‘new’ method of satisfying increasing water needs. The French research and development project SOPOLO (‘WWR a possible local solution?’) takes up part of these, looking to estimate wastewater demand, especially for irrigation purposes, in a Mediterranean fine-wine-growing territory, Pic Saint Loup, located in the South of France (Figure 1). This vineyard is not irrigated yet. There is no WWR project for agriculture nearby. However, after several severe droughts during the last decade, more and more farmers are complaining about lack of water for irrigation. Recycling water from several sewage treatment plants is one of the options for the local public authority (Communauté de Communes du Grand Pic Saint Loup – CCGPSL), which is a partner in our research activity.

As irrigation can be viewed, from the point of view of economics, as an intermediate resource desired not in itself but for growing products later sold in the market, it seems important to assess the level of acceptance by customers of agricultural products (i.e. households). The aim of this paper is to present the attitudes of consumers, buying their fruit, vegetables and wines in this area, toward a WWR project and their intended behaviour concerning local irrigated foods and recreational areas. These elements will give indicators of the risk of consumers turning away from local production, which would jeopardize the viability of the WWR project.

In the first part, we provide a brief overview of the state of the art of WWR's social acceptability issues; we then present the research method before discussing the main results.

‘Social acceptability’ is an ‘embarrassing’ notion (Barbier & Nadaï 2015), but it was soon acknowledged as being a key element in the development of WWR in the same way as technical issues are (Baumann & Kasperson 1974). During the last 40 years, there has been a lot of research investigating factors associated with public acceptance of recycled water in many countries (Fielding et al. 2018), but not in France. In addition, in this country, Berry et al. (2016) revealed difficulties related to the rise of controversies concerning all major infrastructure projects (in their case study on WWR, the interviewees refer in particular to the highly controversial project of a new airport ‘Notre Dame des Landes’). Water is not exempt from these controversies, whether for the justification of irrigation, the quality of drinking water (Hervé-Bazin 2014), or the public interest in new dams for irrigation (Sibertin-Blanc 2019). France has been marked by health scandals affecting food: mad cow disease, children's milk contaminated with salmonella, avian influenza, swine fever, and meat and cheese affected by Escherichia coli. Consumer associations regularly denounce the high levels of pesticide residues on fruit and vegetables, and question the government about emerging pollutants. This context creates a climate of mistrust in food chains that use industrial processes. Our hypothesis is that the treatment of wastewater for agriculture could be seen by consumers as one of these complex industrial processes, susceptible to malfunctions in the elimination of bacteriological, viral or chemical elements. However, WWR has so far received very little media attention in France: the media analysis conducted by Noury et al. (2019) for the period 2000–2017 for 50 national, regional and specialized press titles revealed that few press articles (156) mentioned WWR, with only 30% focusing on it. Most of these, published after 2010, mentioned WWR as a way to deal with repeated droughts without the need for new dams, a particularly sensitive subject, and one in conformity with the political trends of circular economy.

The main French reference to public views of water issues is the annual national barometer of opinions from a representative sample of 2,508 people, about water and wastewater services. In the questionnaire, with over 100 questions, there are only three concerning people's acceptance of using recycled water and one on their knowledge of the water cycle (Centre d'Information sur l'Eau & TNS Sofres 2018). However, this national barometer reports general public opinions on the principle of recycling water, measuring ‘the extent to which a person believes that supporting the recycled water scheme will deliver positive outcomes’ (Nancarrow et al. 2008); this is not an opinion on the desirability of a WWR project close to their homes. Nancarrow et al. (2008) had already documented a difference between supporting the concept of recycling water, because this attitude is in line with the values people stand for, and accepting personal contact or ingestion of food irrigated with such water, because this involves other intimate factors. More precisely, we wanted the answers to five questions, based on literature reviews (Dolnicar et al. 2010; Fielding et al. 2018; Etale et al. 2020), and the French national survey (Centre d'Information sur l'Eau & TNS Sofres 2018):

• What were the main components of the social representation of WWR in this area, if any?

• Were the inhabitants of this area in favour of irrigating agricultural or recreational areas with recycled water?

• Did their opinions vary depending on the agricultural products or recreational spaces considered?

• Did their opinions change if they were provided with basic information on the principles of water treatment and control?

• Was there consistency between their judgment on the appropriateness of a WWR project and their envisaged purchasing of agricultural products and use of recreational areas irrigated with recycled water?

The French national survey indicated that 52% of the population believed that wastewater was already directly used for drinking water, even though this is strictly forbidden by regulation, and so no treated wastewater is directly used for this purpose. This observation, combined with only the very few number of French WWR projects (no one near to Pic Saint Loup Region), and the low media visibility, led us to assume that the local population, too, had very little knowledge of the subject, and that it was an unfamiliar practice. According to social representation theory (Rateau et al. 2011), in this context of unfamiliarity, opinions and beliefs were probably not yet imbedded in knowledge collectively produced by a social group (i.e. a social representation). Identifying the ‘central core’ of cognitive elements constituting the social representation of wastewater treatment, if any, was therefore of fundamental importance. The central core, anchored in people's minds, is particularly difficult to change (Vergès 1992), even with an information campaign. On the contrary, the other (‘peripheral’) elements of a social representation can be changed by ordinary exchanges through social networks and by the media. These elements are also sensitive to ‘information and awareness-raising’ (Noury et al. 2019). We intended to test whether positive information on WWR increases the acceptance rate for people who do not yet have deep-rooted beliefs about this topic (Noury et al. 2019).

At the national level (2018 survey) (Figure 2), 73% of the population considered water as a ‘limited’ resource, and 59% feared water shortages in the future; 83% trusted health authorities to control water quality; 86% would accept using tap water from recycled wastewater for their domestic purposes (hygiene, sanitation, cleaning, etc.), and 75% would be willing to consume vegetables irrigated with treated wastewater. This proportion seems to be increasing slowly in France: in 2013, there were 32% who expressed their opposition to the idea of eating fruit and vegetables watered with treated wastewater (Ben Maïd et al. 2014), without mention of their motivations. We had no reason to assume that the consumers in the Pic Saint Loup region had a different attitude towards WWR in general. However, we wanted to know if they would have the same judgment if a project were set up in their neighbourhood, bringing them into contact with products irrigated with reclaimed water. We aimed to go a step further in the evaluation of the acceptance rate, by determining the purchasing intended behaviour: would they agree to maintain their purchasing habits for local products (wine, fruit and vegetables, cheese) or utilize recreation areas if they knew they had been irrigated with reclaimed water (all economic aspects neutral, such as prices)? In this research project, we did not have the means to make a taste test with a large panel of consumers. We assumed that stated purchasing intentions would provide a fair approximation of actual behaviour.

For this purpose, we used the theoretical framework proposed by the psychologist Ajzen (1991), who distinguished attitudes, the intended behaviour, and the actual behaviour. This framework has already been successfully used by Nancarrow et al. (2008), who established a structural equation model to predict communities' behavioural decisions in relation to the recycling of wastewater for uses involving close personal contact. This framework helped us to select a limited (but observed as relevant in Nancarrow et al. 2008) number of factors influencing attitudes, and expected behaviours to test their relevance to our case study and assess their sensitivity to an information action:

• Disgust. The extent to which a person experiences the ‘yuck’ factor at fruit and vegetables irrigated with treated wastewater.

• Perceived sanitary risk. The sanitary quality of food products is controversial in France, where there have been high-profile health scandals. Moreover media alerts concerning emerging pollutants in water (drug residues, pesticides, plastics, endocrine disruptors…) may be considered by consumers as another source of sanitary risk for WWR.

• Declared sensitivity to environmental issues, as part of a perceived social pressure, widely relayed by the media (promotion of recycling, sobriety, local sourcing).

• Trust. In France, there is both goodwill towards local agriculture, which supplies local markets, and a strong mistrust of intensive agriculture, which is widely reported in the media. Similarly, there is great confidence in the quality of domestic water but mistrust of private water companies (Hervé-Bazin 2014) and the majority of the population believes that water quality in the environment is poor and deteriorating. For instance, 75% of French people believe in 2018 that ‘the authorities take the utmost precautions to ensure that tap water quality standards protect consumers’ health’; 85% of French people are confident of tap water quality and drink it, at least occasionally; while 75% consider that the water in the river and aquifers is polluted (Centre d'Information sur l'Eau & TNS Sofres 2018) (see Figure 2). Therefore, we want to measure consumer confidence in the ability of farmers and professionals of the water sector in the Pic Saint Loup area to control the WWR's health risks.

• Confidence in technology. We want to test whether general confidence in progress could influence attitudes towards WWR that is very scantily practised in France.

• Information habits. This factor has been added with regard to the study of Nancarrow et al. (2008) following the works of Dolnicar et al. (2010), who found a statistical positive change in the acceptance of recycled and desalinated water when respondents are regular newspaper readers. We thus aim to test whether information habits impact French attitudes to WWR.

We conducted a survey in spring 2018 involving 845 consumers, carried out in ten localities belonging to the CCGPSL (Figure 1), near the main markets and supermarkets of the area, before or after they had bought their foodstuffs. Most of them (84%) were inhabitants of the ‘Communauté de communes’, and 7% were living nearby. Interviews were carried out face-to-face in the street with the help of Master students from Montpellier University, wearing ‘Grand Pic Saint Loup community flags’ to inspire confidence. All these people were questioned. When people came together, they were either interviewed independently (using two separate interviewers), or only one person was asked to respond. The refusal rates were very low (less than 10%). The sample did not include persons under the age of 15, as children are not responsible for household fruit and vegetable purchases. Apart from this bias, we afterwards checked that the sociodemographic characteristics of our sample of local inhabitants were close to those of the overall population (around 21,000 inhabitants in the area – see the Supplementary Material for more details of this survey).

The questionnaire, which was designed for completion within ten minutes, comprised only closed questions, and was the same regardless of the level of information previously provided. First of all, consumers were asked to give the three words they spontaneously associate with the expression ‘treated wastewater’ and to rank them in order of importance, following the associative method (Rateau et al. 2011) to qualify the social representations of this resource, if any. (Etale et al. (2020) asked the same type of question, ‘recycled water’ and ‘desalinated water’, then classified the words or expressions into 18 categories, which we tried in our case to follow – adding some categories due to the fact that the question was not exactly the same – and we used it especially in our factorial analysis.) When people were unable to think spontaneously of three-word associations, we did not insist. The rate of people who did not comply with this simple instruction is an indicator of the lack of social mobilization on the subject. A few people asked what treated wastewater was. After giving the word association instructions, we simply told them that it is water from sewage treatment plants.

The questionnaire continued in four different ways:

• A first sample with ‘no information’ (N = 204) simply went on with the questionnaire.

• For the ‘neutral information’ group (N = 310), people received a flyer on which were the characteristics of a WWR project, its objectives, who is involved and for which tasks. There was a very recognizable picture of the Pic Saint Loup at the top of the flyer to associate the project with the site. The rest of the text was not illustrated. The sentences did not include any value judgment (see the Supplementary Material). A student read the main points of the flyer to make sure they were understood. Then the questionnaire went on.

• For the ‘persuasive information’ group (N = 176), the protocol was the same, but the flyer was organized as an advertising medium, highlighting the benefits of WWR through key words on a circle to emphasize the message of recycling.

• Finally, for the ‘commitment’ group (N = 155), students presented to respondents the Pic Saint Loup's map asking the following question: ‘are you in favour of short circuits for food products? If yes, put a green sticker on your district of residence, otherwise a red one’. This type of commitment to a cause was meant to put the respondent in a benign disposition with regard to new information in agreement with the cause to which he had just committed himself (Bernard & Joule 2004). The students then presented the flyer of the persuasive communication on WWR. Then the questionnaire went on.

The questionnaire then focused on consumers' acceptance of nine uses in the village they were living in. We insisted on this point to make sure that they felt directly concerned by the project: three ‘recreational’ uses (golf, roundabout and green spaces) and six ‘agricultural’ uses (shared gardens, market gardening, orchards, olive trees, vine and meadows for livestock). The uses were chosen among the options studied by the local public water authority, and to reflect the range of proximity of skin or mouth contact with the wastewater. We used a four-item system rated on a Likert scale, from strongly agree to strongly disagree, with no ‘neutral’ response to force an agreed or disagreed position, but with an option of ‘I don't know’ for people who had no opinion.

Intended behaviour (purchasing, public garden visits) was assessed through one question per item (wine, fruit and vegetables, goat cheese, parks). Four answers were possible: yes/no/maybe/I don't know (five for wine and cheese with ‘not usually buying’). Individuals were invited to rank their environmental sensitivity on a three-item level Likert scale (with the option to say ‘don't know’). They were then asked to agree/disagree/partially agree or give no opinion on four aspects (on disgust, on health risks, trust in water actors, trust in technology). They were asked about their media information habits, which was further interpreted in four categories on the following basis:

• Well informed: at least two media read/listened to daily

• Informed: at least one media read/listened to daily

• Mildly informed: at least two media read/listened to once or twice a week

• Poorly or not informed: the other answers

• No reply: when no answer to any media sources.

The questionnaire ended with socio-demographic data.

In order to identify the social representation of ‘treated wastewater’, we followed the hierarchical evocation method proposed by Vergès (1992), who differentiates four areas based on two criteria (the level of interest, and the frequency of occurrence): the ‘core’ composed of consensual and fundamental elements (characterized by coherence and stability, which are thus resistant to change), the ‘contrasting elements’ (proposed by few people but with a high rank), and peripheral elements with more personal components (see Table 1 and Montginoul & Vestier (2018) for more details).

The three-word associations with the expression ‘treated wastewater’ and their rank clearly indicated that citizens did not yet have social representation on it:

• The vocabulary associated with ‘treated wastewater’ was poor: the rareness indices (words cited only once by one person) and diversity indices (number of different responses) were 0.14 and 0.22 respectively before lemmatization (which consists of grouping similar expressions, like the terms ‘connection’, ‘pipeline’, ‘water tower’ and ‘pipe’ under the same term ‘network’, in the second step of the analysis, to build Table 1). For comparison, a similar study conducted on the word ‘water’ obtained indices of 0.3 and 0.46 (Montginoul & Vestier 2018). Among the sample, 6% were unable to give a single word, 25% could give only one, 47% only two. These figures show that treated wastewater had not been much discussed among the population, which is at the origin of social representation (Rateau et al. 2011). This weakness of local knowledge and beliefs on this topic may be explained by the low media coverage of this subject in France.

• ‘Treated wastewater’ was strongly associated with ‘wastewater treatment plant’ (Table 1), indicating that citizens related this type of water to its producing entity. This sole strong association with any other, even in the first peripheral perimeter (lower rank but highly cited), indicates interviewers had difficulty in suggesting links. They were able to give only a neutral association, close to a synonym. We can however observe that 9% of the words (and then 21% of respondents) related ‘treated wastewater’ to WWR, which seems high given the WWR French context.

Contrasting elements (low frequency but high rank in Table 1) highlight different and less common ways of thinking. They refer here mainly to people who saw treated wastewater as an opportunity (recycling, water saving, etc.). There are very few elements negatively connoted (i.e. disgust, pollutions, dirt, cost…) appearing mainly in the ‘second peripheral perimeter’ (low frequency and low rank), that is for only a few respondents and at a low rank.

The survey questioned the attitude of citizens towards nine uses: three ‘recreational’ uses (golf, roundabout and green spaces) and six ‘agricultural’ uses (shared gardens, market gardening, orchards, olive trees, wine and meadows for livestock) (Figure 3).

These results support the patterns identified in the literature (Smith et al. 2018) on the relationship between attitude and uses: respondents were less likely to use treated wastewater for market gardening (70%) than for watering green spaces (90%). This distinction is also confirmed when consumer behavioural intentions were questioned: 83% would continue to frequent green spaces irrigated with treated wastewater while only 69% would continue to buy from grocers who sell fruit and vegetables irrigated with treated wastewater.

This corroborates the annual French national water survey in which a very large majority of individuals (75% in 2018) were ready to eat fruit and vegetables irrigated with WWR (Centre d'Information sur l'Eau & TNS Sofres 2018).

A statistical analysis was conducted to identify the sociotechnical variables that could be correlated with these attitudes, among those cited as relevant by various works in other countries (Nancarrow et al. 2008; Fielding et al. 2018; Smith et al. 2018). Table 2 presents results for the irrigation of the three typical uses for which behavioural intentions have been questioned during our survey.

Disgust, environmental sensitivity and risk perception influenced attitudes for all three uses. These results are in line with the synthesis of Fielding et al. (2018) in several dozen studies on the most discriminating factors of attitudes for non-potable uses. In this literature review, disgust, and the perception of health risks were cited as the main factors explaining unfavourable attitudes, while the claim of environmental sensitivity characterized favourable opinions. In our study, people who said they were disgusted with the thought of eating a vegetable irrigated with treated wastewater were unfavourable to all WWR projects; those believing there was no risk in using treated wastewater to irrigate vineyards were more favourable than others to all the WWR projects. The more environmentally sensitive the respondents, the more they were in favour of reusing wastewater to irrigate parks and gardens.

In our survey, confidence in technology also acted on WWR attitudes, but the probability of independence is less significant between these two variables. A gender-related effect is also observed but only for market gardening, for which men were more favourable to WWR than women were. In the synthesis of Fielding et al. (2018), the other variables (age, gender, confidence in technology) had varying effects across studies. When they had a significant effect, it was the same as the one observed in our study.

Information seemed to influence attitudes (Table 3), but not uniformly: it varied according to the type of use and the type of communication.

Information described as neutral significantly strengthened the favourable attitudes for all uses. This result is in line with Fielding et al. (2018) who wrote in their synthesis, ‘providing factual information about recycled water increases knowledge about, and acceptance of, recycled water’. This impact may be explained by the mention of the health authority's controls exclusively on this communication medium. In the national survey, 83% of the population trusted the authority to control water quality. The so-called persuasive information changed the opinion of the very opposed participants, but only as far as agricultural uses were concerned. This brochure explicitly mentioned agriculture in its slogan ‘WWR – a resource for our agriculture’, and as a local solution to food insecurity. Finally, in the commitment scheme, participants gave their opinion on local food systems, as a preparatory act, before being subjected to persuasive information. The results do not validate the scientific hypotheses (Bernard & Joule 2004) assuming that the effects of the information could be reinforced by an act of commitment. There was even a slight decrease in positive opinions compared with the control group (‘no information’). The relevance of the preparatory act must be questioned.

Information seemed to have some influence on the people who answered ‘mitigate’ when we asked them if they were disgusted by the idea of eating a fruit or vegetable irrigated with treated wastewater. With information (whatever its nature), the rate of these hesitating persons decreased significantly (p < 0.02). We can assume that many among them were completely unaware of the process of recycling water. The flyer was sufficient to convince them that the treatment had ‘purified’ the water. However, when a person claimed to be disgusted, he or she was in the realm of emotion, not reasoning, and information did not matter. These findings are in agreement with the work of Rozin et al. (2015), who studied the disgust at drinking treated wastewater.

Information had less of a reassuring effect on the perception of risk. When people first read the flyers, the rate of persons hesitant about their perception of sanitary risk decreased, but not significantly. If information seemed to reassure some of the undecided, others feared contagion or did not express an opinion. We assume that the multiplication of recent health and food scandals (mad cow disease, meat and milk quality, pesticide residues in food, etc.) aroused a great deal of suspicion of the reassuring information provided by public bodies.

The remaining sanitary risk perception was significantly and positively correlated with (i) disgust at the thought of eating food irrigated with treated wastewater and (ii) the lack of trust in water actors (at the 0.001% threshold). Disgust and risk perception seemed independent of environmental sensitivity, and of general confidence in technology.

Media consumption habits (radio, TV, web, newspapers), too, seemed to influence the effect of information among those who had received useful information through the media; their number increased the probability of responding ‘strongly agree’ rather than ‘agree’ (at the 1% threshold) by 25 percentage points. This supports an Australian study (Dolnicar et al. 2010) on WWR, which pointed out that the cause of heterogeneity in information responses was rather the experience of information processing than of socio-demographic data.

The study also found a significant correspondence between attitudes and behavioural intentions (p-value = 0.000). People in favour of WWR would not change purchasing habits if they learned that vineyards are irrigated with treated wastewater. A reassuring effect of the information was also noted. Consumers were less hesitant when they had information, but it did not greatly change unfavourable opinions.

The results presented in Table 4 for wine purchases were similar for vegetables or cheeses from breeding that uses fodder watered with wastewater (cf. the graphical abstract). Without information, 22% to 30% of consumers would switch shops if they knew that the shop buys its products from a farmer who irrigates with treated wastewater. With information, this rate was around 20%.

Again, disgust and the perception of persisting sanitary risk played a major role in the intention of purchasing food (p-value 0.01). Two-thirds of those who would change shops (for vegetables, goat cheese, wine) said they were disgusted by the idea of eating such fruit and vegetables; two-thirds also thought there was a health risk with WWR. As our flyers had an effect only on those who were most ignorant of WWR and hesitant about expressing disgust and about risks, purchasing intentions changed little with the information, in the case of this profile of consumers mainly.

Table 4 also reveals complex judgments; 5.8% of the uninformed and 4.6% of the informed were from in favour to very much in favour of water recycling for irrigating food products, as citizens, but not as consumers. Conversely, 8.9% of uninformed and 5.3% of informed persons, opposed to it as citizens, would tolerate it as consumers. In addition, depending on the products, the motivation of between 20% and 25% of those who would change shops was unknown, since they said they were not disgusted, and that they did not perceive risk of contagion. The very limited questionnaire did not allow for a more in-depth analysis of their motivations.

The cohesion between attitudes in favour of WWR for public parks was less consistent with the envisaged frequenting of them if they were irrigated with reclaimed water (see the Supplementary Material). Information did not significantly influence this level of inconsistency. A majority of those in favour of this use of WWR would use the parks if the project were carried out, but the actions of a larger part of the population would not be consistent with their opinions. Among those who would no longer use the parks, three-quarters of the people were in favour of this use. Conversely, three-quarters of the few initial opponents would eventually continue to frequent these places. Feelings of disgust and health risks could not explain these contradictions between attitudes and intended behaviours.

These apparent inconsistencies confirm the need to go beyond the stated intentions of action by interviewees, the level at which most of the work cited in the literature stops (e.g. Nancarrow et al. 2008). Between stated intention and actual action there are often great differences, as has been demonstrated regarding other pro-environmental behaviours (de Leeuw et al. 2015). It will be necessary to go as far as the observation of actual actions, i.e. here, the actual decision to eat or buy food in a taste test, for example.

Using a factorial analysis (of the 802 respondents who replied to all questions participating in the factorial analysis), we have thus clearly distinguished two opposing sub-populations: half of the total respondents (54%) were in favour to very much in favour of the development of WWR, for all uses. They would not change their consumption habits (more than 80% of them), they were not disgusted (95%), they did not perceive any health risks (67%), they were rather more confident of technological progress (62%), and sensitive to environmental issues (96%). This group is composed of people who had previously received information on WWR (81%), and for 41% of them a neutral kind.

On the other hand, a group representing 17% of the population is made up of people opposed to WWR (both in opinion and intention). Including 22 (out of 23) of the farmers who replied to all questions; this group comprises a majority of retired people (53%), especially more than 75 years old (24%). It includes the more heterogeneous population (cf. the hierarchical clustering – more details in the Supplementary Material), but is highly characterized by persons who have difficulties in associating words with ‘treated wastewater’ (60% did not give three words, 20% of them did not come up with any word).

The last group comprises 28% of the population (37% belongs to the ‘no information’ group, of which only 31% was given ‘neutral’ information); they were resistant to WWR, preferring to stop consuming products that result from the WWR (for more than 50% of them). This group comprised ‘mitigated’ people (on disgust, health risks, technology, and confidence in decision-makers), and a significant proportion of them (12%) declared themselves not very sensitive to the environment, but were concerned about health problems (the word ‘contamination’ was cited by 28%). This includes a majority of women (62%), people between 30 and 44 years of age (28%), working people (61%), and those less open to the media (27%). The great majority of them (80%) had a considerable idea of ‘treated wastewater’ (only 2% did not have anything to say).

The threat of a 20% (26% when no information) drop in potential local buyers would not be acceptable in an agricultural and wine-growing area such as Pic Saint Loup. We did a reconstitution of this work, at the invitation of a mayor. The few winegrowers present confirmed that selling to local customers was essential to the profitability of their business. They stated that they could not accept a 20% reduction in their clientele.

The reduction of current discharges from treatment plants was of interest to drinking water managers (i.e. reducing the risks of aquifer pollution in this karstic area). But the result of this study reinforced other weaknesses in the cost recovery and economic profitability of the WWR project pointed out by other studies conducted in the SOPOLO project. This explains why the Pic Saint Loup community of communes decided not to carry out the project in the near future.

The next research actions on this topic will include (i) a wider range of factors explaining attitudes towards WWR, (ii) an extension of the sample to people with a sound knowledge of wastewater treatment (professionals and students), (iii) an extension of the study of behaviours with tasting tests of products resulting from irrigation with treated wastewater, and (iv) feedback on the behaviour of consumers of agricultural production from some projects currently implemented in France.

WWR projects in France are still rare, and are only rarely the subject of social acceptability studies. In our survey, disgust and the perception of persisting sanitary risks clearly explained the reluctance of 5% to 30% of our sample of consumers. We found three groups of consumers: the promoters of WWR (50%) significantly more sensitive to the environment, and more confident in the technology and in water managers; the opponents (20%) fearing sanitary risks; and the reluctants partly reassured by prompt information on this little-known subject. WWR project leaders will see this as encouragement to launch intensive targeted information campaigns. But the reluctant group, significantly less open to the media and made suspicious by health scandals, will be difficult to convince. This study needs to be deepened, including taste tests, as some inconsistencies between attitudes, food-buying intentions, disgust, and perceived risks have emerged among a significant minority.

We would like to thank the Rhone-Mediterranean Water Agency for its financial support of the SOPOLO project, the Pic Saint Loup Community for its encouragement, the students of 2017–2018 class in the first year of the ‘Water’ Master's program at University of Montpellier who helped to carry out the surveys, and the consumers who agreed to answer.

Data cannot be made publicly available; readers should contact the corresponding author for details.