Nitriﬁed Human Urine as a Sustainable and Socially Acceptable Fertilizer: An Analysis of Consumer Acceptance in Msunduzi, South Africa

: Sustainable smallholder farming is contingent on fertilizer access. Soils across Africa are typically nutrient deﬁcient, a condition exacerbated by long-term nutrient mining. Nitriﬁed urine fertilizer is a nutrient-rich and hygienically safe solution derived from human urine. It has the potential to provide a sustainable source of soil nutrients to low and middle-income countries struggling with food insecurity challenges. This study presents ﬁndings of a survey that assessed public acceptance within Msunduzi, Kwazulu-Natal, South Africa towards the use of nitriﬁed urine fertilizer. Results indicate that in general attitudes were much more positive towards the use of nitriﬁed urine fertilizer than raw urine as a soil amendment. Residents living within rural zones of the municipality (78.5%), as opposed to urban (65.7%) and peri-urban (65.2%), and younger individuals within the sampled population were found to be the most receptive to the use of nitriﬁed urine fertilizer. Our ﬁndings also underscore the complex set of factors that shape attitudes towards a topic such as the use of human waste as a fertilizer, which are crucial in shaping the legitimacy of an emerging technology such as urine nitriﬁcation.


Introduction
Smallholder agricultural production is ubiquitous across Africa. These systems serve as a means of increasing available supplies of food for both subsistence and to diversify revenue streams [1]. From rural to peri-urban and even into urban areas of human settlements on the continent, smallholder production is an important livelihood strategy and provides a key buffer to price volatility, long distances to markets, and low incomes. However, a major limiting factor for smallholder producers in Africa is a chronic lack of fertilizer. Soils across the region are typically nutrient deficient, a condition exacerbated by long-term nutrient mining [2]. This biophysical reality impedes the sustainability of smallholder farming as a livelihood strategy.
Recent advances in sanitation technology have the potential to change this. The Swiss Federal Institute of Aquatic Science and Technology (EAWAG) recently developed a method and technology to separate and process human urine into a concentrated, nutrient-rich solution as part of a distributed sanitation solution for peri-urban communities in South Africa [3]. Just as with raw human urine, which has been used as a source of soil nutrients for millennia in countries across Asia [4], nitrified urine is an effective fertilizer. Recent greenhouse trials show that the uptake of nitrogen and phosphorus from nitrified urine fertilizer (NUF) by plants is similar to commonly used chemical fertilizers [5]. Additionally, this process eliminates pathogen risks, reduces excreted pharmaceuticals to the limit of adequate services (sanitation, housing, power) to its growing population. This process of urbanization has created a distinct rural-urban spectrum, which is rapidly changing traditional land use patterns.
There are approximately 163,993 households in Msunduzi. According to municipal categories of race/ethnicity, 81.1% of the population is Black African, 2.9 % colored, 9.8% Indian/Asian, 6% White, and 0.3% other. The majority of the population speaks either isiZulu or English. The general unemployment rate is 33% and youth unemployment is 43.1% [17]. Furthermore, the region is highly integrated into the global food system, with many of the largest grocery store chains well represented in the city. Despite this, food insecurity is a major challenge facing the population of the region, with 60% of households classified as being food insecure [16].
For this study, we first delineated wards, or geopolitical subdivisions within the municipality, into rural, peri-urban, or urban typologies. This classification was done by analyzing the population density, the percentage of households connected to the municipal sanitation system, and access to municipal garbage services across all the wards of Msunduzi. Once divided, a stratified cluster sampling approach was adopted to randomly identify survey zones ( Figure 1). There are approximately 163,993 households in Msunduzi. According to municipal categories of race/ethnicity, 81.1% of the population is Black African, 2.9 % colored, 9.8% Indian/Asian, 6% White, and 0.3% other. The majority of the population speaks either isiZulu or English. The general unemployment rate is 33% and youth unemployment is 43.1% [17]. Furthermore, the region is highly integrated into the global food system, with many of the largest grocery store chains well represented in the city. Despite this, food insecurity is a major challenge facing the population of the region, with 60% of households classified as being food insecure [16].
For this study, we first delineated wards, or geopolitical subdivisions within the municipality, into rural, peri-urban, or urban typologies. This classification was done by analyzing the population density, the percentage of households connected to the municipal sanitation system, and access to municipal garbage services across all the wards of Msunduzi. Once divided, a stratified cluster sampling approach was adopted to randomly identify survey zones ( Figure 1). A questionnaire was then developed to understand the food security situation of Msunduzi, as well as to identify specific factors that impact residents' attitudes towards food grown with NUF. Relative food insecurity was determined by asking how often the respondent felt it was difficult to provide enough food for him/herself and the household. The respondents were given four possible options: always, most of the time, rarely, or never. A household that responded always or most of the time was considered food insecure, any household that answered either rarely or never was considered food secure. Participation in smallholder agriculture was quantified by asking respondents whether they engage in agricultural production, if so what proportion of their food is sourced from this activity, and whether they grow food for subsistence or income generation. To assess the impact of the treatment process of NUF on public acceptance, a logistic regression model was constructed to identify factors driving this change in opinion. To construct this model, the score difference between respondent answers from questions focused on raw urine and NUF was calculated. These scores were then grouped into two categories: those respondents whose answers did not change (0) and those whose answers changed significantly (>75). This categorical dependent variable was then evaluated against eight demographic predictors (hunger, race, income, religion, age, gender, education, and ward). Willingness to consume food grown with NUF was identified by asking respondents a four-point scale with yes, most likely yes, most likely no, and no being the available answer options. Pearson's χ 2 tests were used to test for independence between pertinent variables. A questionnaire was then developed to understand the food security situation of Msunduzi, as well as to identify specific factors that impact residents' attitudes towards food grown with NUF. Relative food insecurity was determined by asking how often the respondent felt it was difficult to provide enough food for him/herself and the household. The respondents were given four possible options: always, most of the time, rarely, or never. A household that responded always or most of the time was considered food insecure, any household that answered either rarely or never was considered food secure. Participation in smallholder agriculture was quantified by asking respondents whether they engage in agricultural production, if so what proportion of their food is sourced from this activity, and whether they grow food for subsistence or income generation. To assess the impact of the treatment process of NUF on public acceptance, a logistic regression model was constructed to identify factors driving this change in opinion. To construct this model, the score difference between respondent answers from questions focused on raw urine and NUF was calculated. These scores were then grouped into two categories: those respondents whose answers did not change (0) and those whose answers changed significantly (>75). This categorical dependent variable was then evaluated against eight demographic predictors (hunger, race, income, religion, age, gender, education, and ward). Willingness to consume food grown with NUF was identified by asking respondents a four-point scale with yes, most likely yes, most likely no, and no being the available answer options. Pearson's χ 2 tests were used to test for independence between pertinent variables.
Working in collaboration with researchers from the University of Kwazulu-Natal, 11 local bachelor students were recruited to conduct the fieldwork, which occurred over the course of 10 days in October 2017. Data were collected with mobile phones equipped with the software platform "Mobenzi Researcher" and automatically uploaded to a central web console. A sample size of 392 household questionnaires with a response rate of 86 % was obtained during this period. Coding and data analysis was performed using Statistical Package for the Social Sciences (SPSS) version 24 as well as Rstudio1.

Results
392 household questionnaires were collected from rural, peri-urban, and urban wards across Msunduzi during the survey campaign ( Figure 2). Table 1 provides the demographic characteristics of the sample population. Below we present the results on acceptance and for the influencing factors. Working in collaboration with researchers from the University of Kwazulu-Natal, 11 local bachelor students were recruited to conduct the fieldwork, which occurred over the course of 10 days in October 2017. Data were collected with mobile phones equipped with the software platform "Mobenzi Researcher" and automatically uploaded to a central web console. A sample size of 392 household questionnaires with a response rate of 86 % was obtained during this period. Coding and data analysis was performed using Statistical Package for the Social Sciences (SPSS) version 24 as well as Rstudio1.

Results
392 household questionnaires were collected from rural, peri-urban, and urban wards across Msunduzi during the survey campaign ( Figure 2). Table 1 provides the demographic characteristics of the sample population. Below we present the results on acceptance and for the influencing factors.

Food Insecurity
We found that roughly half (47.6%) of the households in Msunduzi can be considered food insecure. When analyzed by ward, the peri-urban communities suffer the highest level of food insecurity ( Figure 3). Interestingly, the urban wards, despite having a large number of food retailers and being highly integrated in the global food network, still suffer from very high levels of food insecurity (38.1%). A χ 2 test of independence conducted between ward and food insecurity was significant (χ 2 (2, N = 382) = 6.843, p = 0.033), indicating that ward does influence rates of food security within Msunduzi. Based on an odds ratio analysis, we found that the likelihood of being food secure in an urban ward of the municipality is 1.95 times higher than compared to a peri-urban ward. Additionally, there is a significant relationship (χ 2 (3, N = 287) = 9.215, p = 0.027) between income and food insecurity. Households surveyed that earn no more than R 1500/month (equals approximately 110 USD/month), which comprised 48.1% of the total sampled population, were more likely to answer that they either always or mostly find it difficult to secure enough food for themselves and their families ( Figure 4). This intuitively makes sense and does indicate that hunger in Msunduzi is inextricably linked with poverty. insecure. When analyzed by ward, the peri-urban communities suffer the highest level of food insecurity ( Figure 3). Interestingly, the urban wards, despite having a large number of food retailers and being highly integrated in the global food network, still suffer from very high levels of food insecurity (38.1%). A χ 2 test of independence conducted between ward and food insecurity was significant (χ 2 (2, N = 382) = 6.843, p = 0.033), indicating that ward does influence rates of food security Additionally, there is a significant relationship (χ 2 (3, N = 287) = 9.215, p = 0.027) between income and food insecurity. Households surveyed that earn no more than R 1,500/month (equals approximately 110 USD/month), which comprised 48.1% of the total sampled population, were more likely to answer that they either always or mostly find it difficult to secure enough food for themselves and their families ( Figure 4). This intuitively makes sense and does indicate that hunger in Msunduzi is inextricably linked with poverty.

Small-Scale Agriculture
Although it may be the case that smallholder production is not as widespread as in the past or in comparison to other African countries, we found that it is still an important factor for many households living in Msunduzi, with 54.7% of all respondents stating that they engage in some sort of agricultural activity ( Table 2). In rural areas, 64.4% of respondents answered that they engage in some type of agricultural activity. Respondents in both the peri-urban (48.2%) and urban (51.5%) zones also stated that they

Small-Scale Agriculture
Although it may be the case that smallholder production is not as widespread as in the past or in comparison to other African countries, we found that it is still an important factor for many households living in Msunduzi, with 54.7% of all respondents stating that they engage in some sort of agricultural activity ( Table 2).
In rural areas, 64.4% of respondents answered that they engage in some type of agricultural activity. Respondents in both the peri-urban (48.2%) and urban (51.5%) zones also stated that they engage in some form of agriculture. By far the major reason given for engaging in this activity is to supplement diet (71%). A quarter of respondents claim to grow food both for subsistence as well as to sell, and only 2.3% stated that they grow solely to supplement income. A χ 2 test of independence found no relationship between a household's relative food security and whether or not they engage in agricultural activity.

Attitude Toward NUF vs Raw Urine
Our findings suggest that acceptance changes drastically when respondents consider NUF rather than raw urine as a proposed soil amendment. The histogram below ( Figure 5) displays the results from two related statements to which respondents were asked to indicate their level of agreement versus disagreement, one regarding the use of treated urine (Question 1), and the other focused on the use of raw urine (Question 2). Our findings suggest that acceptance changes drastically when respondents consider NUF rather than raw urine as a proposed soil amendment. The histogram below ( Figure 5) displays the results from two related statements to which respondents were asked to indicate their level of agreement versus disagreement, one regarding the use of treated urine (Question 1), and the other focused on the use of raw urine (Question 2).
Question 1: Recycled human urine, if treated properly to ensure it is safe, would make a suitable fertilizer.
Question 2: The use of untreated urine, because of its high nutrient content, can serve as a fertilizer for food crops. The mean value respondents gave for question 1, which asked for an opinion on the use of treated urine, was 75.8, while the mean value of respondents towards untreated urine was 15.2. These results suggest that public attitudes towards the utilization of NUF as a fertilizer for food crops are much more favorable than attitudes versus the use of raw urine. Furthermore, a logistic regression model was constructed (Table 3) to identify factors driving this change in opinion. Table 3. Results from logistic regression to identify significant predictors. p < 0.05. Question 1: Recycled human urine, if treated properly to ensure it is safe, would make a suitable fertilizer. Question 2: The use of untreated urine, because of its high nutrient content, can serve as a fertilizer for food crops. The mean value respondents gave for question 1, which asked for an opinion on the use of treated urine, was 75.8, while the mean value of respondents towards untreated urine was 15.2. These results suggest that public attitudes towards the utilization of NUF as a fertilizer for food crops are much more favorable than attitudes versus the use of raw urine. Furthermore, a logistic regression model was constructed (Table 3) to identify factors driving this change in opinion. Table 3. Results from logistic regression to identify significant predictors. * p < 0.10, ** p < 0.05. Within the eight predictors, the youngest age cohort sampled (16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) and respondents living in the rural wards had difference scores that were significant. The proportion of respondents from each ward type that expressed positive attitudes towards the purchase or consumption of food grown with NUF was also assessed ( Figure 6). Within the eight predictors, the youngest age cohort sampled (16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) and respondents living in the rural wards had difference scores that were significant. The proportion of respondents from each ward type that expressed positive attitudes towards the purchase or consumption of food grown with NUF was also assessed ( Figure 6) However, many respondents continued to express reservations towards the use of NUF as a fertilizer. Although multiple reasons were cited for this, a lack of trust in the processing of urine to NUF and thus a continued concern regarding the health implications was the most common source of this reservation (Figure 7). However, many respondents continued to express reservations towards the use of NUF as a fertilizer. Although multiple reasons were cited for this, a lack of trust in the processing of urine to NUF and thus a continued concern regarding the health implications was the most common source of this reservation (Figure 7). . Sources of continued concern regarding the use of NUF as a soil amendment.

Willingness to Purchase Food Grown with NUF
Respondents were also asked whether they would be willing to purchase food if they knew it was grown with NUF. Table 4 displays the results of a series of χ 2 tests that assessed the relationship between willingness to purchase food grown with NUF and the following independent variables: ward, sex, income, education, race, and engagement with agriculture. The only statistically significant results were between willingness to purchase NUF grown food and ward type (χ 2 (6, N = 391) = 16.513, p = 0.011) and willingness to purchase NUF food and race (χ 2 (3, N = 392) = 9.048, p = 0.029). Residents within the rural areas of Msunduzi showed the highest (78.2%) willingness to purchase food grown with NUF ( Figure 7). . Sources of continued concern regarding the use of NUF as a soil amendment.

Willingness to Purchase Food Grown with NUF
Respondents were also asked whether they would be willing to purchase food if they knew it was grown with NUF. Table 4 displays the results of a series of χ 2 tests that assessed the relationship between willingness to purchase food grown with NUF and the following independent variables: ward, sex, income, education, race, and engagement with agriculture. The only statistically significant results were between willingness to purchase NUF grown food and ward type (χ 2 (6, N = 391) = 16.513, p = 0.011) and willingness to purchase NUF food and race (χ 2 (3, N = 392) = 9.048, p = 0.029). Residents within the rural areas of Msunduzi showed the highest (78.2%) willingness to purchase food grown with NUF (Figure 7).

Discussion
A key finding of this research is that overall attitudes within Msunduzi towards the use of NUF are much more positive than towards the use of raw urine as a fertilizer. This is in line with a previous study in South Africa, which found that negative attitudes towards nutrient recycling are shaped primarily by public health concerns [12]. We find that if these concerns are alleviated, it results in a shift towards greater acceptance of nutrient recycling from urine. As the data show, respondents answered much more favorably to the idea of utilization of NUF than towards raw urine as a fertilizer. However, the data also indicate that a lack of trust negatively affects acceptance. For example, 61% of respondents who stated that they would oppose the use of NUF as a fertilizer cited a continued fear of possible health problems as the primary reason. Additionally, when offered the opportunity to specifically state the cause of this fear, many respondents indicated that they do not trust the process involved in the conversion of urine into NUF.
When analyzed more closely, the data showed that two predictors most influenced respondent attitudes towards the use of NUF as a fertilizer: age and ward type. The logistic regression model shows that younger respondents (16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) and citizens living in the rural areas of the municipality displayed the largest magnitude of attitude change between the use of raw and treated urine for fertilizer use. Based on these results, we infer that the younger age cohorts are more open and trusting of the nitrification process employed to treat the urine than older respondents. Similarly, a study focused on the acceptance of urine diversion dry toilets in eThekwini municipality also found that older residents were less receptive to adopting the new technology [18]. Additionally, residents within the rural areas of the municipality also displayed a larger change in attitude due to the treatment process than their peri-urban or urban counterparts. This, in combination with chi-square results showing rural residents being more willing to consume food grown with NUF, indicates a higher level of acceptance for this technology in rural compared to urban and peri-urban zones of the study area.
Research in other parts of the continent indicate that cultural norms and religion serve as important drivers of acceptance of human urine as a fertilizer, which was not the case in our study. For example, a study in Nigeria and Ghana found consumers had specific religious (38%) and cultural objections (26%) to the use of human urine as a fertilizer [9]. In contrast, our study found only 6 % of residents objected to NUF for religious reasons. This is also similar to the percentage of residents (6 %) within the neighboring municipality of Ethekwini that objected to raw urine as fertilizer for religious reasons [12]. Our study and that of Okem [12] were conducted in adjacent municipalities, both of which are located within Kwazulu-Natal, and are comprised of populations with very similar demographic makeups. We can thus infer that for respondents whose attitudes are based on religion, the treatment process has no effect. This supports assertions that behavior towards sanitation and excrement is a complex interaction between individual and societal norms and is often not based on scientific logic or knowledge [19,20].
Our results also indicate that lack of trust in the technology (NUF) is an issue. Many respondents indicated that they were suspicious of the treatment process. Phrases such as "it's new and no proof", "I want to see the process of the urine being cleaned", "assurance of proper cleaning" were common when respondents were given an opportunity to explain their reluctance in their own words. These attitudes are not surprising. Trust in new technologies has been shown to take time [21]. Additionally, issues of trust and fairness have all been shown to influence public attitudes [14,22]. In South Africa, the provision of sanitation services makes this quite clear. Under apartheid, decades of preferential municipal spending on public services led to widespread distrust and perceived unfairness between citizens and the state [23]. This could perhaps help to explain why age was a significant predictor of acceptance towards NUF; the 16-31 cohort did not experience apartheid directly, and are less influenced by its legacies.
Given our results, it seems clear that the 'innovation journey' [24] of NUF is uncertain. According to Deuten et al. [25] the successful incorporation of a new technology within a given society requires that it align closely within three environments: the business environment, regulatory environment, and the wider society. The degree to which an innovation can accomplish this alignment is dependent on the degree of cultural legitimacy it accrues [26] and is of particular importance in the early phases of the innovation journey. As our results show, this will depend in large part on the logic that formed public attitudes towards nutrient recycling. In regions where attitudes towards nutrient recycling are shaped primarily by public health concerns [27,28], it is likely that sanitation technology can accrue higher levels of legitimacy than in regions with attitudes shaped by other factors. Although regulatory bodies in countries such as Switzerland [29] have recently amended existing laws to allow the use of NUF as a fertilizer for food crops, the use of human waste as a fertilizer is currently banned in South Africa. Whether or not other regulatory agencies adopt similar regulatory changes is difficult to predict.
Given this uncertainty, we consider Fuenfschilling and Truffer's [8] suggestion that the successful dissemination of radical innovations requires the creation of a technical niche critical in the context of human based fertilizer use. This protected space allows for "the maturation of new technologies and alignment with a suitable institutional context." Going further, the transition literature [30,31] suggests this can be done through the provision of subsidies and research grants, or the creation of experimental implementation projects [32]. In our opinion, the widespread adoption and scaling up of emerging technologies such as urine nitrification that can improve the sanitation and food security landscape in low and middle-income countries will require the creation of these types of platforms.
Finally, to contextualize the relevance of our results, we have embedded our research in the general state of the food system within Msunduzi. Our findings on acceptance are highly relevant given this study area, as we found that poverty is a major driver of food insecurity in Msunduzi municipality. A χ 2 test found a significant relationship between income and food insecurity (p = 0.027), indicating that households earning no more than R 1500/month (= 113 USD/month) suffer from very high rates of food insecurity. This income cohort comprises 48% of the total population of Msunduzi, indicating the widespread nature of this challenge. Furthermore, despite the large numbers of food retailers in the urban zones, high rates of food insecurity were identified in every area of the municipality. This reinforces the link between poverty and food insecurity within the study area and makes the use of human urine as a fertilizer more pertinent.
Beyond food insecurity, the role of smallholder agriculture also makes acceptance of urine fertilizer salient. While it may be true that this activity plays a more prominent role in other African countries [16], our study found that smallholder production continues to serve as an important livelihood strategy in Msunduzi; 54.7% of all respondents indicated that they engage in some form of agricultural production. Of those, the majority do so primarily for subsistence purposes. Although the rural areas are the most heavily invested in agriculture, the respondents in peri-urban and urban zones indicated a high level of involvement in this activity as well. This finding was unexpected and warrants further study. Interestingly, no significance was found between engagement with agriculture and attitudes towards the use of NUF.

Conclusions
We contribute to a growing body of evidence that suggests that NUF production can play a significant role in the challenge of increasing soil nutrient access for farmers in areas such as South Africa. Our findings indicate that consumer attitudes towards the use of recycled human waste can be positively shifted through the processing of raw waste such as urine into a treated product such as NUF. Specifically, we found that residents in rural wards and younger residents were the most open to the use of treated urine as fertilizer. However, we also conclude that the impact on existing attitudes is highly dependent on the societal norms and underlying logic upon which those attitudes were shaped. This underscores the complex dynamics involved in promoting a shift towards the wide scale adoption of nutrient recycling technologies and the critical importance in understanding the local context.