Survey, sampling, and data preparation

We developed a survey (McOmber et al. 2021) and pilot tested it among 12 greenhouse growers, which were excluded from the final sample. A revised survey was distributed through Qualtrics (Qualtrics LLC, Provo, UT, USA) as a hyperlink to ∼18,000 subscribers of Greenhouse Grower Magazine in Spring 2019. Additionally, the survey was distributed to greenhouse growers through the greenhouse extension specialists’ network at land grant universities throughout the United States in Winter 2019. Recipients of this survey link were requested to participate in the survey by receiving up to three reminders (Dillman et al. 2014). In total, the survey was completed by 421 growers.

To assess representativeness of our sample to the broader population of greenhouse growers, we drew from the 2018 State of the Industry Report [hereafter, Industry Report (Greenhouse Grower 2018)]. The greenhouse operations in our sample reflect a broad range of years of operation as well as size of greenhouse operations, although our sample contains growers that are smaller and on average have fewer years in operation compared with the Industry Report. McOmber et al. (2021) provides additional detail on the comparison of our sample to the Industry Report.

The survey had four sections (see Supplemental Material). The first section collected information about the respondent’s greenhouse operations and asked about growers’ experience and knowledge about recycled water. The second section of the survey asked respondents questions about their knowledge and perceptions of recycled water, tap water, and treated wastewater. The third section queried the respondent’s perceptions, willingness to use, and actual use of MRW, and the last section collected demographic information such as education, age, and gross greenhouse operation income. Although sections 1 and 2 established the respondent’s knowledge and attitudes about recycled water, we knew from pilot testing the survey that growers have diverse understandings of recycled water and would respond to questions using their own diverse frames of reference. To establish a more uniform framing for questions specific to MRW, we introduced an informational diagram between sections 2 and 3 of the survey (see Supplemental Material or McOmber et al. 2021). This figure illustrates where MRW comes from, how it is produced, and various uses of MRW. After viewing the informational diagram about MRW, respondents were asked a series of questions about their perceptions, attitudes, and use of MRW.

For analysis, we denote our variables using “Question” followed by the main question number with any relevant subquestion number noted in parenthesis (Supplemental Material). Given our interest in understanding willingness to use MRW in greenhouse production, we focused our analysis on the 346 respondents who indicated they were involved in greenhouse production (i.e., employees with or without decision power or owners involved or not in daily decisions). The 75 respondents including 60 who did not provide an answer to the question “What is your primary role regarding your greenhouse operation?” and 15 who chose “other” and where their open-ended response indicated they were not involved in greenhouse production were excluded from the analysis. The final regression model included 268 respondents, most of whom had decision-making power either as employees (55.3%) or owners (44.7%). These 268 respondents account for missing values from Question 25 (3) and Question 25 (5), which we combined to create our targeted response variable: willingness to use MRW on food crops. Variable values were defined as 0 if respondents were not willing to use MRW to irrigate food crops whether the water directly contacted the food crop and were defined as 1 if respondents were willing to use MRW to irrigate food crops.

We considered the responses from the following questions as our independent variables: primary role in the greenhouse operation (Question 6); primary customers for greenhouse products (Question 8); primary water source (Question 9); whether any water shortage(s) was experienced over the past 5 years (Question 10); whether water is treated to improve its quality before it is used (Question 11); use of recycled water in greenhouse operation (Question 15); the approximate annual total greenhouse operation net income (Question 29); perceptions of recycled water, tap water, and treated wastewater (Question 22); and motivations and limitations on the use of MRW to grow food crops (Question 26). An additional independent variable was created from size range (Question 20) referring to the square footage floor area of greenhouse operation (1 = than 100,000 ft², 2 = 100,000 to 499,999 ft², 3 = 500,000 to 999,999 ft², 4 = 1 million to 5,999,999 ft², 5 = 6 million to 9,999,999 ft², missing value = no responses).

Statistical analysis

To test whether there exists a statistically significant relationship between each independent variable and the response variable (i.e., greenhouse growers’ willingness to use MRW on food crops), we conducted chi-square tests of independence using a 0.05 level of significance.

A logistic regression model was used to estimate the association of one or more independent variables with the binary response variable, greenhouse growers’ willingness to use MRW on food crops. The logistic regression permits evaluation of the relationship of each independent variable with the response variable while controlling for the effect from other independent variables in the model. An initial logistic regression model was built with all potential predictors and then stepwise Akaike information criterion (AIC) variable selection was used to determine the final model. The exponentiated logistic regression coefficients in the model can be interpreted as odds ratios. Odds ratio explains the odds change for willingness to use MRW on food crops when comparing a certain group of growers based on one predictor vs. a reference category. One way to quantify how well a logistic regression model classifies data is to calculate the area under the receiver operating characteristic (AUC). The value for AUC ranges from 0 to 1. The higher the AUC score, the better the model is at classifying observations into classes. According to Hosmer et al. (2013), an AUC greater than 0.8 is considered excellent discrimination.

Chi-square tests

Using chi-square tests, we found 12 independent variables significantly associated with a willingness to use MRW on food crops (Table 1). Three of these variables captured respondent concerns over water quality that made greenhouse growers less willing to use MRW on food crops. Specifically, concerns over poor quality of MRW and lack of trust in regulatory oversight over the quality of MRW limits growers’ willingness to use MRW. Conversely, when growers trusted the provider that the quality of MRW is suitable for the intended use, they were more willing to use MRW on food crops compared with those who did not trust the quality (chi-square = 6.61, P < 0.01).

Table 1.

Chi-square tests of independence between each survey variable with willingness to use municipal recycled wastewater (MRW) on food crops, based on 268 responses to the Greenhouse Grower survey (McOmber et al. 2021; Supplemental Material).

View Table

Capital or market-based incentives and growers’ primary customers were also significant in affecting grower willingness to use MRW on food crops in the bivariate analysis. We found that growers who are motivated by MRW-focused monetary incentives or operational costs were more willing to use MRW on food crops. We also found that growers who thought that the cost of capital investments (i.e., to enable use of MRW) limits MRW use. Lastly, our results show that growers whose primary customers are restaurants, retail, or farmer’s markets (compared with wholesale) were less willing to use MRW on food crops.

Finally, growers’ knowledge about others’ experiences using recycled water and growers’ environmental concerns were also significant in the bivariate analysis. We observed that greenhouse growers who were motivated by the experience of other greenhouse growers using MRW within the United States were more willing to use MRW on food crops. However, greenhouse growers were less willing to use MRW on food crops when they thought the experience of other greenhouse growers globally using MRW limits use. Environmental and water security concerns also appeared to be significant in the model. Greenhouse growers who thought environmental impacts motivate use of MRW were more willing to use it on food crops, whereas growers who thought that MRW use and its effect on water security limits use were less willing to use MRW on food crops.

Logistic regression results

When controlling for multiple variables in the logistic regression model, five variables are significantly correlated with a willingness to use MRW (Table 2). First, controlling for other variables in the model, the odds that a grower is willing to use MRW on food crops increased by 225% if the quality of MRW motivated use compared with limited use or neither limited nor motivated use. Similarly, the odds that a grower was willing to use MRW on food crops increased by 142% if they perceived that treated wastewater was not disgusting, compared with those who thought it was disgusting. Third, if growers were motivated by others’ experiences using MRW globally, the odds of being willing to use MRW on food crops increased by 173%, compared with growers who expressed others experiences globally limited or neither limited nor motivated MRW use. Fourth, the odds that a grower is willing to use MRW on food crops increased by 125% if they had no experience with water shortages, compared with those who had experienced water shortages. Finally, the odds that a grower is willing to use MRW on food crops decreased by 64.8% if they did not use recycled water (a more general term for reused water such as from onsite ponds) in their greenhouse operations, compared with those who did use it.

Table 2.

Results from the logistic regression model on determining factors of willingness to use municipal recycled wastewater (MRW) on food crops, based on 268 responses of the survey (McOmber et al. 2021; Supplemental Material).

View Table

Water quality

Growers’ positive perceptions about the quality of MRW appear to be important in the model and does increase the odds of growers’ willingness to use MRW to irrigate food crops. Further, as Savchenko et al. (2018, 2019) found, public perceptions of recycled water quality also matter for consumer choices. In turn, growers seem to anticipate consumer perceptions about the quality of MRW, and it seems to impact their willingness to use it on food crops in their greenhouse operations. Growers who sell primarily to wholesale distributors, where the end user is several steps removed from the grower, were more willing to use MRW on food crops. This was in comparison to growers who had more direct contact with end users through farmers markets or restaurants and whose perceptions of MRW might have a more direct impact on their sales.

Direct and indirect experience with MRW

We asked a series of questions to understand how direct or indirect experience using MRW influences growers’ willingness to use it. Growers who lacked experience using MRW were less willing to use it for irrigating food crops in their greenhouse operation. Lack of prior experience using MRW may have negatively influenced perceptions about quality and limitations in implementing the practice. Instead, experience and learning can shape positive and negative perceptions about MRW and increase growers’ willingness to use it within their greenhouse operations (McOmber et al. 2021). However, it is not just direct experience that matters for growers. Indirect experience, through other growers, can also increase growers’ willingness to use MRW. Results indicate that growers who knew about others’ experiences using recycled water throughout the world were more motivated to use it. The influence of global experience may reflect that MRW is more widely adopted and used in agriculture internationally, particularly in countries such as Australia, Italy, and Israel (Craddock et al. 2021; Radcliffe and Page 2020), and that growers may be more familiar with use internationally than they are of other growers using MRW in the United States. Although knowledge of global experiences with MRW remained significant in the regression, knowledge of US growers’ experiences with MRW did not.

Water security concerns

Demand for water and climate change threatens global agricultural water supplies (Mekonnen and Hoekstra 2016), and greenhouse growers are increasingly concerned about water security in terms of quantity and availability (White et al. 2019); yet in our bivariate analysis, we observed that these concerns limit rather than motivate recycled water use. In part, this may be because respondents were defining “water security” differently from definitions in the literature, as we did not provide a definition for the term in the survey. The effect on water security to motivate MRW use among growers was not significant in the regression. Rather, in the regression, not experiencing water shortage motivated growers’ willingness to use MRW. This may suggest that greenhouse growers who do not feel water insecure are willing to experiment with alternative water supplies, whereas those who do feel water insecure are more risk averse. This explanation is consistent with our findings that direct or indirect experience and perceptions of MRW quality (perhaps garnered through that experience) motivate growers’ willingness to use MRW.

Beyond education: making a case for deep engagement to improve MRW uptake

Our research suggests growers’ positive perceptions of MRW quality relate to their direct and indirect experience using MRW and that these together are the most important factors motivating growers’ willingness to adopt MRW in agricultural practice. Personal observation and peer acceptance can be a critical factor in influencing adoption of new innovations (Rogers 1995). Experience and knowledge about water conservation techniques can influence perceptions (Warner et al. 2018, 2020), which, in turn, can increase growers’ willingness to adopt MRW within their greenhouse operations (McOmber et al. 2021). Our findings provide supporting evidence to a growing number of studies that suggest education about water conservation techniques, including alternative water sources like MRW, alone will improve their use (Dery et al. 2019; Lamm et al. 2017; Warner et al. 2018, 2020; White et al. 2019).

Education about MRW absent experience (direct or indirect) may be insufficient because negative perceptions about MRW may be multilayered and difficult to change. Positive perceptions created through trialability and observability (Rogers 1995) with MRW may help to support the intended impact of educational efforts regarding MRW and support behavioral change for adoption within the greenhouse operation. For example, although most growers did not, themselves, think that MRW was “disgusting” or “contained bacteria,” they may be concerned that their customers do (Savchenko et al. 2018, 2019), which may limit their willingness to use MRW. Besides worrying about consumers’ negative perceptions, growers may need to be reassured they can trust the quality of MRW. Increasing knowledge on testing, regulating, and producing MRW could increase adoption and continued use, as observed by Warner et al. (2018). Trusting regulatory authorities’ ability to ensure MRW quality echoes findings from Po et al. (2003), which showed that growers’ skepticism of the testing process and distrust of the responsible authorities providing MRW impeded its use but that these conditions improved when growers increased knowledge and had the opportunity to gain firsthand experience with MRW.

Efforts to improve MRW use among growers should center experience as a starting point for education of MRW. Such programs should include a process of iterative, deep engagement involving growers, extension agents, and regulatory officials that is designed to facilitate experiential and social learning. In this way, these efforts may begin to address knowledge gaps and concerns about MRW quality, while also working to build trusting relationships between all involved. Po et al. (2003) found that deep and continued engagement with growers through their participation in community-based committees provided an opportunity for growers to share their concerns and better understand quality control testing measures used to ensure the safety of recycled water. Deep engagement provided critical iterative feedback loops that allowed for greater attention to growers’ diverse and evolving quality concerns and targeted informational needs in response to their specific concerns. Deep engagement among growers, water authorities, and intermediaries, such as extension agents, create critical pathways for improving adoption of new technologies. Huang et al. (2016) for example, provided insight into how deep engagement is important to changing behavior among the public when perceptions are difficult to change.

They found that participation in extension efforts improve public perceptions of water quality. By demonstrating “personal relevance” to the participant, educators can tailor information to address concerns and experiences with water quality which may be shaping diverse perceptions of recycled water. They find that this engagement is important for overcoming issues of cognitive dissonance around past experiences with water quality among the public. These lessons can also be extended to growers; by identifying points of informational intervention and support around topics of personal relevance, grower perceptions of MRW and ultimately adoption may be improved. Beyond targeting extension outreach to growers’ experience with quality problems, experiential learning opportunities through simulation may also be an effective tool in facilitating learning and acceptance of recycled water practices. For example, Gottlieb et al. (2022) used an online, interactive simulation platform to help growers better understand the benefits (return on investment) of recycled water use as it pertained to each growers’ greenhouse operational needs. Providing this kind of tailored information, directly relevant to the individualized needs of growers in the context of deep engagement, could be a powerful approach to helping the bridge the gap to increased recycled water use in the agricultural sector.