Participatory Farmer Monitoring on Nitrate Loss: Using Farm-Scale Data to Improve Nutrient Management and Water Quality

SARE Results and Discussion

Research Questions and Hypotheses

We hypothesize that participants will not be familiar with what levels of nitrate loss are common for their farm, watershed, or in non-agricultural watersheds or have clear expectations on what times of the year nitrate retention is most or least likely.

This proposed project will address two research questions. First, how does the availability of nitrate data at the farm scale inform farmers’ perceptions on whether and how they can improve nitrate retention? We hypothesize that if farmers underestimate the amount of nitrate they are losing, they will express an expectation that they can reduce their future nitrate losses. Second, based on a farmer’s management priorities and stated rationales, what practices do they indicate they will change following the study? For the second question, we hypothesize that if they under-estimate their nitrate losses they will identify feasible, low-cost management changes that they can make.

Results

Over the course of three years, we had the participation of up to 27 farmers, with 18 farmers participating in on-farm tile drainage sampling for all three years. Farmers were located across 10 counties in central or southern Indiana (Clay, Dubois, Gibson, Knox, Hendricks, Jennings, Posey, Spencer, Vandenburgh, and Warrick).  Overall, we collected 929 water samples from 2020-2023 (this includes a pilot year for eight farmers that remained through 2023). Farmer collected an average of 12 samples per year, while ranging from a minimum of two samples to a maximum of 39 samples for a single year. There were 45 individual tile sites that drained on average 20 acres. The large drainage area a tile drained was 120 acres, while the smallest was 3 acres.

We conducted interviews with farmers at the beginning of the process, before they began sampling, and at the conclusion of the project, when all sampling had been completed. We also maintained informal check-ins in the winter with updated numbers on their drainage loads and nitrate concentrations for their farm and for the group overall. We created graphs to show farmers their individual nitrate concentrations (in parts per million) and instantaneous daily nitrate loads (in pounds per acre). The mean average Nitrate concentration from all samples was 5.36 ppm, while the median was lower at 4.88 ppm. The lowest concentration in any sample was 0.12 ppm, while the highest was 57 ppm. Farmers were also asked to record discharge rates, which was provided for a majority of samples, but not for all samples, so estimates were needed to fill in for missing values. Discharge samples represent a point-in-time estimate rather than continuous monitoring, which were then scaled up to represent a 24-hour instantaneous daily load (pounds per acre). Discharge values had a mean average of 0.94 liters per second (median 0.66 L/S), and ranged from a minimum flow of 0.01 L/S to a maximum of 3.84 L/S. The mean daily load was 0.99 lbs./acre (median of 0.64 lbs./acre). The smallest load was 0.01 lbs./acre, while the largest load was 12.9 lbs./acre.

Hypothesis 1: Unfamiliarity with Nitrate Loss

For our first hypothesis, farmers in interviews expressed nearly universally that they did not know how much nitrate they might be losing or how to even estimate what level might be acceptable from a management standpoint. When asked to give a percentage of the overall fertilizer their estimates ranged anywhere from less than one percent to as high as 40%. A plurality of farmers estimated losses between 5-10%. In this study, we are not able to accurately measure this overall percentage of loss, but we asked the question to elicit farmers’ beliefs to provide an orientation point. We also interviewed extension agents (n=18) and scientists (n=12) the same question as a point of comparison.  The minimum amount of loss estimated by farmers was 3.8% on average, 3.5% by scientists, and 10.2% by extension agents. The maximum amount of loss estimated by farmers was 14.6% on average, 44.6% by scientists, and 53.3% by extension agents.

Q1: How will field-scale data influence farmers’ management?

On the first research question, farmers responded to the concentration and load data with curiosity but offered few interpretations of the data’s meaning. For example, we provided two points of reference for them to compare the data, both on the same graph. The first point of reference between the background nitrate concentrations in the Wabash River Basin, based on previously published U.S. Environmental Protection Agency reference conditions, which were <1 ppm. The second reference point was the 10 ppm, which US EPA uses as the maximum level for nitrate in drinking water under the Safe Drinking Water Act.

Individual mean averages ranged from 2.66 ppm to 10.97 ppm (only two farmers had mean averages above 10 ppm). Overall, 13% (121 samples) have concentrations below 1 ppm. The vast majority of the samples were between 1-10 ppm, with 38% (351 samples) between 1-5 ppm and 40% (376 samples) between 5-10 ppm. The remaining 9% (81 samples) exceeded the 10 ppm level.

Farmers mostly viewed the level of concentrations, which are predominantly under the 10 ppm, as being an acceptable amount, while very few farmers commented on the number of samples below the ecological reference condition, nor made connections with how the majority of samples could contribute to water quality impairment. Farmers also comments on their own samples if those samples were higher than 10 ppm, as those appeared to be outliers when compared to the data overall. Some farmers treated samples between 1-10 ppm as being in an acceptable zone, with those exceeding 10 ppm are problems, but which seemed infrequent compared to the overall number of samples.

Farmers did not offer much in the way of reaction to the daily load levels. These were presented to highlight the potential effect of rainfall on losses throughout the calendar year. Farmers commented during follow-up conversations after the growing season each year that they were surprised by the amount of water flowing through the tiles or by the lack of discharge during the summer and fall from dry weather.

When asked whether the farm-scale data had changed their level of concern about the impacts of their nutrient management, some expressed that the data reaffirmed their concern but indicated that the impacts were not as bad as they had feared, while some expressed that the data indicated that their impacts were lower than they had anticipated.

Q2: What Management Practices Will Farmers Change?

On the second research question, nearly all farmers did not express plans to change their management approaches. This was due in large part to the sample of farmers that wound up participating in the study. Nearly all of the farmers used reduced tillage or no-till, cover crops, and split application of fertilizer. Many of the participants also use variable-rate application as well.

We collected data on tile drainage size, spacing, and depth, in-field management practices, such as nutrient amount, application types, fertilizer types, and cover crops. Due to the messy nature of the data and when farmers decided to collect the data, there were no discernable patterns in the data from which to draw conclusions about the (in)effectiveness of different management practices. Farmers did describe what practices they felt generally farmers should and should not do. These comments nearly exclusively focused on in-field management and cover crops.  

In addition to sharing these data with farmers, we also used the Science Assessment from Iowa’s Nutrient Reduction Strategy in the exit interviews. We include 10 of the Science Assessments’ strategies for reducing nutrient loss to balance in-field, edge-of-field, and land use options. In general, the Science Assessment shows greater reductions in losses from edge-of-field and land use options than from in-field options, which are also effective. We asked farmers to rank these practices from most to least effective for reduction. Farmers more often than not ranked in-field management practices (4Rs) as the most effective, while edge-of-field practices, such as buffers, as less effective.

Additional Research Question: What role do social norms play in the adoption of nutrient reduction practices?

An additional part of the study that emerged following the initial proposal was a survey of farmers in Indiana, Illinois, Maryland, and Pennsylvania on factors influencing their adoption of three nutrient-reduction practices: no-till, cover crops, and not applying fall fertilizer. The survey was sent out in the winter of 2022 to 10 counties in each state, with 585 completed surveys returned. The survey asked farmers around social normative, management, demographic, and farm characteristic questions. We constructed three logit models with the survey data to look at what factors were statistically related to adoption of each of three practices.

For descriptive norms (e.g., common behavior), we asked farmers five-point Likert-scale questions about how many farmers use each of the practices were in their county, ranging from “almost none” to “nearly all.” For dynamic norms (e.g., anticipated to be common in the future), we asked farmers whether the practices had become more common in their county, again on a five-point scale. For injunctive norms (e.g., socially acceptable behaviors), we asked how supportive or opposed family, neighbors, landlords, agronomists, and extension agents would be on each practice. For management, we asked how confident farmers were that they could implement each of the practices (e.g., self efficacy), how effective each of the practices were for a variety of factors (e.g., response efficacy), and then standard questions about age, gender, farm size, income, and education. For the injunctive norms and the response efficacy variables, we created an index to measure the variables.

In our logistic regression models, we found that descriptive norms were statistically significant for no-till and not applying fall fertilizer, but not for cover crops. Dynamic norms were only statistically significant for cover crops. Injunctive norms were not statistically significant in any of the models. Self-efficacy and response efficacy were statistically significant in all three models, as was income. Age, gender, education, and farm size were not statistically significant in any of the models.

Discussion

The farmers in our study are generally using best practices for nutrient management but are still losing fertilizer at concentrations that can be problematic depending on the amount of discharge. Field-scale data provided an interesting opportunity to see the types of concentrations and discharge levels occurring, given the relatively limited amount of data that is available from other studies or through publicly available monitoring. We provided three basic points of reference for farmers to assess their nitrate losses: comparison with other farmers in the study; ecological reference conditions; and Safe Drinking Water Act standards. While these are useful, they also proved to be limited in how much farmers could interpret from the data as we presented it.

One of the difficulties with providing points of reference was that farmers tended to interpret those points of reference as showing that things were not necessarily problematic for water quality rather than seeing that no matter their management practices that some amount of synthetic nitrogen fertilizer is difficult to keep from leaching through tile drains. Farmers had a tendency to depict the higher reference point as a limit to stay under while not paying equal attention to the lower limit showing agriculture’s general impact. This is not an entirely surprising outcome. Farmers tend to be aware of and concerned by regulations that would affect how they are allowed to use fertilizer. The 10 ppm limit under the Safe Drinking Water Act was known by more than half of the farmers in the study but not by all. No farmers were familiar with the reference condition of <1 ppm for the Wabash River Basin. Farmers also, when asked about the different practices that are most effective at reducing nitrate leaching, tended to focus on practices they were more familiar with, such as in-field nutrient management, than approaches that have been demonstrated to have greater impacts on reduction, such as perennial crops or vegetative buffers.

Despite the large number of samples, farmers in exit interviews still talked about the challenges of other non-farm sources of nitrate impairing water quality. The difficulty with tracking nitrogen throughout a farm suggests a couple of promising options for future research studies. First, while this study tended to look at the aggregate picture from sampling, future studies could have farmers develop testable hypotheses around individual tile drains and how those tile drains likely respond to rainfall differently. This study did not attempt to undertake a full nitrogen budget for each farm. Farmers that participated were well aware of the different ways in which nitrogen could leave their farm, through surface runoff, volatilization, and denitrification, in addition to subsurface leaching. For farmers to know how much nitrogen is leaving their farm, accounting for these pathways as well as biomass is crucial to help farmers know where and how their management could further improve. Multiple farmers described a no longer active program called In-Field Advantage that provided information on tissue samples. A complete budget of several working farms, particularly to compare different management types, would be a valuable way to help farmers see where interventions could lead to further improvements in nitrate retention. Education campaigns on the level of nitrate in ecosystems would be valuable to help farmers consider why even seemingly low levels of loss can lead to large downstream problems. Efforts to use field-scale data to inform management decisions would benefit from a stronger baseline understanding of what levels of water quality impairment are problematic for the environment and other industries.

For the survey results, we think the most notable finding is that different norms can matter for different practices. We think the implications of the differences between descriptive and dynamic norms is that no-till and applying fall fertilizer may be more settled practices where farmers have largely decided whether they are going to use those practices or not, while cover crops are becoming more popular and so more farmers are paying attention to see whether it is becoming more popular as they consider trying it out. The importance of self-efficacy and response efficacy also point to a critical aspect of adoption. Both are unsurprising findings as training and knowledge to be able to implement a practice, as well as a belief that the practice to be implemented is useful, seem essential to the basic logic of adoption. Where norms may matter is in getting more farmers interested in learning about new conservation practices and being connected to individuals with knowledge to help them with implementation or encouragement to seek out technical assistance (or vice versa).