Introduction

The adoption of agricultural conservation, or best management, practices is crucial to enhancing water quality and other ecosystem services in agri-environments (Lintern et al. 2020). Practices such as cover crops, riparian buffers, and no-till, among others, can reduce soil erosion, curb nitrogen and phosphorous runoff into waterways, and sequester carbon (Kaye and Quemada 2017; Skaalsveen et al. 2019; Cole et al. 2020). To facilitate farmers’ adoption of such practices, governments and non-governmental organizations often offer financial and technical assistance (Reimer et al. 2018; Bowman and Lynch 2019; Berthet et al. 2021; Pathak et al. 2021). Technical assistance is provided by conservation practitioners, who, among other things, meet with farmers one-on-one to help them plan, design, apply for, and implement conservation practices (CBC 2017). The trust and relationships between these practitioners and farmers, along with the level of financial incentives, are often characterized as primary influences on farmers’ decisions to adopt conservation practices (Eanes et al. 2017; Brock et al. 2018; Ranjan et al. 2019, 2024; Morris and Arbuckle 2021).

The effectiveness of technical and financial assistance is commonly assessed by examining correlations between conservation practice adoption, measured as practices/projects in the ground, and a set of covariates (e.g. Mishra et al. 2018; Prokopy et al. 2019). These covariates often include some measure of technical or financial assistance, along with socio-demographic and production characteristics (e.g. Jäger et al. 2022). Such research has suggested that factors including access to varying levels and types of financial incentives (Mezzatesta et al. 2013; Fleming 2017; Claassen et al. 2018; Plastina et al. 2018), attendance at outreach events (Singh et al. 2018; Read et al. 2021), and one-on-one assistance (Lee et al. 2018) are effective tools for promoting the adoption of conservation practices. In such studies, adoption is most often measured as a binary variable representing whether a farmer adopted one or more conservation practices (Reimer et al. 2014). Other measures of adoption include whether adoption persists over time (Dayer et al. 2018; Defrancesco et al. 2018) and the extent of land over which adoption happens (Pannell and Claassen 2020).

This correlational approach that measures adoption as a binary variable forms the basis of many frameworks aimed at understanding and increasing conservation practice adoption rates (Read and Wainger 2023). Behavioral frameworks often test the influence of different personality traits and preferences, social interactions and pressures, and perceptions of costs and benefits of adoption (Dessart et al. 2019). For example, several studies have shown that those with greater environmental concern are more likely to report intentions to adopt conservation practices (Floress et al. 2017; Palm-Forster et al. 2017). Segmentation frameworks aim to discover how adopters and non-adopters systematically differ, so that adopters can be more quickly identified and interventions can be specifically tailored to different groups (Prokopy et al. 2019). Reviews of such work have identified that larger farms and greater education are associated with adoption (Prokopy et al. 2008). Economic frameworks have evaluated different policy designs aimed at reducing transaction costs and increasing additionality, for example, and have largely found that easier contracts and larger payments result in greater adoption (McCann and Claassen 2016; Fleming 2017; Claassen et al. 2018).

One important drawback of using such correlations to assess the effectiveness of financial and technical assistance is that they often obscure the sequence of events that lead to adoption, which can potentially confound their results (Read and Wainger 2023). That is, deciding whether to implement conservation practices is only one decision in an adoption process (Rogers 2003). Everett Rogers’ foundational work (Rogers 1961) and work from other fields (Ettlie 1980; Weinstein et al. 1998; Prochaska 2008) have long recognized that the adoption of new behaviors or practices is a process with stages. For example, the precaution adoption process model used in medicine and public health describes how people move through six stages from ignorance to action, and suggests different interventions for each stage to promote movement to the next (Weinstein et al. 1998, 2020). Recently, some studies have emerged applying such an understanding to the adoption of agricultural conservation practices. These studies have highlighted how farmers move through a set of sequential stages, including decisions about whether to learn about conservation practices, trial them, fully adopt them, and continue to use them long-term (Brown et al. 2017; de Oca Munguia et al. 2021). Conservation practitioners go through similar stages in their guidance, including building relationships with farmers, diagnosing resource concerns, influencing farmer intentions to adopt, translating those intentions into actions, and ensuring stable use of the adopted practices over time (Rogers 2003). Some work has shown that the significance of the covariates commonly used to predict adoption, (e.g. farm size, risk preferences, and education levels), varies across adoption stages (Doran et al. 2022), suggesting that treating adoption as a sequence will clarify the effectiveness of different techniques and their timing for promoting adoption.

Treating adoption as a process with distinct stages requires testing intervention effectiveness and expected outcome per stage. For example, Braddick (2017) documented variation in how helpful Australian farmers found different extension methods at different stages of adoption, with field days being helpful when first considering the practices, decision support tools being helpful when trialing practices, and financial incentives being helpful when implementing the practices. Some experimental studies have attempted to measure effectiveness within a stage by examining an intervention’s effect on the most immediate outcome of that adoption stage, rather than on some measure of final adoption. Examples of such work include the effect of different message framings on farmers’ response to mailings (Czap et al. 2019; Reddy et al. 2020), and different incentive designs on farmers’ willingness to accept contracts (Palm-Forster et al. 2017). Implicitly, these experimental studies assess the effectiveness of interventions at encouraging farmers to move from one adoption stage to the next. However, these studies rarely make explicit connections to other stages of adoption, and, as yet, there is little evidence to suggest what practices are effective at different stages for motivating farmers to continue toward long-term adoption.

In this paper, we present the results of two studies that explore variation in the effectiveness of engagement techniques across the agricultural conservation practice adoption process. First, through semi-structured interviews, we explored the perspectives of conservation practitioners (CPs) in the Chesapeake Bay Watershed (CBW), USA, on what they have found to be effective at engaging farmers at different adoption stages. Our goals for these interviews were to outline the main adoption stages and to summarize what techniques CPs found effective at moving from one stage to the next. Second, we conducted an online experiment to test the effectiveness of an early-stage intervention on two subsequent outcomes in the adoption process. Specifically, we tested the effectiveness of including before-and-after pictures of conservation practice adoption in advertisements for financial assistance on two sequential outcomes. These outcomes were whether viewers clicked the ad to receive more information and, if so, whether they personally contacted a CP. Through these two studies, we aim to build a foundation for further investigation into how intervention effectiveness varies by stage in the adoption process. Understanding what techniques CPs find effective under different circumstances, and experimentally testing such interventions, can provide researchers with the empirical grounding necessary to more accurately inform CPs’ work (Ranjan and Witter 2020; Morris et al. 2021).

Study 1: Interviewing Conservation Practitioners about Effective Techniques Across Adoption Stages

The aim of the first study was to understand conservation practitioners’ (CPs) perspectives on what engagement techniques were effective at different adoption stages. We first describe the method used to collect and analyze data, and then present results concerning stages in the adoption process, effective techniques at those stages, and an illustrative example of those techniques in practice.

Methods

Study Site

We interviewed CPs working to improve water quality across the Chesapeake Bay Watershed (CBW), which spans 166,000 sq. km. in the eastern United States. While still a highly productive ecosystem, water quality, particularly low dissolved oxygen and excessive algal growth, have degraded aquatic habitat (Kemp et al. 2005). A major contributing factor to this degradation is runoff of agricultural fertilizers, in addition to urban runoff and atmospheric deposition (Hagy et al. 2004; Fisher et al. 2006). In 2010, the US federal government set a Total Maximum Daily Load (TMDL) for nitrogen, phosphorus and sediment for the CBW and established a 2025 deadline for reaching these caps (Linker et al. 2013).

Implementing agricultural conservation practices can reduce farm nutrient and sediment runoff at the watershed scale (Kleinman et al. 2019; Sabo et al. 2022). However, because of the expense and complexity of many practices, financial and technical assistance is often necessary for farmers to afford adoption and to navigate the adoption process. A range of CPs in the public and private sectors provide such assistance (CBC 2017). Most public CPs are staffed in federal Natural Resources Conservation Service (NRCS) offices or local Soil and Water Conservation Districts (SWCD), which have offices in most counties across the CBW. Often, public CPs work closely with those in non-profit or private companies, who are typically hired to write and sometimes implement approved conservation plans. Non-profit CPs often bring separate financial assistance that allows more flexibility in practice design specifications and quicker installation timelines than government-funded practices. Excluding private CPs, most technical assistance is free to farmers.

There are approximately 83,000 farms in the CBW. The average farm size in each of the three main Chesapeake Bay states (Pennsylvania—145 acres, Maryland—159 acres, Virginia—187 acres) is smaller than the national average of 463 acres (NASS 2022). This smaller acreage means that CPs in the CBW have more landowners to interact with per farmland area than elsewhere. Soybean and corn production make up the majority of agricultural land use, though vegetables, fruit, and other crops are prevalent. Poultry production is substantial and has been expanding, especially on Maryland’s Eastern Shore. The adoption rate for some conservation practices, like cover crops, is higher in the CBW than elsewhere in the US (Wallander et al. 2021), but other practices thought to be highly effective, such as riparian forest buffers, are not being widely adopted (CBPWT 2022). While some regions of the CBW have seen measured reductions in nitrogen (Ator et al. 2020; Sekellick et al. 2023), the watershed is not on track to meet the 2025 TMDL, suggesting that conservation adoption rates need to accelerate.

Data Collection

We sampled an initial group of CPs to achieve equal representation in the three main CBW states and diversity among public, non-profit, and private organizations. We constructed a list of practitioners (N = 9) across the watershed using our own and our colleagues’ contacts, and used snowball sampling to request suggestions for additional potential participants from previous participants (final sample size, N = 30). Practitioners were contacted by email to schedule semi-structured phone interviews. Each interview began with questions about how they came to their current position. We then asked for descriptions of one successful and one unsuccessful experience working with a farmer on conservation practices. Next, we asked each practitioner to look at an outline of adoption stages, which had been emailed beforehand (Fig. 1A). The diagram represented different adoption stages at which a CP and farmer would interact. The stages were distinguished by farmers’ decisions over which CPs would have some influence. After explaining the diagram, we asked if there were any changes that should be made to increase its accuracy (see Fig. 1B for revised diagram), and what they find to be effective at encouraging movement from each stage to the next. Thus, we considered effective techniques to be those that CPs identified as helping to move farmers to each subsequent stage. We ended each interview with a broad question about what the practitioners thought needed to change to make technical assistance more effective at increasing conservation practice adoption rates (see Supplementary Material for interview guide). With respondents’ permission, we recorded, and later transcribed, audio from all interviews.

Fig. 1
figure 1

Diagram of adoption stages and outcomes leading to the next stage before (A) and after (B) conservation practitioners’ feedback

Full size image

Data Analysis

We largely followed Deterding and Waters’ (2021) approach to flexible, open coding of the interview transcripts. Two coders marked where each new question began. We then identified and coded each suggested change to the diagram of adoption stages, iteratively refining codes to see if new suggestions were novel, the same, or a variant of a previously coded suggestion. We used these codes to revise the diagram and adoption stages. We then identified each technique respondents said they used when working with farmers. We used the same iterative process for coding techniques as we did changes to the diagram, revising codes until reaching a high inter-rater reliability (mean Cohen’s kappa = 0.908; Hruschka et al. 2004). To be considered a technique, the respondent had to discuss something that they had done or said, or had seen done or said by another practitioner, to a farmer. We did not include general statements about farmer preferences as techniques. For each technique, we coded at what adoption stage it occurred and whether the respondent characterized it as effective or not for proceeding to the next adoption stage. We considered techniques to be effective if respondents said that they helped farmers to progress through the adoption process or when they said that not doing a technique hindered such progression, a circumstance we considered as a ‘double negative’. We considered techniques to be ineffective when respondents characterized their use as hindering progress through the adoption process. All data were analyzed in NVivo 12 (QSR International Pty Ltd 2020).

Results

We conducted 30 interviews (mean duration 63 min) with 11 practitioners working in Pennsylvania, 9 in Maryland, and 10 in Virginia. Five respondents were women and 25 were men. Five respondents worked in federal agencies (NRCS and FWS), 8 in soil and water conservation districts, 3 in university extension, 8 in non-profits, and 6 in private consultancies.

Adoption Stages

According to respondents’ suggestions, we made several changes to the diagram of adoption stages, the most important of which changed the sequence of steps through which farmers progress (Fig. 1A). We added an arrow looping from “Post-Implementation” back to “One-on-one” because many CPs said that they work on different practices with the same farmers over time. Similarly, we changed the flow from the “Mass Outreach” stage to go both to the “Outreach Event” and “One-on-one” stages. We also we added door knocking as a technique that then begins the process at the “One-on-one” stage, rather than the “Mass Outreach” or “Outreach Event” stages. Many respondents mentioned that their interactions with farmers often start at the “One-on-one” stage because the farmer was referred to the practitioner or sought them out on their own, or because the CPs prioritized working with certain farmers. These changes resulted in an outline of the adoption process with 7 stages (Table 1).

Table 1 Description of adoption stages after conservation practitioner’s feedback
Full size table

Effective Techniques

We identified 74 unique techniques that practitioners use across different adoption stages. Most techniques were characterized as effective, with only 8 being described as ineffective by at least one practitioner. The majority of techniques (N = 60) were only mentioned at one adoption stage (Table 2), while the rest were mentioned at multiple stages.

Table 2 Number of respondents identifying effective/ineffective techniques mentioned at only one stage in the adoption process
Full size table

Of all stages, mass outreach had the fewest techniques and highest proportion of ineffective techniques mentioned by respondents. Mailers were the most often mentioned technique, but a third of respondents found them ineffective. Two respondents mentioned more specific techniques about targeting particular audiences with mailers. One reported that using a county list of farmers to target recipients was ineffective, while the other said that writing personalized cards and letters to each farmer in their database was effective.

Most outreach event techniques were characterized as effective by the respondents. Six respondents suggested that presenting at other organizations’ meetings, rather than hosting their own conservation event, was effective. Several such respondents specifically mentioned partnering with organizations who host meetings that are required for some farmers to attend, such as university extension events that give required credits toward pesticide application permits and Farm Service Agency events required to receive crop insurance. Respondents said that these events tended to attract larger and more diverse audiences, helping them reach farmers with whom they do not normally interact. Demonstrations showing how to implement practices or the effect of those practices on soil and water were also mentioned as effective, as was having an experienced farmer, rather than a practitioner, as the main speaker for an event. The three ineffective techniques mentioned for outreach events pertained to environmental themes and framings, or hosting events about common practices about which most farmers already knew.

Half of the techniques that respondents mentioned for the one-on-one stage were also mentioned for the property visit stage (Table 3). Often, these stages occur at different venues, but have a similar purpose, which is for farmers to meet individually with a CP and get a sense of what practices might suit them best. The most commonly mentioned techniques were to focus on practices that match farmers’ interests and to summarize the process and timeline involved in applying for and implementing practices. No respondent mentioned any ineffective practices at either of these stages.

Table 3 Techniques used at multiple adoption stages
Full size table

Whereas some respondents had suggested de-emphasizing financial incentives during the previous two stages, discussions of funding featured in 2 of 7 techniques discussed during the drafting conservation implementation plan stage. In particular, one frequently mentioned technique was finding alternative sources of funding so that farmers’ expenses are reimbursed as much as possible, such as combining different cost-share programs or public and private funds. These techniques are important because at the end of this stage the farmer must decide which practices to apply for, a decision largely driven by available funding. Two other frequently mentioned techniques used at this stage emphasized the voluntary choices of farmers about which practices to implement. One respondent explained the importance of this technique saying, “It’s important for them [farmers] to understand that this is a voluntary plan. ‘Anything I put in here are just suggestions. You don’t have to do them.’”

Once funding has been secured, CPs collaborate with several partners to implement them and most of the effective techniques at this stage had to do with ensuring that everyone had a similar understanding of the project. These techniques included having a pre-construction meeting with all contractors and inspectors, double-checking the plan before installation, and ensuring that the CP was on the farm during construction. Other techniques sought to reduce farmer or contractor costs by shifting responsibilities to others. For instance, one CP from a non-profit organization said that, for installing riparian forest buffers, he organizes, “a volunteer event to get 50 to 80 people out on the farm and do a tree planting after the fencing is installed.”

Most of the techniques mentioned by CPs at the post-implementation stage aimed to ensure that the practices are maintained properly. These techniques included sending reminders to the farmers, ensuring that they have the proper resources, and providing some kind of social recognition, such as an award, to reinforce their decision to adopt and encourage them to maintain the practices in the future. The most frequently mentioned technique at this stage was to offset the burden of maintenance from the farmer to someone else, particularly for riparian forest buffers. This is similar to the technique from the previous stage of having volunteers install practices, except that at this stage they work to maintain the practices.

There were several techniques that respondents reported using at multiple stages in the adoption process (Table 3). Adapting to farmer time preferences was mentioned across four stages. One respondent mentioned using this technique at outreach events, which translated to hosting early morning breakfast events that are amenable to farmers’ schedules. During other stages, this meant quickly responding to farmers’ requests, or at property visits, respecting farmers’ time by balancing the speed and thoroughness of the visit. One respondent even mentioned allowing farmers to return to their work while the CP finished the property visit on their own.

Of all techniques, ‘Setting expectations for funding and roles’ was the most frequently mentioned, and pertained both to the Draft Implementation Plan and Implement Plan stages. Many respondents said that this involves getting full drafts of plans and cost-estimates before asking the farmer to sign any contract or apply for funding. The CPs suggested that setting these expectations helps both to prevent future misunderstandings and to demonstrate that cost-share contracts require responsibilities from both the farmer and the funding agency such that each have “buy-in” with the project. However, one respondent suggested that setting these expectations can backfire when doing so over-emphasizes the consequences of not adhering to the contract.

Compliance framings and profitability framings were mentioned at three stages each. The compliance framing aimed to present the choice of conservation practice adoption as one of reducing regulatory risk. This technique was mentioned most frequently among respondents who work in Pennsylvania, where the Department of Environmental Quality had begun inspections of conservation plans. One respondent explained that compliance framing emphasized how he was helping farmers, saying, “We come at it from the helpful hand… ‘We’re going to send Phil out to visit your farm and find out if your plan’s current, and help you get a plan so that you are compliant.’” Profitability framing was the most frequently mentioned framing, and emphasized how adopting conservation practices can increase farmers overall profit margins. Some respondents made this point by highlighting how using conservation practices could increase farmers access to customers who pay more for sustainably farmed products. Others pointed to how soil health practices, like cover crops and precision nutrient application, can reduce farmers fertilizer input costs. One respondent explained this point by saying, “Even though you might not be growing as many bushels an acre of corn, you’re making more money per acre, and that’s, from a business standpoint, ultimately what you are worried about.”

Example of Effective Techniques through Adoption Stages

An example from the interviews helps to demonstrate how these techniques are sequenced by a CP throughout the adoption process. The quotes below are verbatim from the interview, but we eliminated digressions, stop words (e.g. “you know,” “um,” etc.) and applied pseudonyms, as appropriate. The example includes all steps except the Mass outreach, Outreach event, and One-on-one stages, as the farmer was approached directly at their property through a door knocking. Techniques and stages in the adoption process are marked in [italicized brackets].

“My district manager… said, ‘There’s one farm…we have been very unsuccessful with. They don’t want to cooperate with us. It’s probably one of the worst situations in the entire district. And we just don’t know how to deal with them, how to reach them. And if you could get them on board, that in itself would be one of the biggest success stories that our district would have.’ It was that bad. It was really bad.

[Stage: Property visit] “So, I started trying to reach out to them – made a field visit…I was very lucky to have a DC [district conservationist] and an extension agent who had had dealings with this farmer, and they were able to make the introduction… [Technique: Trusted messenger]. We just kind of said, ‘Hey, we’re going to come out. We want to introduce you to [someone], she’s new here’ [Technique: Use friendly conversational tone]. So, I wasn’t really invited to look over the farm, but I had an opportunity to talk to him and say, ‘Hey, I have these programs if you’re interested’ And no interest whatsoever.

“About that time…we had the Ag Stewardship Act was coming about in Virginia… It allows citizens and/or agency to lodge a complaint against an agricultural operation and this is through our Department of Ag and Consumer Services, which is regulatory.

“So, they were doing some monitoring of some of the creeks…and this one particular creek…was showing high levels of fecal coliform. And so, they…called us up and said, ‘Hey, we think we know where all this fecal matter is coming from. And it’s this dairy farm.’ And it was the one I had just gone out to meet with, and they’re like, ‘You know, we’re gonna lodge an Ag Stewardship Act complaint.’

“So, I called up the farmer to let them know that the state was going to be lodging a complaint against their operation, which is not something we like to have to do. But I wanted to let them know that before this happened, ‘Let me onto your farm and let me see what the situation is so maybe we can head it off before the state comes in and tells you you have to do something, before you’re forced to do it’ [Technique: Compliance framing].

And so they did agree for me to come on the farm, and… I was shocked…Cows were standing knee-deep in mud and manure. They did not have any kind of lagoon or anything at the time, so all of the puddle water was point source just basically funneled right into the nearest draw that fed the creek that was coming up high.… And there was manure piled up everywhere. It was just awful.

But I did at that point get to talk to the owner, and started telling him about some of the programs that we had. And, I wouldn’t say that…I was warmly received, but I think that kind of started the conversation. So, if you want to say regulation maybe pushed him that way, that probably was the impetus for him to start working with us.

[Stage: Draft conservation implementation plan] So we did some initial analysis of the farm. We started saying, ‘Ok, these are some of the BMPs that you’ll need to put in place’ [Technique: Balance resource concerns with farmer interest]. We started giving him cost estimates of the BMPs [Technique: Set expectations for funding and roles], which he just totally said, ‘Oh there’s no way I can afford that. No way.’ But at least it got him thinking about what was out there. And eventually the Ag Stewardship Act complaint did come down the pipe. The complaint was formally lodged…so, at that point, the farm had six months to comply…

So, we went out, and again the [lead ag] officer, luckily he was local. So, they knew him, or knew of him. He had a farming background as well [Technique: Trusted messenger]. So, he and I went out and again talked over options and solutions [Technique: Educate farmer about design options], and between the two of us, we got them to agree to sign up for the cost-share. We spent, I think it was probably, a good year and a half – well beyond the six months – trying to come up with a plan that they were happy with that they could afford.

[Stage: Implement plan] But they eventually put in the lagoon. They put in a dairy loafing system, which [supports] probably upwards of twenty-three waterers. They got all kinds of exclusion fencing. They did all of this, or he did most of the work himself, which helped too, because they can bill us for their work, which provided the 25% offset. So, really, they had no out-of-pocket expense [Technique: Access alternative funding]. And so they put in the system. I don’t know if he would have shared at the time that he was happy with it…

[Stage: Post-implementation] About two years later, we decided to honor them with [an]… award, because they had done so much work [Technique: Social recognition]. And when we did the interview, you would have never known that he was kind of forced to do it. Everything was his idea, the greatest system next to sliced bread. And a lot of that came about because we let him be involved in the planning process. We didn’t come out and say, ‘You have to do it this particular way’ [Technique: Emphasize voluntariness]. We let him work with us to help us design it, and it took a long time, but he had a big part in designing it to meet his needs [Technique: Balance resource concerns with farmer interest]…After that award, they decided that they wanted to do more [Stage: One-on-one]. And so, they started to come to us for, ‘Hey, I want to extend my system. I want to do more grazing. I want to do it in my heifer lot. I want to do it for my beef. I want to do wildlife areas. I want to do upland conversions and cover crops.’ And so, it kind of snowballed from there to the point where today they are probably one of the nicest, most conservation-oriented farms in my entire district.

Study 1 Conclusions

The findings of Study 1 suggest that CPs use a diversity of techniques to promote agricultural conservation practices with farmers, and that these techniques vary greatly across the stages of the adoption process. However, CPs largely limited their discussion to techniques that they found effective. Because interviews alone provide no accurate way to compare the effectiveness of these techniques across stages, we proceeded with Study 2, an online experiment that measured these differences more precisely.

Study 2: Behavioral Test of Technique Effectiveness at Multiple Adoption Stages

The aim of the second study was to test whether, and if so to what extent, the effectiveness of outreach techniques promoting agricultural conservation practice adoption varies by adoption stage. To do so, we conducted an online experiment in Pennsylvania, for which we measured the effectiveness of the treatment on two sequential outcomes. Specifically, we placed an advertisement in an online farming magazine. The advertisement randomly rotated between either a treatment version, which showed pictures of farms before and after implementing conservation practices, and a control version, which had identical text and format but no pictures. We tested the effect of these pictures on (1) whether viewers clicked on the advertisement to visit a non-profit’s website offering technical and financial assistance for conservation practice adoption, and (2) whether those viewers who did click the advertisement filled out a form on the website requesting to meet with a CP.

In tandem with the goal of testing how intervention effectiveness varies by adoption stage, this online experiment also sought to address a shortcoming of mass outreach techniques noted in Study 1. Not only did CPs mention the fewest techniques for mass outreach but these techniques were most frequently characterized as ineffective. Experimental research has tested a wide variety of mass outreach messages, largely through direct mailings to producers (e.g. Czap et al. 2019; Reddy et al. 2020). One study tested ads promoting outreach events on Facebook, finding that business framings were most cost-effective overall, and that science and steward framings were more effective at engaging women (Witzling et al. 2023). Online information sources are becoming more commonly trusted by farmers (Rust et al. 2022) and have been shown to be cost-effective at recruiting potential adopters, though with many of the same biases as traditional recruitment methods (Burke et al. 2021, 2023). Thus, further building the evidence for what works in designing online advertisements could improve effectiveness of mass outreach techniques.

Although we did not find any usage by practitioners, some research has suggested that before and after photos of conservation practice implementation could be an effective tool to promote adoption during other stages of adoption. Exploratory results suggested that photos are particularly effective in improving farmers’ understanding of or comfort with practices that are unfamiliar or that substantially change the landscape (Schattman et al. 2019, 2020). Because photos increase the observability of conservation practices to potential adopters (Rogers 2003), farmers can better visualize the practices and understand the extent to which practice adoption will necessitate changing operations. Further, showing already installed practices may suggest that adoption is a social norm among their peers, and therefore a less risky behavior (Streletskaya et al. 2020). Although photos are infrequently tested within the broader conservation literature, two studies testing different visual frames (positive vs. negative pictures, and pictures of wildlife vs. people) could not detect any effect of the photos on zoo visitors’ donations to conservation (Salazar et al. 2021, 2022). However, one of those studies found that negative pictures increased engagement with conservation issues. These exploratory findings suggest that photos may be an important and cost-effective tool to increase the observability of conservation practices and promote their adoption.

Methods

Study Site

We conducted an online experiment in Pennsylvania, USA, of which nearly half the land area falls within the CBW. Pennsylvania is not projected to meet its 2025 nutrient and sediment reduction targets set under the TMDL (USEPA, U.E.P.A. 2022). Since the majority of nutrient and sediment runoff from Pennsylvania’s Susquehanna watershed appears to originate from agricultural land, insufficient implementation rates of agricultural conservation plans and practices are considered the main impediment to achieving runoff reduction goals (PA DEP, P.D. of E.P. 2022).

Data Collection and Experimental Materials

We designed our experimental treatment with input from CPs, many of whom participated in Study 1. During the interviews, 5 CPs said suggested that ‘Visualizing farm management outcomes’ (Section “Effective techniques”), or showing some media to help demonstrate how the practices would affect farm operations, at the Property Visit stage was an effective technique. We collaboratively decided that showing media would be a low-cost technique to test. But due to the need for a large sample size and to prevent overburdening CPs with the logistics of running an experiment during their property visits, we decided to test this intervention at the Mass Outreach stage. We asked CPs for illustrative before and after photos of conservation practices and searched online for additional photos. We then selected the practices and photos to test based on photo availability and the extent to which practices result in substantial change to the landscape or may be unfamiliar to farmers, where photos have been suggested to more strongly influence decisions (Schattman et al. 2019).

We designed ads that highlighted riparian forest buffers, grassed waterways, or stream crossings with livestock exclusion fencing. We chose photos of farms within the CBW watershed or that had features consistent with such farms. To control for the influence of photograph color and contrast (e.g. Bishop and Miller 2007; Probst et al. 2012), we color-matched the shadows and highlights of all photos in Adobe Photoshop. The entirety of the advertisements were clickable and a “contact us” prompt in the ads invited clicks (Supplementary Information).

The ad text was modified from pamphlets used by CPs at the Stroud Water Research Center (Stroud), a non-profit organization, and stated each practice’s farm and water quality benefits. All text and formatting for the ads was reviewed by CPs for clarity and accuracy and included Stroud’s logo (Fig. 2). Viewers who clicked the ads were taken to Stroud’s watershed restoration website (stroudcenter.org/restoration), which provided more information and invited visitors to contact Stroud by filling in a form or using the email and phone contact information provided. Stroud partners with government agencies and other non-profits across Pennsylvania to conduct scientific research and farmer outreach regarding conservation practice adoption. We chose Stroud as the ad contact because of their willingness to engage in research, non-governmental status, and minimal advocacy by the organization. While farmers tend to have less trust in environmental non-profit organizations than government agencies, farmers associations, and agribusiness representatives (Arbuckle et al. 2015), Stroud staff suggested that trust in their organization could be relatively strong because they do not engage in regulatory advocacy.

Fig. 2
figure 2

Treatment (top) and control (bottom) versions of the grassed waterway ads. See Supplementary Information for copies of all ads. Photo Credit: Mike Staton, Michigan State University Extension and Michigan Soybean Committee

Full size image

We placed the ads in LancasterFarming.com, an online agricultural magazine whose viewership is about two-thirds farmers. The ads were viewable online from 30 August to 26 October, 2022, and randomly circulated between treatment and control versions for the three different conservation practices. Anticipating a small effect size (d < 0.10), in line with similar pro-environmental choice architecture experiments (Osbaldiston and Schott 2012; Czap et al. 2019; Salazar et al. 2021), we arranged to have the ads viewed a total of approximately 300,000 times to promote adequate statistical power. We measured two sequential behavioral outcomes: information seeking via the click-through rate for each ad, and contacting CPs via the online form on Stroud’s website.

Data Analysis

We analyzed the click-through data by performing a one-sided z-test for significant differences between web users receiving the treatment and control ads. We also conducted a power analysis to identify the probability of correctly rejecting the null hypothesis that the photos would have no effect. All analysis was performed in R, 4.2.2 (R Core Team 2022), using the ‘pwr’ package, 1.3–0 (Champely 2020) for the power analysis. The rates of form filling on the Stroud website were not analyzed statistically due to insufficient data.

Results

The ads were viewed a total of 300,594 times, with the treatment and control versions viewed in about equal numbers. The average click-through rate for all ads was 0.1%. We found that showing farmland photos before and after conservation practice implementation resulted in a small, significant increase on viewers’ click-through rate (Cohen’s d = 0.214, p < 0.001, Table 4). The power analysis revealed that these results are adequately likely to be true (1-β > 0.80). The separate tests for individual conservation practices had low power (1-β < 0.80), and therefore are not reported. The form for requesting more information on the Stroud website received three submissions during the period in which the ads were viewable. Two were from residential landowners and one was from a farmer. All respondents requested assistance with installing conservation practices, including buffers and stream crossings, and all reported to Stroud that they had seen the treatment versions of the ads.

Table 4 Effect of conservation practice photos on viewers’ click-through rate
Full size table

Discussion

Together, the results of both studies suggest that the effectiveness of techniques to promote agricultural conservation practices varies by stage in the adoption process. Our interviews with conservation practitioners (CPs) suggested that they use different techniques at different stages, with only a small proportion of all techniques being used at more than one stage. The greatest number of techniques were mentioned at the property visit stage suggesting that this stage is critical to further progress. Indeed, the bulk of the example in 2.3.3 consists of the CP working with the farmer at the property visit and draft implementation stages to find the right fit between the practices that will resolve the resource concerns and what the landowner is interested in and can afford. Our online experiment suggested that the same technique had significant effectiveness on one outcome, but not on the next sequential outcome. Together, these results suggest that, as has been done in other fields (Weinstein et al. 2020), the stage in the adoption process should be considered both when planning which interventions to test and when estimating their effectiveness. Simple correlations between covariates and binary measures of adoption may obscure important variation and details in landowner decisions and CP interventions occurring at different stages in the adoption process.

In the interviews, we found that respondents agreed most about the effectiveness of techniques that clarify the adoption process, such as ‘Summarize process and timeline’ and ‘Set clear expectations for funding and roles’. This agreement suggests that interventions aimed at helping farmers to understand the complexity of conservation practice contracts and rules could be particularly effectives at encouraging their progression through the adoption process. This finding aligns with reviews of other behavioral interventions that show providing instructions and other high-engagement techniques are particularly effective for promoting pro-environmental behaviors that require more effort (Osbaldiston and Schott 2012). Such interventions help to build individual’s decision-making competence by clearly delineating what choices they will need to make throughout the process (Grüne-Yanoff and Hertwig 2016).

Many of the techniques listed by the interviewees align with economic literature on the effectiveness of reducing the transaction costs associated with adopting agricultural conservation practices (McCann and Claassen 2016; Palm-Forster et al. 2017). The frequency with which interviewees mentioned techniques that reduce transaction costs, including summarizing the process and timeline and setting clear expectations for funding and roles, suggests that these transaction costs are a significant barrier to adoption. While the techniques used by CPs can help to reduce some of these costs, they will likely remain unless larger policy changes are made (McCann 2013; DeBoe and Stephenson 2016), given that the transaction costs are rooted in the lengthy application process and burdensome paperwork (Reimer and Prokopy 2014). Techniques such as ‘Present at other organization’s meeting’ and ‘Distinguish conservation organizations’ at the Outreach Event stage reduce the number of meetings farmers must attend and their effort in deciding whom to reach out to about conservation. At the property visit and draft implementation plan stages, techniques such as ‘Explain sources of funding’, ‘Access alternative sources of funding,’ and ‘Bring partners to first visit’, reduce time needed to decide between funding options and the division of responsibilities. Last, during the ‘Implement Plan’ and ‘Post-implementation’ stages, many interviewees listed techniques that reduce the time and effort needed of farmers to install and maintain practices, including ‘Have volunteers do installation’, ‘Help with permit applications,’ ‘Have others do maintenance,’ and ‘Send reminder about follow-up.’

We also found relatively high agreement among respondents regarding the effectiveness of leveraging farmer and CP networks. Respondents mentioned such techniques across the adoption process, including having experts or experienced farmers speak at outreach events, presenting at events organized by other groups, providing peer examples to farmers, coordinating outreach with other organizations, and letting others do maintenance. Leveraging practitioner networks can help to spread effort across different partners, share resources, and identify the best messengers for individual farmers (Galik and Grala 2017). Leveraging farmer networks can help farmers to understand their peers’ past experiences with conservation practices, and enhance trust in CPs given other farmers’ approval of their past work (Bressler et al. 2021; Parks 2022). At nearly every step in the adoption process, techniques that involved leveraging networks were some of the most agreed upon practices among respondents.

Our online experiment represents a way to test hypotheses of intervention effectiveness by adoption stage. We measured outcomes of providing visuals on two sequential, observable behaviors, information seeking and contacting a CP. Most field and laboratory experimental evidence on agricultural conservation adoption only examine outcomes at one stage that may occur much later than the intervention stage. Examples of such studies include mass outreach interventions meant to encourage landowners to submit applications to different programs (Czap et al. 2019; Reddy et al. 2020; Weigel et al. 2021, 2022) and tests of different contract designs at the draft implementation plan stage (Palm-Forster et al. 2017; Lim and Wachenheim 2022).

The online experiment suggests that mass outreach might be made more cost-effective by using online advertisements but further research is needed to understand how to leverage initial interest. We found that adding before-and-after pictures to the ads significantly increased their effectiveness at gaining clicks, but had no effect on whether those who did click contacted a CP. The difference in click rate on the two ad versions suggested that using pictures saved about $738.23 USD compared to the ads without pictures. The ads without pictures would have been required 73,971 additional views (at a cost of $0.00998 USD per view) to achieve the same result. However, the treatment ads cannot be said to have been cost-effective at encouraging viewers to contacts CPs. Rather, an important future direction for research, which has been examined in the health literature (Vandelanotte et al. 2008), is to examine how interventions such as the online ads might be sequenced with other interventions to best maintain progression through the adoption process.

More work is needed on how engagement techniques complement institutional and regulatory mechanisms encouraging farmers’ adoption of agricultural conservation practices (Ranjan et al. 2019). The example in Section “Example of Effective Techniques through Adoption Stages” highlighted the importance of regulatory oversight overcoming that farmer’s hesitancy to work with the CP; it was only after a complaint was lodged against him that he began considering conservation practice adoption. Once the adoption process progressed, the CP’s conversational tone, patience and willingness to collaborate on how best to fit practices to his farm heavily influenced the farmers’ continued progression through the adoption process. As additional engagement techniques are tested, taking account of the institutional and regulatory environments within which they are embedded will be helpful for future implementation research and syntheses that examine how intervention effectiveness varies by context (Yoder et al. 2019).

There are several limitations to both our studies. In Study 1, our small sample of conservation practitioners was not representative of all CPs in the CBW. We likely oversampled CPs in non-profit organizations and did not select our respondents randomly. Because of this, the techniques listed here are likely not exhaustive and we could not investigate how their use varies across different types of CPs. Safety precautions during the COVID-19 pandemic also necessitated that we collect data remotely, limiting our ability to accompany CPs during their interactions with farmers to observe techniques in practices. Additionally, the respondents mentioned very few ineffective techniques, suggesting either a social desirability bias from respondents wanting to appear effective or a lack of awareness that some of what they do with farmers may be ineffective (see Mengak et al. 2019). In Study 2, as with other tests of online advertisements (Kubo et al. 2022), certain limitations apply (see Bruce et al. 2017). In particular, the lack of viewer-level data means that we are unable to know whether a viewer had seen either or both versions of the ad multiple times before clicking on it and could not confirm if viewers were farmers. These limitations might be overcome by designing experiments that measure the number of times respondents are exposed to either the treatment or control.

Conclusion

We report results from two studies that suggest the importance of considering stages in the adoption process when estimating the effectiveness of techniques to promote agricultural conservation practices. First, we used interviews to elicit approaches that conservation practitioners (CPs) have found to be effective at engaging farmers about agricultural conservation practices. We were able to distinguish techniques that are useful to first introducing farmers to conservation practices from those used to draft plans and implement them. Although we did not find high levels of consensus on specific techniques, CPs generally agreed that clearly explaining the steps in the adoption process, reducing transaction costs, and leveraging practitioner and farmer networks were effective. CPs also agreed that mass outreach techniques to spur early interest had low success, suggesting a need for new methods. Second, we used an online advertising experiment to estimate the effectiveness of before and after pictures of conservation practice implementation on two sequential outcomes that occur early in the adoption process. Pictures had a significant effect on information seeking but no significant effect on the second step of filling out a form to arrange contact from a CP. Our results highlight a suite of interventions, by adoption stage, that CPs have used to encourage farmers’ adoption of conservation practices and thus may help guide CP actions. This suite of interventions also provides empirically-grounded hypotheses for further testing of the magnitude of intervention effects on farmers’ progression through the adoption process, and when in that process they are likely to be most effective.