Author Information.

Introduction

The U.S. Corn Belt is a ~ 1,000,000 km2 region that stretches from the Dakotas and Kansas to as far east as Ohio (Green et al., 2018). The majority of land is the region is used for agriculture, predominantly corn and soybeans, as well as secondary products like ethanol, eggs, beef, chicken, and pork (Green et al., 2018). While highly efficient and economically important (Riccetto et al., 2020), these annual cropping systems also leave the region’s land bare for six months of the year and therefore prone to erosion (Thaler et al., 2021), nutrient loss (McLellan et al., 2015; Nowatzke et al., 2022), and greenhouse gas emissions (Griffis et al., 2013; Lawrence et al., 2021). The limited number of crops raised leaves farmers prone to economic vulnerability as individual commodity markets fluctuate (Wright, 2011), a fact well-known to farmers (Roesch-McNally et al., 2018). In tandem, not all farmland is equally suitable for production (e.g., marginal) and it has been suggested these areas should be prioritized for perennial conservation (e.g. Figure 1) (Basso et al., 2019; Meng et al., 2022). Diversifying the Corn Belt through implementation of additional crops and perennial conservation practices such as prairie and wetlands has potential for strengthening farm resiliency, reducing inputs, increasing profits and yield, and improving soil health, wildlife habitat, and multiple other agroecosystem services (Bowles et al., 2020; Brandes et al., 2018; Esquivel et al., 2021; Meng et al., 2022; Schulte et al., 2017; Tamburini et al., 2020; Weisberger et al., 2019, 2021).

However, despite the strong evidence for agronomic and environmental benefits of diversification, crop and landscape diversity in the Corn Belt has continued to decrease over the past century as corn and soybean acres have increased (Bigelow et al., 2020; Lark et al., 2020; Morefield et al., 2016; Weisberger et al., 2021). While largely driven by structural farm policy influences on commodity supply and prices (Coppess, 2018, 2017), increasing landscape simplification is also influenced by individual socioeconomic and cultural factors: productivist vs. multi-objective mindsets; the increasing amount of farmland that is rented; differing concerns by landowners about land use; constraints on farm resources; lack of alternative markets; learning curves and potential negative experiences that come with trial and error of new practices; and the present general status quo of corn and soybeans (Drescher & Warriner, 2022; Esquivel et al., 2021; Lark et al., 2022; Larsen et al., 2019; Luther et al., 2020; Prokopy et al., 2019; Ranjan et al., 2019; Roesch-McNally et al., 2018; Wauters et al., 2021). Given the variety and complexity of factors that dictate agricultural decision-making, it is clear that facilitating these decisions requires a robust and evidenced framework.

One way to facilitate diversification decision-making is through agricultural decision support systems (DSS). While there are many definitions, broadly speaking a DSS can be defined as an interactive, computer-based system designed to support decision-making (Liu et al., 2010). Agricultural DSS are often coupled with precision agriculture (PA) or digital agriculture (DA) technologies, and are designed to help the end-user make more informed on-farm decisions to address complex problems. While goals of DSS will vary by commodity, product, and/or type of end user, past applications have included using optimal amounts of fertilizer and pesticides; improved animal husbandry and grazing management; E. coli risk management; optimal grain storage and transportation; carbon sequestration and market decision-making; and general landscape management, to name a few (Eastwood et al., 2017; Lundström & Lindblom, 2018; Mardaneh et al., 2021; Oliver et al., 2017; Rose et al., 2016; Rose & Bruce, 2018; Stitzlein et al., 2020; Zhai et al., 2020).

It is now more common to consider DSS and PA to be a maturing if not mature aspect of agriculture; however, the paradigm of using these approaches to individually optimize specific outcomes may be changing rapidly (Antle et al., 2017; Zhai et al., 2020). Increasingly, DSS are enlisted to assist farmers in making evidence-based diversification and conservation decisions (Ranjan et al., 2020), especially as measuring, monitoring, reporting and verification (MMRV) tools required by private and government sustainability programs such as the Partnerships for Climate-Smart Commodities (USDA, 2022). Now more than ever, technological advances in remote sensing and real-time operations monitoring expansions are being coupled cloud-based information that is increasingly integrated in terms of ecosystem monitoring and accounting (e.g. carbon intensity) (Jones et al., 2017).

However, adoption of DSS by the farming community has remained low, in what is often termed the implementation problem (McFadden et al., 2023; Rossi et al., 2014). A study which interviewed farmers and farm advisers in the UK (Rose et al., 2016) identified 15 factors required for successful DSS design, including benefit to the end user, ease of use, evidence-based, compatible between farmers and advisers, and flexible enough to fit each individual farming operation. These 15 factors also included modifiers to adoption such as user age, farm scale, IT education, and recommendations from peers. A study with Swedish farmers and a new nitrogen fertilizer DSS (Lindblom et al., 2017) recommended that future agricultural DSS designers first ask what information is needed for the given task; when, where, and why is the information needed; and who needs the information?

One common DSS pitfall, and an increasing consensus in the literature, is that many DSS do not explicitly incorporate end-user input or observation insights during the design process (Gutiérrez et al., 2019; Lindblom et al., 2017; Lundström & Lindblom, 2018; Rose et al., 2017; Stitzlein et al., 2020). A review of 61 agricultural DSS (Gutiérrez et al., 2019) found that only 32% of studies had any interaction with end users (e.g., observation, interview, survey, co-design) during the design process, and of all studies only one conducted a formal assessment of the qualitative participant data, indicating that even when end-user interactions occur the information is not used optimally or evaluated thoroughly. As a result there has been a tendency for DSS design to focus on data the developers find ‘cool’ first and user needs second (Findlater et al., 2021; Lundström & Lindblom, 2018). Prior research has shown that involving end-users in the DSS design process improves tool usefulness and usability (Gutiérrez et al., 2019; Lindblom et al., 2017; Lundström & Lindblom, 2018; Rose et al., 2017; Stitzlein et al., 2020), and could therefore help facilitate landscape diversification informed via DSS.

The goal of this research project was to bridge two research areas—conservation farming DSS design and cropland diversification barriers—to: (1) better understand the tasks, tools, and people involved in land conversion from corn and/or soybeans to diversified practices such as perennial grasses, prairie, pasture, alternative crops, and wetlands, (2) identify and understand the nuances of structural barriers and opportunities for diversification, and (3) identify ways these findings could be translated to DSS and elucidate which barriers cannot likely be overcome by DSS. The aim of this study was not to quantitatively survey DSS broadly nor evaluate specific DSS, but to qualitatively characterize how diversification was being implemented and how it can be translated to DSS. To understand the decision-making and implementation process of diversification, semi-structured interviews were conducted with 11 farmers in the US corn belt who had already taken steps to diversify their farming operation. Through this process, 469 relevant pieces of data were identified and then synthesized using an affinity diagram that produced 12 overarching themes relating to how landscape diversification decisions are made and implemented as well as barriers to diversification. Five of these themes were directly applicable to DSS design, while the remaining seven primarily related to social and structural barriers. While participant size was limited, findings from this study can be used to both inform the feature sets of future agricultural DSS, highlights potential policy that could address structural barriers to diversification, and can help inform future research questions with a larger sample (e.g. survey).

Methods

Semi-structured, in-depth interviews were conducted with farmers who had already diversified land on their operation from corn and soybeans to a constrained list of other land management uses (Fig. 1). Semi-structured interviews allow researchers to identify significant factors underpinning research goals and to explore emergent themes as they arise during the interview process (Creswell, 2013). While prior studies have explored conservation and diversification barriers among general farming populations (Prokopy et al., 2019; Weisberger et al., 2021), participants who had already diversified their operation were chosen for this study to better understand the actual process of diversification implementation and how future DSS could support or augment this process. To analyze the data, an affinity diagram (Holtzblatt & Beyer, 2017; Scupin, 1997) was created as a group. While originally developed in anthropology as the KJ method (Kawakita, 1991), affinity diagramming is useful in making sense out of large volumes of qualitative data. In addition to anthropology, the method has been used in many disciplines, including human-computer interaction, sociology, business, and design (Harboe & Huang, 2015). However, based on a literature search, there did not appear to be many direct applications of affinity diagramming in agronomy or agricultural diversification, and this study appears to be a novel application of the method in this domain. Approval for this study was obtained from the Iowa State University Human Subjects Internal Review Board. Participants received monetary compensation for their time.

Research questions

Interviews addressed four research questions: (1) What are the tasks, tools, and people involved in land-management conversion from corn and/or soybeans to diversification practices such as perennial grasses, prairie, pasture, alternative crops, and wetlands? (2) What are the structural barriers and opportunities available for agricultural landscape diversification? (3) How are participants currently using DA/PA software and equipment on their farms? and (4) How can these findings be translated to agricultural DSS?

Participants

Practical Farmers of Iowa (https://practicalfarmers.org/), a farmer-led, non-government organization focused on ‘equipping farmers to build resilient farms and communities’, was partnered with to send out a one-page call for initial participation. Snowball sampling (Parker et al., 2019) was then used to identify additional participants via recommendations from participants and other agricultural professionals. To participate in the study, farmers must have transitioned corn and/or soybean row crop acres to one or more conservation or cropland diversification practices in the past four years. To verify eligibility in the study, participants were screened via a pre-survey and again verbally before starting the interview process. The list of suitable practices included conservation reserve program (CRP), prairie, pasture, wetlands, or alternative crops such as perennials or small grains (Fig. 1). While cover crops were not included as a practice in the initial call for participation, all participants were incorporating cover crops into their operation in some manner and would naturally discuss their experiences with them. Relatedly, it should be noted that the ‘Acres Diversified’ column in Table 1 does not include cover crops.

Eleven farmers working on ten total operations were interviewed. The majority of interviews had a single participant except for two sessions: one session in which both husband and wife were interviewed, and another with a father and son. The husband and wife were treated as a single participant since they co-operated their farm equally, and the father and son participants were treated as separate as they both had distinct farming operations, though they did farm a shared portion with other family members. The average participant age was 44, with the youngest participant being 26 and the oldest 71 (Table 1). Participant average number of years farming was 16, with the shortest tenure of one year and longest of 44 years. The average number of acres farmed was 824, with the fewest acres farmed at 90 and most at 3,200. All participants self-identified as white. All participants self-identified as male except for the participant couple in which the wife identified as female. All participants were from Iowa, USA except for one participant from Illinois, USA. A full list of participant demographics can be found in Table 1.

Fig. 1
figure 1

Graphical depictions of cropland diversification practices explored in the interviews: Conservation Reserve Program (CRP), Grass Waterways (GW), Cover Crops (CC), Perennial Crops (PC), Prairie (PRE), Pasture (PAS), Wetlands (WET), and Small Grains (SG)

Full size image
Table 1 Participant demographic information
Full size table

Interview questions

Five themes were explored in the interviews using a predetermined protocol and script. The first focused on personal and farm demographic information to validate pre-survey results and identify additional pertinent demographic information. The second theme focused on the land-use diversification practices that had been implemented on participant farms within the previous four years. The third theme focused on the steps and tasks involved in the process of land-use diversification for a given practice, such as signing up for a conservation program, seed selection, sowing, management, etc. The fourth focused on the people involved in decision-making and implementation, including government officials, family, and anyone else identified. The last theme explored how current or prior decision-making tools (e.g., spreadsheets, resources, etc.) and DA/PA were being used on participant farming operations, though explicit use of DA/PA was not a requirement to participate in the study.

Procedure

A pre-survey was sent to eligible participants asking for demographic information, land-use diversification practices implemented, and DA, PA, or DSS tools used, if any. The pre-survey reiterated basic information about the study and was the primary means in which participant consent was obtained. Interviews were conducted over several months from August to December 2021. The primary and secondary authors conducted the interviews through online meeting software (Zoom). After consent was again verbally obtained from participants, interviews were conducted and the audio recorded. The interviews lasted around 60–90 min and were guided by the study protocol and script. Following Charmaz’s (2006) operational guide, notes and interview highlights were drafted and shared among the researchers the same day after each interview. The primary and secondary authors adjusted the list of interview questions as new themes arose. It should be noted that the semi-structured interviews tended to move between topics or themes relevant to the research goals, including those not directly asked about in the script. The interview audio was transcribed via an online transcription service (Rev) and transcriptions and audio were moved to a secure university cloud storage. Identifying information was removed from transcripts and respondents were instead assigned individual participant numbers.

Data analysis

The data from interviews was analyzed in two stages. After all interviews were conducted and transcribed, the primary author analyzed each transcript for individual quotes and information related to the research goals. These individual quotes were then transferred to a document which contained the quote text, a unique quote identification number, and the respective quote line start and line end in the transcript. Quotes ranged in length between a single sentence to an entire paragraph. The primary author then distilled each quote into a shorter, more easily readable quote. The distilled quotes were then given their own identification number, which consisted of the participant number and the quote number. These identification numbers allowed each distilled quote to be linked back to the original transcript during the formal analysis, so additional context or clarification about the quote could be easily accessed during the affinity diagramming session.

The affinity diagramming session was conducted with the full author group. Each distilled quote was printed and attached to an individual sticky note. All 469 stick notes were then attached to a blank whiteboard. Following affinity diagramming guidelines (Holtzblatt & Beyer, 2017), researchers then silently placed quotes into similar groupings and themes. Researchers were allowed to move and shuffle quotes that had already been placed. After all quotes had been grouped, the researchers then discussed each group. Some quotes were again moved and groups combined until grouping labels were created. In a separate meeting, final overarching themes were created from the labeled groupings.

Results

Themes identified via affinity diagramming

At the end of the first set of silent groupings, the researchers identified a total of 46 unique groups. Two groups, which consisted of only one observation each and were very specific to the individual participants, were later removed for a remaining 44 groupings. In a separate meeting, the researchers organized the 44 groups into 12 overarching themes (Table 2). The first fives themes are directly applicable to DSS design, which the remaining seven dealt with issues that DSS likely cannot solve.

Theme 1: reasons to use digital agriculture tools/ benefits

Things we like/ work well about software; tech positives/ farmer uses of tech on farm

For the participants who were using PA, DA, and decision-making tools, two themes were identified: the first related to mostly usability and quality of life features and the second related to what participants used the technology for. One participant mentioned enjoying web-based applications and not having them tied to a single computer. In relation, participants enjoyed cloud syncing across equipment, so that any data layer generated would appear in a central location. Participants also liked software that was intuitive and easy to use.

The biggest use of precision data was to track information: general productivity, hybrid seed trials, application trials, field operations, and equipment performance. One participant mentioned using yield monitor data to break individual fields into management zones instead of using generalized field-scale prescriptions (given enough years of observations). Most of the precision data was used at the end of the growing season, with very few instances during the growing season. Decision-making using this data largely boiled down to comparison and selection of management treatments (e.g., hybrids, fertilizer rates, pesticides) to use in the next growing season.

Theme 2: issues with or interacting with digital agriculture tools

Tool Design Flaws; Age/ Time Barriers to Precision Agriculture Use/ Adoption; Software Choice Drivers; Costs/ Value of Precision Software; Tools Not Suited for Conservation Planning.

As with any software, participants identified issues they had with the current suite of DA/PA. This included design flaws, usage barriers, tool feature sets, reasons they were or were not using certain software, and the associated costs and value of using software. Many tool design flaws identified were in relation to poor usability or steep learning curves. “It’s just clunky… If you want to do X, how many steps and screens you have to go through to do that. Or if something goes wrong… Can it tell you what’s wrong? Can it tell you how to fix it?”

Multiple participants identified the youngest person in the operation as the person who generally handled the technology aspects of the operations. One participant indicated that some older farmers were very excited about using PA/DA, but also indicated that it is younger farmers who typically manage that aspect of the operation. Apart from age, the amount of time available was also a barrier, especially among participants who worked more than one job and therefore had less time. Participants also talked about learning curves and difficulties that occurred when software would have major version changes. Several participants also expressed that they felt they were not using the software to its full potential and need to take the time or a training course to learn it. “It’s got tremendous capability, if you know how to run it, but it’s a steep curve to climb to learn how to run it… It’s for a professional… and people that do it all the time.”

Most participants indicated that they were using whichever PA software was associated with the equipment they had, even if they would rather be using a different software; generally, PA software decisions were forced, not elective.

Depending on the size of the operation, the costs and benefits associated with PA varied greatly. Subscription services did not seem overly cost prohibitive for participants with mid to large operations, though cost was somewhat prohibitive for participants with smaller operations. Several participants mentioned that they were in their first or second year of using new precision software and would need to see if the costs provided a suitable return on investment before continuing to subscribe. The biggest costs were for the precision equipment itself, which was very prohibitive for smaller operations. ‘When I first bought GPS it was $15,000 for auto-steer and another $5,000 for the monitor and dome. Then… $1,200 a year for their subscription service. Then if you wanted more things, of course it was more… You can’t justify that on less than 1,000 acres a crop.’ Lastly, one participant indicated that most available precision agriculture software does not consider conservation, profitability, or soil health metrics, and is therefore not well-suited for decision-making in those spaces.

Theme 3: data management and sharing

Farmer data sharing/ privacy; data management

Two observations arose in relation to data sharing and privacy, both in regard to farmer discussions with conservation professionals. One participant, who also worked full time as a conservationist, indicated that farmers were reluctant to share data. Another participant indicated a willingness to share data like yield maps with NRCS, but had never been asked by NRCS about sharing them.

Data management varied depending on the platform and farmer needs. One general problem was the inability of certain data to sync between platforms, even among individual companies’ own software offerings. “Not as easy [to transfer data] as it could be… I do a lot of manual data in-boxing, moving from SMS into Ag or into Climate.” Another participant commented, “[It’s] all flash drive. Nothing is automatic except for FieldView does upload its stuff when we use it.” Overall, only one participant expressed ease-of-use in managing their data, and that was largely because they had the same brand of technology on their equipment which synced.

Theme 4: placement

Rented land; less productive; spatial and farming logistic considerations

Placement refers to where row-crop farmers decide to place conservation practices or diversify their crops. When deciding where to place perennial conservation practices, farmers considered if the land was owned or rented, the productivity of a field or field area, and spatial and farming logistics.

Farmers who mentioned rental considerations seemed more likely to put annual practices (such as cover crops) on rented acres rather than perennial practices such as CRP, unless they were renting the land from a close family member. Generally, participants sought to diversify acres that were less productive, prone to flooding or erosion, lower in soil organic matter, and lower yielding than other field areas (Fig. 2, Items 3 & 8). By diversifying these areas, participants hoped to increase soil organic matter, reduce extra time and inputs spent on these areas, and raise overall average field yields for crop insurance and/or marketing purposes.

Practice placement—especially when it came to CRP placement— depended on spatial and farming logistic considerations. A large part of these considerations entailed being able to access areas for proper management (e.g., multiple mowings in the first year for CRP prairie), which relegated these establishments to field borders or within-field areas that were accessible without damage to annual crops (Fig. 2, Items 8 & 9). “So those are the three things I look at: soil conserving, access, a lower yielding area in the crops that I raise.” It should be noted that two participants indicated a willingness to run over soybeans to access a within-field conservation plating during the establishment year, but neither participant had yet done so. Other spatial considerations included fitting diversification practices to equipment size (Fig. 2, Items 7 & 8) and being more likely to diversify fields close to the homestead to crops like small grains (Fig. 2, Item 3), and less likely to implement alternative crops on fields that were far away.

Fig. 2
figure 2

Conceptual depiction of diversification practices implemented by interviewees in relation to identified themes. Pasture (2) placed close to where cattle are housed (1). Mustard seed (3) planted in an area with low organic matter soil and on few acres as farmer gained familiarity with this new alternative crop. Mustard seed (3) is buffered from neighboring field and associated pesticides (5) by Conservation Reserve Program (CRP) perennial border (4) and is close to pollinator habitat (6) and homestead (1) for more active management. CRP border (4) is easily accessed for management (e.g., mowing) and takes compacted turn rows out of production. Prairie strip (6) and within-field CRP (8) areas are sized to allow for balanced equipment passes (7) on remaining row-crop acres. Within-field CRP (8) removed from production due to frequent flooding; removing allows for more timely field operations and increased ROI; is easily accessed for in-season management (e.g., mowing, burning) via grass waterway (9) and CRP border (4)

Full size image
Table 2 Twelve themes and their respective subgroups identified from affinity diagramming session
Full size table

“It’s a good size farm that’s further away, it’s easier to just keep that in a corn soybean system that we… know how to manage it on a year-to-year basis. We’re not looking at that farm everyday…” Lastly, both conventional and organic farmers indicated a willingness to convert field borders to mitigate pesticide drift from neighbors. For conventional farmers, this would be a corn-specific herbicide that would drift and damage soybeans or vice-versa. The organic grower was required to have a certain width of field border for this purpose. They even increased the width of the buffer to be larger than the minimum width as an extra precaution against pesticide drift.

Theme 5: data-driven decision-making

Financial Planning; Important to Combine Qualitative and Quantitative in Decision-Making; Using Multi-Year Data for Decisions; Yield Monitor Calibration and Data Layering; Fertility Assessment; Data not Adequate.

Data-driven decision-making relates to how farmers use on-farm data and observations to make decisions in relation to farmland productivity, fertility, prescriptions, and financials. An important point is that farmers do not appear to be using one source of data in making decisions, but several, and often using multiple years’ worth of data.

Financial planning was incorporated by farmers in several aspects of diversification decision-making. This included growing cover crops to harvest and offset seeding costs; deciding if a new crop was profitable before implementing it; tracking yearly costs and sales; negotiating custom farming rates; and calculating (or not calculating) break-even yields. “Basically before I started doing any of this, I needed to make sure that it looked like it was financially feasible to raise corn or soybeans. Then the same thing to go to a small grain…” Spreadsheets were the predominant tool used to track financial data and production costs. Land grant university extension documents (e.g., ISU Ag Decision Maker) were frequently mentioned as well, both for calculating costs of crop production as well as negotiating custom farming rates or renting machinery. One farmer indicated that they wished they could see CRP and conservation rates to aid them in decision-making without having to rely on direct communication with the Farm Service Agency (FSA) for such numbers, often requiring an in-person visit to the local FSA office.

Several participants emphasized not just relying on DA technologies but combining DA with in-field observations to make sure what was seen on the computer matched their perceptions. “Go dig in the dirt. Get dirty. Then put their knowledge of the lay of the land, together with the technology. As opposed to just relying on the technology. That’s the big one.” Using multiple years of data in decision-making was a common theme across farmers of all sizes, including examination of good and bad years to sufficiently evaluate field productivity trends. Farmers expressed using anywhere from the previous five to 15 years of data (e.g., yield monitor, satellite imagery) or more frequent intervals such as every few weeks of visual inspection to evaluate land productivity trends. Layering yield monitor data with other data sources was also common for participants who had yield monitor equipment. This included layering yield monitor with soil maps, satellite imagery, and elevation to create profitability estimates, agronomic prescriptions, explanations of subfield yield variation, and validating that what the yield monitor showed was true. “The elevation map is more of, I guess, an unbiased way to see where’s the water sitting, which way does it travel. Whereas yield, so many things can impact yield that it’s nice to double check that with an elevation map.” The majority of farmers were doing some form of fertility assessments to identify field areas with higher organic matter as well as create fertility prescriptions like variable rate nitrogen applications.

Lastly, there were instances where participants indicated that current available data was inaccurate or not adequate for their needs. This included the hit-or-miss nature of publicly available soil maps, PA/DA data not being sufficient for real-time decision-making (i.e., time lags in data acquisition and interpretation), or how one piece of data is often not sufficient to get a holistic picture of problems in the field.

Non-DSS Related Themes

Seven of the identified themes did not directly relate to DSS design, but instead dealt with internal and external influences when it comes to cropland diversification. This included indirect benefits from diversification that were harder to quantify, such as improved timeliness of field operations, enhanced in-field pollination, increased soil fertility and resiliency, reductions in fertilizer use, and enhanced landscape beauty. Farmers displayed different tactics and strategies for diversification, such as which seed mixes to select, management of perennials, starting small, and managing expectations when working in perennial systems versus quick turnaround annual systems. Farmers also identified internal and external reasons not to adopt diversification, such as continuing to cultivate flooded areas, not wanting to put in minimum acreage for certain conservation programs, perceived limitations in alternative crops (e.g., breeding), difficulty in procuring the necessary equipment, lack of markets, and general incentives to maintain the status quo. There were also social pressures to adopt or not adopt, both from within the operation and outside (e.g., the community). Lastly, participants had both positive and negative interactions with government organizations which could impact conservation adoption, and finding adequate and digestible information on conservation practices and management was also variable. More information on these themes (6–12) is presented in Table 2, the Discussion, and in the supplementary.

Discussion

Diversifying the agricultural landscape via alternative crops (e.g., small grains, pasture) and perennial conservation practices (e.g., CRP prairie, wetlands) has great potential for improving on-farm and landscape health. Digital agriculture and DSS can facilitate decision-making on choice of alternative practices and fitting them to the needs of individual farming operations. However, even with a hypothetical perfect DSS, there are still other factors that may influence or limit diversification. This study investigated how diversification is currently being implemented on the landscape via interviews with farmers who had already taken steps to diversify their operation. Study objectives were aimed at adding insight into the use of DSS for diversification and providing resolution over which of the current barriers to diversification could be better addressed with improved DSS versus those that will likely require more broad level interventions. Affinity diagramming was used to analyze interview data and identify ways in which DSS could help facilitate diversification decision-making and which barriers to diversification DSS cannot or only partially address. Overall, findings from this study suggest that there are a number of potential avenues that improved DSS can pursue to support diversification; however the number of structural barriers that were identified outside the perceived envelope for DSS may perhaps be even greater.

How DSS can facilitate diversification decision-making

One common theme among participants was targeting lower-producing areas of fields to be diversified. While participants did not seem to need DSS tools to identify these low-producing areas, support in knowing what to implement, how to manage it based on farming operation and field characteristics, and costs to implement and/or ROI were important in the decision to diversify. Similar to (Zimmerman et al., 2019), participants indicated concern at managing crops when perennial conservation practices like CRP are integrated into fields. Being able to spatially orient desired practices on the landscape is an area where DSS could help, recommending different practices based on field area access for management, soil properties, farmer goals, farming operation characteristics, and ensuring conservation practices like CRP fit program size constraints.

Ensuring that practices made financial sense was emphasized by participants, though the depth of these calculations seemed to vary. Some farmers spent significant time inputting all costs of production to get accurate estimates of profitability for new or existing crops, while others were simply taking rough numbers or ignoring calculations entirely. Nevertheless, DSS should at least incorporate basic financial information for alternative crops such as cost to implement, potential ROI, and custom farming rates and/or additional machinery rental needs. In terms of conservation implementation, being able to see similar costs of implementation as well as potential payments is important, as participants seemed unwilling to implement practices like CRP without long-term payments. Unfortunately, exact payment estimates for practices like CRP are currently only accessible by the FSA, though county average CRP rates are available in the FSA website. Nevertheless, it is important for DSS to incorporate this economic information. In tandem, having local cost share and conservation payment programs geolocated and searchable would be valuable to end users.

Aligning practices with farm goals is another area where DSS can help. Similar to (Weisberger et al., 2021), participants talked about many indirect benefits gained by diversifying: increased pollinator species; decreases in inputs, weeds, pests, and time spent managing unproductive acres; increasing soil fertility, organic matter, and water infiltration; and supplementing livestock bedding and/or feed. Decision-support systems can act to align user goals with these indirect benefits that are likely lesser emphasized in current PA/DA tools. Other ways DSS could help users is by filling knowledge gaps identified by participants. This could include how to establish native prairie/pollinator habitat; managing visual expectations in establishment years; provide pictures to help identify the new plant species; and provide information on proper management and potential issues.

While most farmers were using PA/DA to track yield trends, field operations, and management trials, there are ways in which this information can be transferred to DSS for conservation farming and landscape diversification. Almost all participants were analyzing their fields using multiple years’ worth of yield data and layering other data sources such as soil and fertility maps, satellite imagery, and elevation. One participant commented, “Having that much water stress in low lying ground, it shows up on a satellite really well… I was able to see the areas that, to me, I thought if you looked at the yield monitor data, it shows, all the way across the end rows are all bad. But on the satellite, you could see the crop yellowing in a more dramatic pattern here, and not so much over here. I could say, ‘Okay, well that’s actually good ground.’” As such, conservation farming DSS should allow for layering of data sources and multiple years of data to help identify productivity trends, the root cause of persistent field issues, and allow farmers to check for data source inconsistencies.

Multiple participants emphasized using qualitative, boots-on-the-ground observations, similar to (Rose et al., 2016). Thus, DSS should allow input of qualitative observations with precision data. To further facilitate this data layering, it will be important for DSS to be able to cloud sync data across devices, though this may be a challenge given the mainly forced adoption of software via equipment brand ecosystems. Based on the conversations with participants, it seems that the ability for different equipment to synchronize is limited by application programming interface (API) development and the desire of companies to enable cross-platform sharing, a problem that has been identified in several prior studies (Aubert et al., 2012; Eastwood et al., 2017; Jakku et al., 2019). It seems likely that not allowing differing equipment to communicate with a centralized software will prevent end users from maximizing the software’s potential value. Lastly, DSS need to be affordable for all farming operation sizes. Similar to findings in (Hackfort, 2021), while current PA systems did not appear cost prohibitive for medium to large operations, they were prohibitive to small operations. Similar again to (Hackfort, 2021), participants did not seem to be choosing PA/DA software based on preferences such as available features or usability, but rather were again locked in to whichever software based on their equipment brand. Having affordable, open-access DSS is an area where academic researchers and extension can particularly help by developing products not tied to expensive subscription models or machinery.

It will be important to keep end users involved in the design and development of DSS. Many frustrations with current PA tools were in relation to a high learning curve and/or poor usability. These findings are particularly important given the age barriers to DSS as identified by this study’s participants and other studies (Rose et al., 2016); having DSS that are accessible for an aging audience will be necessary if they are to succeed in adoption and use.

Barriers that DSS likely cannot, or can only partially, address

While areas that DSS could help in diversification decision-making were identified, this study also identified lingering barriers to diversification that DSS cannot or only partially address in the current agricultural and policy system. Participants seemed less likely to put long-term practices on rented land, and instead preferred to implement additional annual practices like cover crops to address resource concerns on rented acres. This would also align with a similar study which found that farmers were less likely to put pasture or alfalfa on rented ground due to high rents and low ROI of alfalfa/pasture compared to corn and soy (Roesch-McNally et al., 2018). Participants also indicated an unwillingness to implement practices like CRP unless there were significant financial motivations, even on acres where crops were repeatedly failing. When asked why they would not transition the failing acres to a perennial conservation program, participants indicated that high commodity prices of corn and soybeans along with crop insurance payments did not necessitate transition of low producing cropland.

As similarly indicated in Weisberger et al. (2021) which surveyed 1k Iowa farmers, lack of available crop insurance seemingly discourages the raising of alternative crops like small grains. Participants indicated additional crop insurance pain points via the subsequent negation of crop insurance on primary crops like soybeans when a small grain was relay cropped (e.g., winter wheat before soybeans). These findings are important given their contradiction to related work. A conservation practice adoption literature review by Prokopy et al. (2019) found a slight positive effect of having crop insurance on implementing conservation practices, though the review only identified six studies which explored the topic. Fleckenstein et al. (2020) meanwhile examined crop insurance’s effect on conservation tillage and cover crop adoption via interviews and a survey. The study did not find that crop insurance inhibited cover crop or conservation tillage adoption; however, it also did not explore its effect on practices such as CRP, extended rotations, or small grains, which this study would indicate crop insurance is a potential deterrent of adopting.

Rye, I raise rye. They can’t insure rye, because it’s not a crop that can grow in Iowa, according to crop insurance… Oats? F*** you, no no no. No oats in [Participant] County. You can grow them in [Second] County. You can grow them in [Third] County. But not [Participant] County… They won’t cover that.

A previous economic study had suggested that reallocation of crop insurance subsidy payment could have a positive effect on CRP enrollment acres by allocating more funding to CRP, including the significant administrative and operating costs associated with crop insurance (Miao et al., 2016). A similar modeling study found that crop insurance did not have a significant impact on conversion of perennials or CRP to cropland (Claassen et al., 2017). Future research should explore these contrasting findings and potential crop insurance synergies for expanding diversification.

Participants indicated that the ability to easily sell corn and soybeans in any given county but not alternatives like small grains also reinforced the status quo. Roesch-McNally (2018) and Weisberger et al. (2021) also cited lack of markets as a barrier to alternative crop adoption. Similar to Weisberger et al. (2021), participants also indicated hesitancy for alternative crops like small grains given the comparatively limited amount of breeding and research that has been conducted on them compared to corn and soybeans. Decision-support systems could partially address alternative crop insurance coverage issues by helping farmers identify which crops are insured in which county. Findings suggest that beyond improving DSS, more research is needed on the effect of crop insurance on both adoption and implementation for a wider set of conservation practices. Further, the current crop insurance structure must be changed to insure small grains and practices like relay cropping which would provide more landscape cover during the year.

Similar to (Rose et al., 2016), participants emphasized that having an adequate social network and being able to see other persons succeed, especially in adoption of alternative crops, were important factors in adoption. While DSS could assist in this effort by attempting to connect farmers looking to adopt a practice with those who have successfully implemented it, there is only so much that can be done in local regions that do not have adequate adoption. Participants also indicated within-operation and external pressures from neighbors and community members, both for and against diversification, and it may be difficult for persons to implement new practices if other persons involved in the operation (especially family members) do not want new practices on the landscape. Participants shared anecdotes about how the general farming populace was either not aware of conservation farming practices or misinformed about them (e.g., not aware that CRP no longer has to be an entire field transition but can be partial areas and smaller acres). Until the broader farming community is aware of conservation farming and how it operates, the movement seems unlikely to go far.

Discussion of participant interactions with government entities revealed both opportunities and barriers to diversification. One barrier identified was poor knowledge of conservation programs, farming practices, and/or general support from FSA/NRCS. This is especially problematic if the county FSA office is ill-equipped for conservation program sign up since they are the obligate gateway to program enrollment. Likewise, one participant commented that NRCS staff can be ill-informed of farming practices, and not understand how to best fit conservation into a farming operation. However, there were also instances where participants stressed the benefits from their positive experiences interacting with these entities, especially NRCS. Morris and Arbuckle (2021) found that visiting a USDA Service Center (centers where FSA, NRCS, and Rural Development agencies are housed together) more than five times was a positive indicator of adoption of five different conservation practices by farmers. While this could very likely be correlation and not causation, it would seem to indicate that having a good relationship with these entities is predictive of conservation behavior. It therefore seems also likely that poor interactions would be associated with decreased adoption by farmers.

Lastly, participants expressed frustration at the amount of time it can take from conservation program application to implementation. Participants also commented that USDA and NRCS documents were generally dense and hard to interpret for the layperson and necessitated FSA or NRCS personnel to interpret; again, problematic if government personnel themselves are ill-equipped to help. This presents an education opportunity for land-grant universities to close conservation and alternative crop knowledge gaps for farmers and agriculture professionals through relevant undergraduate classwork, professional certificates, and extension. Decision-support systems could also assist in this manner by removing unnecessary information and streamline diversification implementation, and open-source DSS could increase the potential for use in educational contexts.

As with any qualitative study, it should be noted that these findings may not represent US or Corn Belt farmers at large, especially given the limitations in the study’s sample size. Instead, this study was meant to provide in-depth analysis into the implementation and decision-making process on cropland diversification, and should be used to inform future studies that seek to quantify these effects (e.g., surveys) in a mixed-methods approach (Creswell & Creswell, 2023; Creswell, 2013). This study was also originally meant to be conducted at farm operation premises when the proposal was first submitted in March 2020. It would be useful for future studies in this research area to conduct on-farm observational visits with more participants to better understand farmer decision-making, operational constraints, and further enhance research findings (Rose et al., 2018; Stitzlein et al., 2020).

Conclusion

Diversifying the US corn belt by adding additional crops and conservation land-use practices has opportunity to increase farm resiliency while reducing environmental degradation. This research study has identified real opportunities for DSS to assist in diversifying the Midwest agricultural landscape. By analyzing interview data with farmers who have already taken steps to diversify their operation from corn and soybeans, several pinch points in current DSS were identified, which if relieved, could increase DSS utility in cropland diversification. These pinch points included: alignment of DSS with users’ conservation and financial goals; ease of use; interoperability between data platforms; DSS and equipment cost; and an accessible user community. Participants tended to use multiple years of data in assessing farmland; already know which areas would be opportune for management change; are driven by monetary benefits of alternative crops and conservation programs to the status quo; and are driven by indirect benefits such as increased pollinators, personal ethics, livestock, and increasing soil health. This study also identified structural and societal issues such as crop insurance structure, lack of markets, interactions with government entities, access to equipment, and general disinterest and misinformation about conservation farming, which could make DSS a null point unless also addressed. In conclusion, results from this study indicate future research and product engineering should engage with farmers and end users before, during, and after the development of any conservation agriculture DSS to ensure that tools are useful and usable.