We focus on a region within the U.S. Corn Belt called the Upper Mississippi River (UMR) Foodscape. This area, situated in the UMR Basin, encompasses 83 counties across southeast Minnesota, northeast Iowa, southwest Wisconsin, and northwestern Illinois (Figure 1a; Supplementary Table 1). Of the 13.9 million ha in the UMR Foodscape, 9.1 million ha are cropland, comprising 65% of total land (Figure 1b). According to the 2022 USDA NASS Cropland Data Layer, cropland in the UMR foodscape is dominated by maize and soybeans, which occupy 6.6 million ha, or 72.5% of cropland. Hay and pasture take up another 1.88 million ha, or 20.7% of cropland. Other legumes and grain make up < 1 million ha (6.5% cropland), and fruits and vegetables make up 0.3% of cropland. Tree crops comprise < 1% of cropland.

We defined maize/soybean (MS) acreage in the UMR as hectares of maize and soybean occurring in the last 9 of 15 years, identified using the 2008–2022 USDA Cropland Data Layers (Figure 1a; USDA National Agricultural Statistical Service, 2008; USDA NASS, 2022). Marginally productive MS hectares were defined using the National Commodity Crop Productivity Index (NCCPI 3.0), which ranks land on a scale from 0 (low productivity) to 1 (high productivity; Albers et al., 2022). The average NCCPI score across the UMR Foodscape was 0.65 (Figure 1c).

This study focuses on four continuous cover transition scenarios: winter annual cover crops, winter annual oilseeds, agroforestry, and pastured livestock (Table 1). We define winter annual cover crops as overwintering, non-harvested cover crops on MS hectares. This definition captures a variety of cover cropping practices and species, including cover crops grown as monocultures or in mixtures. We define a winter oilseed as an overwintering oilseed crop, either winter camelina (Camelina sativa L.) or pennycress (Thlaspi arvense L.), relay-cropped on MS hectares. These winter annual oilseeds have been proposed as new winter-hardy cash crops for the U.S. Midwest that can be grown in conjunction with a maize/soybean system (Forever Green Initiative, 2021).

For these two scenarios involving overwintering crops, we estimated environmental outcomes if these practices were adopted on 50% of MS hectares, in line with goals to implement regenerative practices on at least 50% of U.S. row crop hectares (Prokopy et al., 2020; Bossio et al., 2021; World Wildlife Fund, 2023). Because winter annual cover crops have already been adopted on some MS hectares, we estimated environmental outcomes for additional acreage that would equal 50% of MS hectares in total. We defined cover crop adoption on MS hectares as the 3-year average of cover crop use from 2017 to 2019, identified through data from Operational Tillage Information System (OpTIS) version 3.0. In the Foodscape, we estimate that cover crops are currently adopted on 2.1 × 10⁵ ha, or 8% of MS hectares. Cover crops would need to be adopted on an additional 42% of acres to reach 50% of MS hectares in the Foodscape.

We define agroforestry for this study as alley cropping with a woody crop like hazelnut (Corylus americana) on MS hectares, converting at least 20% of MS hectares in a farm to an agroforestry system, based on the U.S. Department of Agriculture Natural Resource Conservation Service (USDA NRCS) agroforestry Conservation Practiced Standard (CPS) 311. We define pastured livestock systems as marginal MS hectares converted to a perennial grass/legume system rotationally grazed by cattle. In this system, we assume cattle are on pasture from calving to finishing at one farm, rather than finished on grain at a feed lot. For both of these scenarios, we estimate adoption on 50% of marginally productive MS hectares, defined as land with an NCCPI score < 0.55. Unlike winter annual crops, these scenarios represent a more significant transformation of MS land, and thus we estimate these practices would be adopted on land in the UMR Foodscape with the lowest opportunity cost.

In this work we focused on three environmental outcomes: nitrate loss, soil erosion, and GHG emissions. When possible, we estimated environmental outcomes with publicly available tools to make the methodology accessible and useful for a variety of applications (Table 2). For the best estimates for the UMR Foodscape, we employed tools that were either developed specifically for the region or that could be spatially targeted to the region. We used the web platform AgEvidence (https://www.agevidence.org/) to estimate environmental outcomes related to cover crop adoption. AgEvidence is a web-based tool developed by The Nature Conservancy that uses meta-analyses and data visualizations to communicate the environmental and food production outcomes that result from implementing a variety of conservation agricultural practices, relative to a conventional row crop system (Atwood and Wood, 2020). The meta-analyses are based on peer-reviewed papers filtered by geography (the U.S. Corn Belt) and cropping system, published from 1980 to 2020 (Atwood et al., 2024).

We used the CarbOn Management and Emissions Tool (COMET)-Planner (http://comet-planner.com/) to estimate changes in GHG emissions due to transitioning MS hectares to winter annual cover crops, agroforestry, and pastured livestock systems. COMET-Planner is a web-based tool that provides estimates of GHG emissions of conservation practices across various agricultural landscapes (Swan et al., 2023). Created by Colorado State University in partnership with the USDA NRCS, the tool is widely used to estimate carbon benefits of conservation practices at large geographical scales (Swan et al., 2023). However, it is worth noting that the model relies on national and regional averages, rather than site-specific data on soils, climate, and management, so the model may over- or under-estimate carbon sequestration for specific parcels (Castle et al., 2025).

When publicly available tools were not available for a specific transition scenario/environmental outcome combination, we used published meta-analyses or empirical studies conducted in the UMR Foodscape region. For certain combinations of transition scenarios and environmental outcomes, we found no empirical studies specifically based in the UMR Foodscape. In those cases, we limited results to studies in states bordering the U.S. Corn Belt. The authors also looked at reputable gray literature (e.g., Ecotone Analytics et al., 2022) to identify other peer-reviewed literature relevant to our systems of interest.

To estimate baseline nitrate loss on MS hectares, we assumed that 18–36% of N applied as fertilizer is leached from the system, based on studies of MS systems in UMR Foodscape states (Randall and Iragavarapu, 1995; Sexton et al., 1996; Masarik et al., 2014). We estimated N fertilizer application rates using the most recent (2016) county fertilizer application data from the Nutrient Use Geographic Information System (NuGIS; Rund et al., 2010). To estimate annual top soil loss, we assumed an average annual top soil loss on MS hectares as 7.2 MT ha¹ yr¹ (Thaler et al., 2022).

We estimated per-hectare coefficients for changes in nitrate loss and soil erosion with AgEvidence for the winter annual cover crop scenario and with empirical studies for the remaining three scenarios; when applicable, citations are provided in Table 2 for empirical studies. For pastured livestock systems, scant literature exists to directly compare row-crop MS systems to pastured livestock because this is currently an uncommon transition in the region. As such, we first estimated the change in nitrate loss when MS hectares were converted to un-grazed, perennial systems. Then we estimated the change in nitrate loss in grazed systems vs. un-grazed systems to account for grazing-specific differences nitrate loss, such as biomass and manure accumulation. We combined these two estimates to calculate total change in nitrate loss and soil erosion.

To estimate baseline GHG emissions across MS hectares in the UMR Foodscape, we used the USDA U.S. Agriculture and Forestry Greenhouse Gas Inventory: 1990–2018 (U.S. Department of Agriculture, 2021). The GHG Inventory includes data on soil carbon stock changes, direct N₂O emissions, and indirect N₂O emissions by cropping system and state. We summed emissions across the UMR Foodscape counties and averaged total emissions by state across the years 2013–2015, the most recent data available. This approach allows us to estimate a percent change in emissions, representing a novel use of COMET-Planner. Typically, the tool is used to estimate a reduction in emissions, measured in MT CO₂e, but our calculation of baseline GHG emissions allows us to convert that figure to a percent reduction.

We used COMET-Planner to estimate changes in GHG emissions when converting MS hectares to continuous living cover (Swan et al., 2023). COMET-Planner aligns field-based studies to USDA NRCS CPS to estimate GHG effects of implementing these conservation practices on farms. For this analysis, we selected the CPS most closely aligned with each transition scenario. When possible, we calculated the range of GHG emissions changes by using COMET-Planner's estimates of maximum and minimum GHG reductions. In some cases, maximum and minimum values were not provided, so we used total GHG emission reductions. For the two CPSs associated with agroforestry, reduction estimates were not available for every county in the UMR Foodscape (values missing for 17 out of 83 counties). If COMET-Planner did not have data for a specific county, we used the average value of all UMR Foodscape counties within the same state. The estimation of GHG emissions through COMET-Planner incorporates multiple sources: soil carbon sequestration or soil organic carbon loss, woody biomass carbon sequestration, and CO₂/CO/N₂O/CH₄ emissions from biomass burning, liming, urea fertilization, and drained organic soils.

To estimate GHG emissions reductions related to winter annual cover crops, we modeled outcomes with two CPS options: “Add Non-Legume Seasonal Cover Crop (with 25% Fertilizer N Reduction) to Non-Irrigated Cropland” (CPS 340) and “Add Legume Seasonal Cover Crop (with 50% Fertilizer N Reduction) to Non-Irrigated Cropland” (CPS 340). For the winter oilseed scenario, the COMET-Planner tool did not have a sufficiently applicable CSP due to the complicated climate change potential of these crops: because their intended use is as aviation fuel (Resurreccion et al., 2021), lifecycle assessments of these crops also include the potential reduction in emissions due to oilseed-derived jet fuel relative to petroleum-derived jet fuel (Berti et al., 2017; Cecchin et al., 2021; Ecotone Analytics et al., 2022). For only this crop, we included life cycle assessments from peer-reviewed studies to estimate GHG emissions.

For the agroforestry scenario, we selected two CPS options to represent a range of management choices: “Conversion of Annual Cropland to a Farm Woodlot” (CPS 612) and “Replace 20% of Annual Cropland with Woody Plants” (CPS 311). For the pastured livestock scenario, we used both COMET-Planner and empirical data. First, we calculated the change in GHG emissions due to conversion of annual row crops to a perennial system (“Conversion of Annual Cropland to Non-Irrigated Grass/Legume Forage/Biomass Crops”; CPS 512). Then, we used per-steer emissions estimates for a pasture-based system to estimate additional GHG emissions from steer, assuming 2.7 steers/ha (Stanley et al., 2018). This method does not consider emissions from manure, so it could represent an under-estimation of GHG emissions from the system.

In the UMR Foodscape, winter annual cover crops were used on 2.1 × 10⁵ ha in 9 of the last 15 years, equal to 7% of total MS area (OpTIS v2023), aligning with previous research (Zhou et al., 2022). To reach 50% of MS hectares total, we calculated environmental outcomes for winter annual cover crop adoption on an additional 43% of MS acres (1.3 × 10⁶ ha). For winter oilseeds, we calculated environmental outcomes given adoption on 50% of acres (1.5 × 10⁶ ha). We estimate no current adoption of winter oilseeds in the UMR foodscape given lack of data, though some estimates suggest there are perhaps 2,000 ha in the region currently (Ecotone Analytics et al., 2022). For the agroforestry and pastured livestock scenarios, we estimated environmental outcomes based on adoption of these practices on 50% of marginally productive MS land in the UMR Foodscape. Counties varied in their quantity of marginal lands (Figure 1c), which could provide an opportunity for county-level targeting of agroforestry or pastured livestock practices. Most land in the UMR foodscape is highly productive; 50% of marginal land equals 9.6 × 10⁴ ha or 3.1% of MS area.

Adoption of winter annual cover crops or winter oilseeds could represent a range of nitrate loss outcomes, but winter annual cover crops are generally expected to reduce nitrate loss to a greater extent than winter oilseeds (Table 3; Figure 2). The use of winter annual cover crops could lead to a range of nitrate loss outcomes on a per hectare basis: from a 74% decrease to an 8% increase relative to the current system (Table 2). If cover crops were adopted on 50% of MS acreage total in the UMR foodscape, that would translate to a 3.6 × 10⁴ MT decrease in nitrate-N leaching to a 3.9 × 10³ MT increase in leaching (Table 3; Figure 2). This wide range echoes findings of a global meta-analysis that found that many factors—including cover crop species, soil type, soil sand content, and tillage intensity—can affect the extent to which cover crops prevent nitrate loss, if at all (Nouri et al., 2022).

While winter annual cover crops are typically not fertilized, fertilizer application is recommended for winter annual oilseeds, which limits their utility in reducing nitrate loss. Fertilizer application for common winter annual oilseeds can range from 67 to 89 kg ha⁻¹ (Ott et al., 2019; Gregg et al., 2024). However, winter annual oilseeds still provide cover and nutrient demand in the fallow season, which can reduce nutrient loss (Berti et al., 2017; Cecchin et al., 2021). As such, our estimates suggest a range of possible nitrate loss outcomes for winter oilseed systems, relative to MS systems: a 25% decrease to a 27% increase in leaching per hectare (Table 2). If winter oilseeds were adopted across 50% of MS acreage in the UMR foodscape, this could translate to a 1.4 × 10⁴ MT-decrease to a 1.5 × 10⁴ MT-increase in nitrate-N loss (Table 3; Figure 2).

Agroforestry and pastured livestock systems represent more transformative agricultural scenarios and have higher potential to reduce nitrate loss. Agroforestry systems could reduce nitrate loss by 82–91% per hectare (Table 2). If implemented across 50% of low-medium productive MS hectares in the UMR foodscape (3.1% of total MS area), nitrate loss could decrease by 1.3–2.9 × 10³ MT (Table 3; Figure 2). On a per-hectare basis, pastured livestock could decrease nitrate loss by 68–86% relative to MS (Table 2). If implemented across the same hectares as the agroforestry scenario, pastured livestock could prevent 1.1–2.8 × 10³ MT nitrate-N loss (Table 3; Figure 2).

Winter annual cover crops and winter annual oilseed crops both offer benefits of preventing topsoil loss. In systems with winter annual cover crops, soil erosion could decrease by 28–92% per hectare relative to MS systems (Table 2). Adoption of winter annual cover crops on an additional 43% of MS hectare could prevent 2.6–8.5 × 10⁶ MT of soil loss across the UMR foodscape (Table 3; Figure 2). Winter annual oilseeds could reduce soil erosion by 39–50% per hectare based on empirical evidence (Table 2). We estimate that adoption of winter oilseeds on 50% of MS hectares could represent 4.2–5.4 × 10⁶ MT soil loss avoided (Table 3; Figure 2).

Agroforestry and pastured livestock systems also have the potential to reduce soil erosion. Available data suggest agroforestry systems could reduce topsoil loss by 30–51% per hectare relative to a MS system (Table 2). If agroforestry were adopted across 50% of low-medium productive MS acreage in the UMR foodscape, soil erosion could decrease by 2.1–3.5 × 10⁵ MT (Table 3; Figure 2). Pastured livestock could reduce soil erosion by 72–88.4% relative to MS systems, which would translate to 4.9–6.0 × 10⁵ MT avoided soil loss if implemented across the same acreage as agroforestry in the UMR foodscape (Table 3; Figure 2).

Winter annual cover crops are likely to reduce GHG emissions, while winter annual oilseeds could lead to a range of emissions outcomes. If implemented on an additional 43% of hectares in the foodscape, winter annual cover crops could reduce GHG emissions by 4.9–7.7 × 10⁵ MT CO₂e yr⁻¹ (Table 3; Figure 2). For winter oilseeds implemented across 50% of MS hectares, GHG emissions could decrease by 1.4 × 10⁶ MT CO₂e yr⁻¹ or could increase by 2.8 × 10⁶ MT CO₂e yr⁻¹ (Table 3; Figure 2). This range represents uncertainty around these new crops, especially over their fertilization rates which will affect emissions, as well as their downstream use—and potential emissions savings—as replacements for petroleum-based aviation fuel.

Agroforestry systems represent a potentially large reduction in GHG emissions, even with limited adoption. This range was created using two different CPSs in COMET-Planner: CPS 612 (convert the whole property to woodlot) and CPS 311 (convert 20% of annual cropland to agroforestry). If agroforestry were adopted across 50% of low-medium productive MS acreage in the UMR foodscape, GHG emissions could potentially be reduced by 3.1 × 10⁵-1.1 × 10⁶ MT CO₂e per year (Table 3; Figure 2).

Livestock systems represent a potential increase in GHG emissions. Emissions from enteric fermentation largely offset expected reductions due to converting MS hectares to a perennial system. If implemented across the same area as the agroforestry scenario, pastured livestock systems could increase GHG emissions by 2.5–5.6 × 10⁵ MT CO₂e per year (Table 3; Figure 2). We estimated the conversion of an annual row crop system to a perennial forage system or un-grazed pasture to reduce GHG emissions (1.3–4.5 × 10⁵ MT CO₂e yr⁻¹ avoided), but grass-fed cattle are estimated to emit 2,700 kg CO₂e yr⁻¹ per steer. At an assumed stocking rate of 2.7 steers/ha (Stanley et al., 2018), we estimate the emissions of livestock to outpace the carbon sequestration benefits.

Our approach, using a combination of empirical studies and publicly available tools, represents an accessible and useful way to scale up spatially based models for landscape-level change at the scale of a foodscape. Our results provide three major outcomes with utility for stakeholders: (1) A broad overview of potential environmental outcomes for an agricultural transitional scenario focused on continuous cover, (2) a preliminary overview of areas with marginally productive maize and soybean production that could be targeted for more transformative transition scenarios, and (3) an example methodology for foodscape-scale projects in other regions to assess environmental outcomes and their potential implications. This work has the potential to set the stage for broader, systems-level interventions (e.g., development of farmer-to-farmer and advisor learning, creation of markets and infrastructure, policy creation and implementation, etc.) to facilitate changes in our agriculture and food systems.

We recognize numerous tools exist to estimate environmental outcomes across various spatial scales, from sub-field, field, small watershed, to large basin (e.g., Perez and Cole, 2020). Many of those tools presently lack the capacity to assess the continuous cover practices and their associated environmental outcomes assessed here, and/or require extensive technical skills and empirical data, often for validation and calibration, to run. The approach presented here overcomes those challenges to provide estimations that can guide future conservation planning efforts and public-private investment. The applications of this work are particularly timely, in order to address the challenge of ensuring that public funds are applied in the right practices and places to achieve desired environmental outcomes (e.g., Jones et al., 2018).

Our approach to analyzing environmental outcomes allows high-level comparisons between transition scenarios, which are useful for conservation organizations considering where to invest their time and resources in a regenerative agriculture transition. Winter annual cover crops and winter oilseeds were compared directly because they are two conservation practices that fit into maize/soybean rotations in a manner that would minimally disrupt the system. The wide range of outcomes related to winter annual cover crops, relative to winter annual oilseeds, could represent the variety of winter cover crop species studied across multiple locations (Atwood and Wood, 2020). This study also does not take into account potential changes in N fertilizer demand for subsequent maize/soybean crops after cover crop use. Winter annual oilseeds, by comparison, have been studied much less extensively (Cecchin et al., 2021; Emmett et al., 2022). Further field studies on winter oilseeds would elucidate the true range of nitrate loss and soil erosion outcomes. So far, life cycle estimates for winter annual oilseed crops assume their use as jet fuel and account for lower emissions from oilseed-derived jet fuel relative to petroleum-derived jet fuel (Resurreccion et al., 2021). At the same time, winter oilseeds have the potential to increase GHG emissions relative to maize/soybean systems, especially due to their fertilizer use (Cecchin et al., 2021).

This analysis focused on environmental outcomes across transition scenarios but did not incorporate economic data. Winter oilseeds have the greatest economic potential of the four transition scenarios we investigated, given our current agricultural system and its associated policy, markets, and infrastructure. Soybean systems relay-cropped with winter camelina or pennycress oilseed crops have been found to generate an equivalent net income to monoculture soybean systems (Ott et al., 2019). Fertilizer application on oilseeds constitutes a major expense (48% of material costs; Ott et al., 2019), so reducing fertilization rate could make the system more profitable and reduce nitrate loss from the system. The system could prove profitable over the longer term given that: (1) winter cash crops only minorly disrupt summer cash crop yields and, thus, profits, (2) winter cash crops do not require capital expenditures on specialized equipment, and (3) these crops could be harvested to generate revenue greater than the additional costs to include them in a rotation. Given the lack of market for these oilseeds currently, future prices for these crops will affect the profitability of the system (Gesch et al., 2014). As of now, winter oilseeds are more successful when relay-cropped with soybean than maize because of better light penetration during establishment in soybean systems (Mohammed et al., 2020); future cultivars could be developed for successfully interseeding in maize systems.

Winter annual cover crops have both costs and benefits associated with them, though they lack the direct profits associated winter annual oilseeds. The largest direct costs of cover crops are associated with planting and management (Bergtold et al., 2019). High seed cost was identified as one of the highest barriers of adoption for farmers [Conservation Technology Information Center (CTIC) and Sustainable Agriculture Research and Education (SARE), 2013]. Cover crops can provide direct benefits, such as higher cash crop yields, though that is highly dependent on environmental factors (Bergtold et al., 2019). Cover crops can provide indirect benefits by adding or recovering nutrients to the soil to reduce N input need, though cost savings are dependent on fertilizer costs (Snapp et al., 2005). Cover crops can also provide indirect benefits through weed control and consequent herbicide savings. Cost-share programs are often cited as a method to incentivize cover crop adoption by reducing short-term risks (Plastina et al., 2020). Payments through the Environmental Quality Incentives Program (EQIP) were found to have a statistically significant, positive effect on cover crop adoption at the county-level in the U.S. Corn Belt, but payments through the Conservation Stewardship Program (CSP) reduced county-level cover crop adoption (Park et al., 2023; Surdoval et al., 2024). These opposing effects underscore that not all cost-share programs are created equal. One proposed reason for the success of EQIP is its narrow focus on new projects likely to be successful in encouraging new conservation practices (Park et al., 2023). However, cost-share programs should not just remove barriers to initial adoption but also help provide a path toward long-scale feasibility (Thompson et al., 2021).

Because agroforestry and pastured livestock systems represent a more significant transformation in the current MS system than either winter annual cropping system, we conservatively estimate their environmental impacts on 50% of marginally productive MS acreage. We estimate that pastured livestock systems and agroforestry systems would offer similar environmental benefits in terms of nitrate loss, but pastured livestock systems could have an order-of-magnitude greater reduction in topsoil loss. These results account for the presence of animals by considering grazed vs. un-grazed systems, but there is uncertainty around these outcomes because of the paucity of studies directly comparing maize/soybean systems to pastured livestock systems. More research is necessary to fully understand this relationship and its impacts on environmental outcomes.

While pastured livestock systems have tremendous potential for C sequestration—more than any other tested agricultural practice on the Mollisol soils of the region (Sanford, 2014)—they are also expected to increase GHG emissions relative to a MS system due to the emissions from steers. Some research indicates that emissions per steer can be offset by soil carbon sequestration (Stanley et al., 2018). In this analysis, we only account for C sequestration due to conversion of row crops to perennial forage; we do not estimate additional C sequestration due to the specific impacts of adding cattle to the landscape. Most grazing-mediated effects on soil C sequestration have been attributed to effects on the relative abundance of C3 vs. C4 grasses in grassland systems (Derner et al., 2006, 2019), which may not be relevant in managed forage system. However, livestock enteric fermentation plus CH₄ and N₂O from manure management account for about a third of agricultural GHG emissions in the U.S. (U.S. Department of Agriculture, 2021). The decrease in pasture-based livestock operations in the Midwest over the last 60 years demonstrates the economic challenges of this system, given current agricultural incentives (Sulc and Tracy, 2007), and the proposed increase in pastured livestock in the UMR Foodscape would outstrip current abattoir capacity. However, with cost-sharing programs, these systems can represent a cost-reduction in rearing livestock due to the reduction in feed costs compared to an operation which must purchase grain feed (Winsten, 2024). Several policy interventions have been proposed to lower the cost of production for small and midsize regenerative grazing operations, such as improving grass-fed labeling standards to better accommodate the needs of small-scale farmers and encouraging supply chain innovations like cost-share for new infrastructure or state-run meat processing (Spratt et al., 2021).

Our analysis suggests agroforestry systems have the greatest potential to reduce GHG emissions compared to MS systems on a per-hectare basis, due to the combined potential of trees to sequester C (Udawatta and Jose, 2011) and the reduction in fertilizer needs for the system (Kim and Isaac, 2022). Agroforestry also reduces nitrate and topsoil loss, suggesting the greatest environmental value of the four systems studied. However, there is currently limited economic viability for agroforestry systems in the region. The most likely agroforestry crop for the region is hybrid hazelnut (C. americana × C. avellana; Braun et al., 2019). Despite growing global demand for hazelnuts, most hazelnuts grown in the U.S. are cultivated in Oregon (Smith and Mehlenbacher, 2023), and the current yield of hybrid hazelnuts in the UMR foodscape region is too low to support commercial production (Fischbach et al., 2011). Efforts are underway to improve hybrid hazelnut germplasm (Braun et al., 2019) for cultivation in this region, with additional financial support from USDA Climate-Smart Commodity grants for agroforestry establishment and production. Hazelnut production requires large upfront expenditures, and trees take at least 4–5 years to bear fruit. Hazelnuts must be harvested and de-husked with specialized equipment that is not yet widespread in the region, though groups in the UMR Foodscape are working to develop a processing line (Braun and Jensen, 2015). The broad environmental benefits of agroforestry systems suggest that there would be regional benefits to implementing cost-share programs to assist producers with start-up costs alongside investments into hazelnut processing infrastructure.

We estimated environmental outcomes with the best available sources, but limitations exist in data availability for this work. For one, the small number of empirical studies and large standard deviations in data underscore the importance of continuing to collect data on these novel agricultural systems, especially through field trials. Also, this work focused on agricultural transition scenarios in isolation, but producers could, and will likely need to if we are to meet ambitious environmental goals, incorporate several transition pathways at once, such as incorporating livestock into an agroforestry system. Evidence suggests that implementing multiple diversification strategies, rather than one strategy, improves cash crop yields, biodiversity, and ecosystem services (Estrada-Carmona et al., 2022; Jones et al., 2023; Rasmussen et al., 2024). Additional studies on the environmental outcomes of stacking practices, as well as effective long-term monitoring, are lacking and will be critical to transformation.

To scale up any of these transition scenarios, the region would need to undergo significant infrastructure changes, and many approaches would come with their own potential environmental externalities. For example, scaling up winter annual cover crop adoption to 50% of all land currently occupied by MS systems would require a proportional increase in seed production, possibly necessitating the conversion of cultivated lands from cash crops or natural systems to produce cover crop seeds (Runck et al., 2020). While maize seed production takes < 0.5% of land devoted to the crop, even a high-seed-yielding cover crop, such as rye, would need an average of 12 times as much land (Runck et al., 2020). Scaling up winter annual oilseed adoption would also require a proportional increase in seed production, though there are currently fewer estimates of seed yield among these plants to predict how much land would be needed. This reality underscores the critical role that cover crop and oilseed breeding research can play in improving seed yield, though pennycress and winter camelina are both currently bred with seed yield as a consideration (Ott et al., 2019).

These realities highlight the importance of a systems-level approach to agricultural transformation. None of the proposed transition scenarios can be implemented without tremendous changes to infrastructure, markets, and agricultural policy, or without increasing farmers' participation in decision-making and adoption of practices. To carry out this transition in the UMR foodscape, an assessment of environmental outcomes is one early step in a multi-sector effort that must combine scientific research into improved cultivars, economic incentives to build out supply chains, and opportunities for farmers to test these strategies without risking their livelihoods. We have focused on reducing local impacts, including local production of greenhouse gases. But widespread adoption of these new systems would have global consequences, which are critical to explore in future work.