Financial incentive programs and farm diversification with cover crops: assessing opportunities and challenges

Farmers in the Great Lakes region of the U.S. face tremendous pressure to reduce nutrient losses from agriculture. Increasing crop rotation diversity with overwintering cover crops can support ecological processes that maintain productivity while improving multiple ecosystem functions, including nutrient retention. We conducted a mixed-methods study to understand how financial incentive programs impact transitions to cover cropping in Michigan. Michigan farms span a wide range of soil types, climate conditions, and cropping systems that create opportunities for cover crop adoption in the state. We tested the relationship between Environmental Quality Incentives Program (EQIP) payments for cover crops and cover crop adoption between 2008–2019, as measured by remote sensing. We coupled this quantitative analysis with interviews with 21 farmers in the Lake Erie watershed to understand farmers’ perspectives on how incentive programs could support greater cover crop adoption. Panel fixed effects regressions showed that EQIP increased winter cover crop presence. Every EQIP dollar for cover crops was associated with a 0.01 hectare increase in winter cover, while each hectare enrolled in an EQIP contract for cover crops was associated with a 0.86–0.93 hectare increase in winter cover. In semi-structured interviews, farmers reported that financial incentives were instrumental to cover crop adoption, but that program outcomes fall short of intended goals due to policy design problems that may limit widespread participation and effectiveness. Thus, strengthening EQIP and related conservation programs could support broader transitions to diversified farming systems that are more sustainable and resilient.


Introduction
Simplified agricultural systems depend on synthetic inputs to sustain yields, causing water pollution, soil degradation, biodiversity loss, and greenhouse gas emissions [1,2].Intensified agriculture also harms farmers who take on record levels of debt while facing narrow profit margins [3], contributing to the decline of rural communities [4].Both research and farmer experience demonstrate that diversified farming systems can help address these crises [5].Farmers can manage crop diversity to support beneficial ecological interactions that maintain crop yield while improving ecosystem functions, which reduces input costs and increases resilience to shocks [6][7][8].
Cover crops (i.e.non-harvested crops) are a diversification practice that supports key functions such as minimizing soil erosion, supplying and recycling soil nutrients, improving water quality, and enhancing soil health [9][10][11].They can also mitigate greenhouse gas emissions by reducing synthetic fertilizer inputs and restoring soil organic carbon [12,13].In temperate climates, overwintering cover crops can maximize these potential benefits by covering the soil for a longer period compared to cover crops that are killed by frost.Replacing winter bare fallows with living plant cover is essential for reducing nutrient losses that cause eutrophication and harmful algal blooms [14].As a result, cover crops are receiving renewed attention and relatively high levels of government funding compared to other conservation practices [15].Yet, to date, only 7.5% of U.S. farms include cover crops on about 1.7% of U.S. cropland [16].We conducted a mixed-methods study in the Great Lakes region to understand whether financial incentive programs for cover crops are increasing cover crop presence on the ground, as well as farmer perceptions of the benefits and challenges of these programs.Our analysis focused on the state of Michigan, which contains hotspots of cover crop adoption in the northern U.S [17,18].
A complex suite of environmental, social, economic, and political factors interacts across scales to influence, and ultimately constrain, farmers' decisions about land management [19][20][21][22].Farmers in the U.S. face major structural barriers to adoption of diversification practices like cover cropping due to federal farm policies (i.e. the Farm Bill), concentrated markets, and corporate power that work in concert to disproportionately support simplified cropping systems [23].Adoption of conservation practices therefore typically occurs through grassroots networks that emphasize peer-to-peer learning [24].A large body of research has focused on behavioral and demographic factors that influence practice adoption, while structural factors, including agri-environmental policies, are understudied [20,21,25].
Over the past 30 years, new agri-environmental policies in the U.S. emerged in response to challenges to state-centered farm support alongside growing interest in environmental conservation [26,27].These voluntary conservation programs include the Conservation Reserve Program (CRP), the Environmental Quality Incentives Program (EQIP), and the Conservation Stewardship Program (CSP).Working lands programs like EQIP and CSP provide cost-sharing for farmers to adopt conservation practices on land in agricultural production.These programs have been durable because they address multiple political goals, yet they are also criticized for limited environmental effectiveness [26,27].For instance, the capacity for these programs to scale up practices that are new or risky, such as cover cropping, is likely constrained by their limited funding.Conservation programs comprised only 7% of projected 2018 Farm Bill spending [3], and funding for commodity and crop insurance programs is more than twofold greater than funding for conservation programs [28].
To assess patterns of cover crop adoption on the ground, a growing number of studies use satellite data products to measure cover crop adoption at large spatio-temporal scales [29,30].Satellite data can also be used to estimate area under 'successful' cover crop performance, measured as high levels of biomass [31][32][33], which is important because realizing the benefits from cover crops depends on their biomass production [10].Studies have used satellite estimates of cover crop area to identify temporal trends and hotspots of adoption in the U.S. [17,29,30,34] and show that corn and soybean yields are reduced, on average, following winter cover crops [35], which is a wellknown barrier to their adoption [9].Only a few studies have used remote sensing to assess relationships between cost-share programs and cover crop adoption, for both state [33] and federal programs [30,36].To our knowledge, only one study has tested the relationship between Farm Bill conservation programs and cover crop adoption (as measured with satellite data) with panel fixed-effects regressions that address issues of omitted variable bias [36].This previous study, however, examined the influence of all conservation payments, rather than payments specifically for cover crops, even though only a small proportion of conservation payments is allocated to cover crops [15,37].
We used panel fixed effects regression models to test the relationship between EQIP payments for cover crops and overwintering cover crop adoption in Michigan from 2008-2019, as measured by remote sensing.Our remote sensing approach distinguished cover cropped fields with high spring greenness (a proxy for cover crop biomass), to identify fields where cover crops performed well in the overwintering niche, which could inform opportunities to adapt this program for increased effectiveness [38].We coupled this analysis with in-depth information collected from semi-structured interviews with 21 farmers in the Lake Erie watershed who are experienced with cover cropping.Our interviews aimed to understand the role that financial incentives play in cover crop adoption, and farmers' perspectives on changes to incentive programs that would increase adoption.To our knowledge, this is the first study integrating quantitative and qualitative methods-including novel remote sensing approaches-to advance an interdisciplinary understanding of the outcomes, opportunities, and limitations of financial incentive programs for supporting farm transitions to cover cropping.We hypothesized that EQIP cover crop assistance would have a moderate, positive effect on winter cover crop presence because a complex suite of social and environmental factors influences and constrains cover crop adoption, and the growth of cover crops after they are planted.

Study area
Our remote sensing analysis encompassed Michigan's Lower Peninsula.Michigan is the second most agriculturally diverse state in the country [39], with farms spanning a wide range of soil types, cropping systems, and climate conditions, although nearly half of Michigan farmland is planted to corn and soybeans [39].Within row crop rotations, the relatively high presence of small grain crops such as winter wheat creates a greater opportunity for cover cropping compared to farms that only produce corn and soybeans, because small grains are harvested in the summer.
EQIP is currently the largest working lands program [27] and during the study period, EQIP contracts covered more farmland in Michigan than other Farm Bill conservation programs.In most years, CRP payments, which are for land set-aside, exceeded EQIP and CSP, although CRP payments declined while EQIP and CSP payments increased from 2008 to 2019 (SI, table S1).Our analysis focused on EQIP payments for cover crops, specifically, which increased incrementally from 2008-2018 and had a large increase in funding in 2019 (figure 1).Data on CSP payments are not readily available at the county level, however, EQIP and CRP comprise over 80% of conservation payments in Michigan (SI, table S1).
To complement the quantitative analysis, we conducted in-depth interviews with farmers who use cover crops in the Lake Erie watershed, where nutrient losses from agriculture contribute to recurring algal blooms in Lake Erie [40].The interviewees were in the Maumee River and the River Raisin sub-watersheds.The River Raisin watershed (2776 km 2 ) contains parts of five counties in southern Michigan and most of the watershed's land is in row crop agriculture [41].The Maumee River is the largest watershed in the Great Lakes region (21 538 km 2 ) spanning Ohio, Michigan, and Indiana [42].Farms in our study were in the northern portion near the Ohio-Michigan border.Roughly two-thirds of the watershed is in agricultural production, and the Maumee River is the primary source of phosphorus to Lake Erie [43].
Farmers in the Lake Erie watershed face tremendous pressure to mitigate the environmental pollution caused by intensified agriculture [44].They also increasingly experience management challenges due to climate change [45].For instance, in spring 2019, extreme rainfall and historic flooding prevented farmers from planting large areas of agricultural land (e.g.Michigan farmers reported not planting 357 620 ha [46]).The Natural Resource Conservation Service (NRCS) created a disaster relief program for farmers to grow cover crops on unplanted land to provide financial support, maintain soil health, and reduce water pollution.Our study site thus experienced a highly visible impact of climate change, and an unusually high level of funding for planting cover crops, during the period when we interviewed farmers.

Data
We compiled the following datasets at the county scale from 2008-2019: area under successful (high biomass) overwintering cover crops, EQIP payments for cover crops (and all other EQIP payments), area associated with EQIP contracts for cover crops, CRP payments, crop subsidy and insurance payments, and two environmental factors that affect cover crop growth (growing degree days (GDD) and precipitation (table 1; SI, appendix S1)) [47].

Cover crop area
To quantify the area under successful overwintering cover crops, we produced annual maps at 30 m resolution of the main classes of agricultural land cover that are present from September to April.To create these maps, we combined information from two data For our statistical analyses, we only considered the overwintering cover crop class, and used this as our dependent variable in all regression models.Specifically, we aggregated these 30 m satellite data to the county scale for each year in two ways (appendix S1).The first quantified the area under overwintering cover crops for all agricultural parcels, and the second measure quantified the overwintering cover crop area only for parcels that were classified as row crops (corn, soybeans, or wheat) for any year from 2008-2019.We used the raster [48] and exactextractr [49] packages in R project software for all data layer creation, calculations, and extraction [50].

Federal payments
County-level data on EQIP payments and contracted cover crop area (figure 1) were provided by the NRCS Resource Economics, Analysis, and Policy Data Team.From 2008 to 2019 in Michigan, about 13% ($21,709,959) of EQIP payments were for cover crops (SI, table S3).We also included conservation payments from CRP [51], and government payments from crop insurance, commodity payments, and subsidies for corn, soybeans, and wheat (provided by the Environmental Working Group [52]) (table 1).

Modeling and analysis
We used Mann Kendall tests to identify temporal trends in overwintering cover crops for each county during the study period.We then used panel fixed effects regressions to test how changes in EQIP payments (or contracted cover crop area) over time affected the area of agricultural land under overwintering cover crops, by estimating equation (1): where cover_crop it indicates the area (ha) under overwintering cover crops in county i at time t, EQIP it-1 denotes EQIP payments (or contracted cover crop area (ha)) in county i in the year of cover crop planting (t−1), Env it denotes the environmental variables in county i during the cover crop growing season t, govt_pay it-1 signifies the payment variables (conservation, subsidy, and insurance) in county i for the cover crop planting year (t−1), φ it are the site (county) and year fixed effects, and ε it is the error (table 2).We ran four separate models-two with overwintering cover crops on all agricultural land and two with overwintering cover crops on land in row crops (i.e.corn, soybean, or wheat in each study year).Models were run using the plm package in R 4.0.5.Software and included the 58 counties with both EQIP cover crop contracts and parcelaggregated winter cover crop data (figure S1).

Farmer interviews
We conducted 21 semi-structured interviews between July 2019 and March 2020 with farmers who plant cover crops in the Maumee River and River Raisin Watersheds (IRB: HUM00165334).We used purposive sampling and an in-depth, qualitative approach to data collection to better understand the perspectives of our sample [53] and provide context for our quantitative results.To identify farmers, we contacted Soil and Water Conservation District personnel who connected us with row crop farmers who use cover crops.We identified additional participants by asking interviewees for other contacts.In the River Raisin we also interviewed farmers who were participating in research on cover crops led by one of the authors.We focused on farmers who grow cover crops because they could provide rich information from their personal experience using the practice.However, interviewees were heterogeneous in their years of experience with cover cropping, the size and physical characteristics of their farms, their attitudes toward conservation, and their level of engagement in government programs (SI, table S4).Sixteen interviewees had participated in EQIP, and all but two had participated in a financial incentive program to support growing cover crops.We did not exclude farmers if they had not participated in EQIP or another program; this broader approach allowed us to learn why those farmers had not participated.
Four interviews were conducted in-person, and the remaining interviews were conducted by phone.Interviews lasted approximately 80 min and included questions on farm characteristics and experience with cover cropping and cost-share policies and programs (SI, appendix S2).One farmer did not consent to being recorded, so we took hand-written notes during this interview.Remaining interviews were recorded and transcribed by one of the authors.Using a standardized template, memos were written for each interview to capture emerging themes.Transcripts were imported into NVivo and thematically coded by one of the authors [54].Our analysis was an iterative process, with interviews, transcriptions, memos, and conceptual analysis occurring simultaneously at times.Interviews were coded multiple times to ensure all relevant data were captured and consistently categorized into appropriate codes to identify salient themes.

Adoption of overwintering cover crops
Remote sensing results confirmed that most agricultural land in the lower peninsula of Michigan is bare in winter (figure 2(a)).The area planted to alfalfa/hay and winter wheat was low throughout the study period, with a slight downward trend for wheat.The overwintering cover and low biomass (weedy fallow) classes were more variable, and the area in overwintering cover crops increased in 2016 and 2017, corresponding with a decline in the area in bare fallow.Overall, the area in successful overwintering cover crops ranged from 239 261-456 242 ha on all cropland in lower Michigan during the study period, with a mean of 325 310 ha, which represents approximately 9% of cropland in lower Michigan (figure 2(b)).
At the county scale, the area in overwintering cover crops increased in just six counties in lower Michigan between 2008-2019 (figure 3), two of which are in the Lake Erie watershed.In contrast, 14 counties in central and northern Michigan exhibited decreases in overwintering cover crop area, with the remaining counties having no change.

Panel fixed effects model results
We ran four models to assess the impacts of EQIP cover crop payments, EQIP contracted area, other conservation payments, and other federal payments on the area in overwintering cover crops (table 2).Both EQIP cover crop payments and EQIP contracted area were significantly associated with overwintering cover crop area for all models.The effect of EQIP area on the area in overwintering cover crops was larger for all crops (0.93) compared to row crops only (0.86), while the effect of EQIP payments was the same for all cropland and land in row crops (0.01).All other variables considered in the models, including GDD, precipitation, and conservation and nonconservation payments were not significant.

Financial incentives support cover crop adoption
Farmers described the financial risk of disrupting cash crop production as an overarching barrier to cover crop adoption.For instance, farmers must successfully terminate overwintering cover crops in the spring in time to plant their primary crop.Fourteen farmers cited challenges with termination, which reduced crop yields by delaying planting or immobilizing nitrogen.More broadly, ten farmers noted it can be hard to justify the cost of the practice with no guaranteed return on investment.'All [farmers] see is the seed cost.They do not see the downstream bene-fits…reduced herbicide, increase in health and organic matter; those benefits are going to come down the line' [Farmer 10].
Farmers uniformly described financial incentive programs as playing an important role in buffering against the risks of cover crop adoption.They explained that, in an ideal scenario, farmers would enroll in a program for several years, gain experience with cover crop management, begin to witness soil health benefits, and thus become motivated to continue the practice once the program ends.One  the area they planted to cover crops.The combination of financial incentives and on-farm experimentation allowed farmers to learn a knowledge-intensive skill and observe benefits firsthand while limiting their financial vulnerability during the transition period.

Incentive program outcomes fall short of intended goals
Despite widespread agreement that financial incentive programs support cover cropping, the interviewed farmers also outlined four fundamental design problems that may limit participation in EQIP and similar programs and cause farmers to discontinue the practice once financial incentives end.First, and perhaps surprisingly given that funding for cover crops is limited, a main criticism was that incentive programs provide too much money per area.This was mentioned by nine farmers (43%), who cautioned that if farmers view these programs as a money-making venture-instead of a soil health investment-they will be less likely to learn how to manage cover crops for maximum environmental benefits.Instead, they may terminate the cover crop too early and diminish its ecological potential, making farmers less likely to continue the practice: 'People are looking at it as a moneymaker.They can put cover out there, do a haphazard job of it, throw it out there.Apply it for 25 dollars an acre, and they are gonna get paid 50, 60, 70 dollars an acre to do it ' [Farmer 4].These farmers suggested that lowering the payment would cause farmers to take the practice more seriously to reap its full benefits.
A second shortcoming noted by eight farmers (38%) in our sample was the burden of enrolling in incentive programs combined with their lack of flexibility.Farmers were dissatisfied with the amount of paperwork and with specific program requirements, especially for federal programs like EQIP, saying that these bureaucratic requirements can be confusing and out of touch with the realities of farming.These farmers noted that EQIP's seeding rate requirements were too high for their region and management systems, or the planting window was too narrow, making management more difficult and limiting potential benefits.Seven farmers expressed uncertainty about the expectations, as deadlines and seeding rates vary by program and year.
A third problem was related to scale.EQIP tends to favor larger-scale farmers who are familiar with government programs.Seven farmers (33%) noted the lack of guaranteed acceptance into a financial incentive program as a problem, as these are often competitive and use a ranking system.With EQIP, for example, applications that enroll more land may receive a higher score and have a greater chance of being funded.This may result in some farmers not even applying: 'I think the policy end-they do not do a very good job of sorting out people like me.You know, they would much rather give an EQIP contract to a 2500 acre farm because they knocked out a bunch of acres, and it does not even really matter if people are really passionate about it' [ Farmer 15].Seven farmers described their preference for programs administered at the local level because these have fewer 'hoops to jump through' and because district employees know their local farming community's needs.However, the smaller budget for these programs made other farmers believe the effort to apply is more work than it is worth.
Finally, four farmers (19%) suggested that program contracts should be longer to achieve the full benefits of cover crops.Cost-share contracts generally last one to three years (Claassen, 2008), but interviewees noted that short contracts may not allow farmers to gain sufficient confidence with cover crop management or experience the soil health benefits and associated financial benefits that come from reduced input costs or increased yields.Because visual indicators like weed suppression, earthworms, and soil aggregation convey to farmers that they are getting a return on their investment, they believed that longer contracts would make it more likely for farmers to witness benefits and continue cover cropping.

Relationship between EQIP and winter cover crops
The results of this study demonstrate that financial incentives support transitions to cover cropping, despite limitations of program design and other barriers that constrain widespread adoption.Our analysis focused on Michigan, which is a diverse agricultural state in a region that sits at the center of debates about how to mitigate nutrient losses from agriculture [45].We used a fixed effects regression approach, which better identifies a causal relationship compared to traditional regressions by assessing whether inter-annual changes in EQIP payments (or contracts) and cover crop area are tightly coupled at the county level after controlling for time invariant factors.We found a strong relationship between EQIP contracted area and cover crop area in lower Michigan, and that every $1 of EQIP cover crop payments led to a 0.01 ha increase in overwintering cover crop presence.Indeed, prior studies have shown a significant effect of EQIP payments on cover crop adoption as reported in farmer surveys [18,55], and these studies find high additionality, meaning that high levels of observed cover cropping are attributed to economic incentives [56].
Our remote sensing approach adds a novel aspect to these findings-by using an NDVI threshold based on ground data, we focused on successful cover crops in the overwintering niche.Without appreciable biomass, environmental benefits such as nutrient retention will not be realized from the taxpayer dollars allocated to cover cropping [31,57,58].In temperate crop rotations, it is difficult to achieve high cover crop biomass and many cover crops fail, which may help explain why we found a stronger correlation between EQIP contracted area and the area in successful winter cover for all cropland compared to land in corn, soybeans, or wheat.The U.S. Census of Agriculture estimates that approximately 13% of farms and 7% of cropland in Michigan use cover crops [16], which is similar to the area we found in the overwinter cover class (9%; figure 2).This is surprising because we focused on winter cover crops with successful spring biomass (a subset of all cover crops), suggesting that we are only capturing a partial estimate of all cover crop area.However, our estimate may be slightly higher than the Census of Agriculture because we only considered lower Michigan, which is where most cover crops are planted, or because our remote sensing algorithm may have overestimated cover crops through time.
Farmers in the Great Lakes experience increasing management challenges from climate change, such as the historic flooding in the year that we conducted farmer interviews.Prior studies have shown that crop rotation diversity, including use of cover crops, is associated with resilience to extreme weather events [8].Overwintering cover crops support climate change resilience by building soil organic matter [12,59,60], which can improve key soil functions including water infiltration and storage [61] that buffer against drought and flooding.By using a threshold approach to identify successful cover crops, our work demonstrates how ecological information can be incorporated into remote sensing studies to better understand the ecosystem services from cropping systems, such as soil carbon storage and nutrient conservation [60,62,63].For instance, models based on satellite imagery of legume cover crops can successfully predict nitrogen inputs from their aboveground biomass [64], which can inform reducing fertilizer inputs.More broadly, there is a lack of research evaluating EQIP and other federal agrienvironmental policies [27], which have been criticized for limited environmental benefits in part because they provide incentives for practices rather than outcomes [26].Remote sensing could be applied as a monitoring tool to distribute more payments to farms with successful cover crop growth, which is a strong proxy for environmental outcomes.

Policy implications
Using qualitative methods, we assessed how farmers interact with and perceive incentive payments [65].The farmers we interviewed had three to more than 30 years of experience with cover cropping.However, they could all be considered innovators and early adopters of cover crops, given that cover crop adoption is still so low [66,67].Although the earliest adopters discussed the long-term positive effects of cover crops, the more recent adopters in our sample explained how incentives supported their transitions to cover cropping by buffering against large financial risks and allowing them to experiment with new management strategies on a portion of their farm.This trial-and-error approach was reflected in the data showing a small median size (41 ha) for EQIP contracts in Michigan (SI, table S3).These farmers also suggested that there is an ideal payment per area that could broaden participation while ensuring that farmers are invested in soil health, which would likely improve cover crop management and outcomes.This suggestion, however, may reflect these farmers' ideological commitments as early adopters (i.e.their motivations beyond subsidies), while nonadopters may have different perspectives.This should be confirmed with future research that includes a broader sample of farmers.Such research may reveal additional opportunities for engaging non-adopters.For instance, previous studies suggest that communication about cover crops within farmer networks can be crafted to resonate with non-adopters [68].Early adopters may also benefit from non-monetary rewards such as other forms of social recognition that strengthen social norms about cover cropping and conservation [69,70].
In Michigan, over $400 million was spent between 2010-2020 on projects to address nutrient pollution, with about 10% of the funds flowing to six southeastern counties in the Lake Erie Basin [71].These targeted investments may explain the Mann Kendall trends showing that two of the counties with consistent increases in overwintering cover crops were in the River Raisin watershed, which faces large pressure to reduce agricultural nutrient losses.Overall, it is perhaps surprising that the six counties where cover crop area increased are in the more intensively farmed region of Michigan, where corn-soybean rotations constrain the use of cover crops.However, prior studies have shown that large row-crop farms are often more likely to participate in federal conservation programs [25,72].This was also noted by some of the smaller-scale cover croppers we interviewed who experienced challenges with accessing these programs and expressed frustrations about ineligibility and program equity [27].The counties that showed declines in overwintering cover crops, in contrast, had much less row crop acreage [73].Nationally, NRCS allocates more than half of EQIP payments to livestock-related practices and facilities [15,37].While this suggests that EQIP funds reach lands that contribute to nutrient pollution, this approach has limitations related to the potential environmental benefits, social equity, and diversity of farm types supported by working lands conservation programs.Our analysis isolated the effects of EQIP payments for cover crops; however, broader political constraints limit EQIP's transformative potential [26,28].
In the U.S., farmer and community-level innovation is the main pathway for transitions to diversified farming systems because policies and markets that promote diversification are limited [23,24].Here, we focused on a federal conservation program that could help alleviate the large structural constraints to cover cropping.While widespread adoption of cover crops is unlikely to occur without broad changes to agricultural policy, modifying the EQIP program could increase participation and better target currently limited funds to improve cover crop success.Drawing from our interviews with farmers who were experienced with EQIP and other financial incentive programs, the following specific changes could increase program impact (SI, table S5): • Create a 'transitional' program for farmers to learn to manage cover crops that focuses on small areas and covers just the cost of implementation.Participation during this transition period would not subsequently disqualify a farmer from enrolling larger areas once they have gained experience and confidence with cover cropping.• Increase flexibility in program requirements, such as planting dates and seeding rates, which could then be tailored to different contexts through a more decentralized approach.Rather than stricter rules about practice implementation, program effectiveness could be enhanced by supporting peer networks to build local knowledge, offering social rewards, and strengthening networks of farmers and researchers to improve site-specific recommendations.• Better coordinate among agencies to streamline the process of enrolling in financial incentive programs, such as reducing paperwork and the number of acronyms.Given that the labor demands for cover crop adoption are high, barriers to enrollment need to be low to facilitate scaling up the practice.• Longer contracts are needed to match the timeframe of biophysical processes, such as building soil organic matter and restoring key ecosystem functions to levels that enable reduced input costs (or higher yields), and thus higher profitability.
Overall, the experiences and perspectives of the farmers we interviewed offer broader lessons for improving conservation incentive programs in regions that have high stakes for mitigating global change.

Conclusion
This study demonstrates that EQIP incentives for cover crops are tightly coupled with cover crop presence on the ground in Michigan, and that incentive programs supported adoption of cover cropping by reducing financial risks.Our results thus suggest that shifting more resources to these programs would increase the area in cover crops, which are currently planted on a small fraction of cropland in the U.S. Our remote sensing approach focused on cover crops that performed well in the overwintering niche because their biomass plays a key role in nutrient retention and other ecosystem functions.Overall, row crop farmers in the Great Lakes region experience large structural barriers to increasing the diversity of their cropping systems.However, this context is quickly changing as the urgent need to address climate change increases the political will and resources available for cover crops as part of a 'climate-smart agriculture' policy agenda.Drawing on their years of experience with cover cropping and incentive programs, the farmers who we interviewed suggested key policy changes that could reduce barriers to enrollment and improve program outcomes.Our findings suggest that strategic, top-down changes to strengthen conservation programs could amplify farmer-led innovation and catalyze transitions to diversified farming systems that are more sustainable and resilient.

Figure 1 .
Figure 1.EQIP payments for all EQIP contracts (dashed line) and for cover crops (solid line) in Michigan.

Figure 2 .
Figure 2. (a) Area in bare fallow and four classes of agricultural land cover from 2008-2019 in lower Michigan, and (b) area in the four winter cover classes, excluding bare fallow.

Figure 3 .
Figure 3. County-scale trends in overwintering cover crop presence on row crop fields in Michigan between 2008-2019, measured with Landsat satellite imagery.Hash marks represent counties not included in the analysis due to missing parcel or EQIP data.

Table 1 .
Variables included in regression models with descriptive statistics, expected relationship to dependent variable (i.e.overwintering cover crop area on all cropland, and on land in row crops), and data sources.All data are county-scale.
Acronyms: CRP = Conservation Reserve Program; EQIP = Environmental Quality Incentives Program; EWG = Environmental Working Group; FSA = Farm Service Agency; GDD = growing degree days; NRCS = Natural Resource Conservation Service; NOAA = National Oceanic and Atmospheric Administration; PSL = physical sciences laboratory.aThisvariable leaves out CSP payments because those data are not readily available at the county level.However, EQIP and CRP comprise over 80% of conservation payments in Michigan (SI, tableS1).

Table 2 .
Results of panel fixed effects regressions assessing the impacts of EQIP payments and EQIP contracted area ('planned cover crop') on adoption of overwintering cover crops in Michigan.
* 2.40 * 3.04 * 2.68 * Note: * p < 0.05; * * p < 0.01; * * * p < 0.001.layersdeveloped by the United States Department of Agriculture (USDA) National Agricultural Statistics Service Information and a map of winter cover that we produced using Landsat satellite data (appendix S1).We specifically mapped five different classes of agricultural land cover: bare/fallow, winter wheat, alfalfa hay, low biomass cover that represented weedy fallow or unsuccessful cover crops, and high biomass cover that represented successful cover crops (e.g.cereal rye, ryegrass), which obtained an appreciable amount of biomass during the overwintering period (referred to as overwintering crop crops in the rest of this paper).