6

Fireside Chat with Industry Representatives

The first session of the workshop’s second day was a “fireside chat” with three representatives from the private sector: Teddy Bekele, the chief technology officer of the farmer-owned cooperative Land O’Lakes, Inc.; Adrian Percy, the chief technology officer of UPL Limited, a major crop protection company, and venture partner at Finistere Ventures; and Karsten Temme, the co-founder and the chief executive officer of Pivot Bio, which produces a microbial nitrogen solution to use in place of synthetic fertilizer. Organizing committee member Ranveer Chandra moderated the session and asked a series of questions about soil health and soil maps, giving each panelist the opportunity to respond. There was no follow-up or discussion beyond those individual answers. This section is organized by the five specific questions posed to the panelists.

Question 1: Why do you care about soil health?

Chandra opened the discussions with a basic question for each panelist: Why do you care about soil health?

Answering first, Bekele stated that Land O’Lakes is a farmer-owned, farm-to-fork cooperative with three divisions: crop inputs, animal nutrition, and dairy products. Crop inputs is the largest division, accounting for nearly 50 percent of the company’s sales. He described Land O’Lakes as a “value-added distributor:” “We buy products from a variety of different agricultural manufacturers, and we sell that to ag retailers that then work with farmers to put these inputs into the ground and then manage the crop throughout the life cycle.”

One service that the company offers to set itself apart from the competition is the provision of recommendations to farmers on ways to best manage their fields and their crops to maximize profit while sustaining the environment. Naturally, soil plays a key role in those recommendations.

To gather data on which to base these recommendations, Land O’Lakes uses “answer plots” to conduct research on agricultural land in all areas where it conducts business, which

represents the majority of the row crop production in the United States, Bekele said. The company plants different seed varieties in different soil types under different soil conditions and different environmental conditions and managed under different farming practices. Land O’Lakes shares the data collected from these answer plots with farmers to help them make the best choices in their own fields. “Making sure that we provide the best possible recommendation,” Bekele said, “really goes back to ‘Do we understand that soil? Do we understand the dynamic properties of it?’” Furthermore, he said, the recommendations must consider not only the farmer’s economic interests and the crop’s productivity, but also the health of the soil.

Temme began his response by saying that he is trained as an engineer, but in the biological sciences. Thus, he has a passion for agriculture and, specifically, for soil health and its dynamic aspects. He stated, “At Pivot we’re trying to help every grower produce a more bountiful crop in a more efficient way,” he said. “We want to be able to move new ideas and innovation into the marketplace fast as possible. And at the end of the day, we’d like to be able to leave things better than we found them.” In short, Temme and Pivot share the goal of making land more productive and sustainable in a way that benefits not only the grower, but also all inhabitants of the planet.

Percy commented that soil has been a focus of many workshop participants and himself for decades, but has only recently come to be seen as “the new frontier.” Industry now realizes the value of helping growers to better understand their soils and then sell them products that can help them maximize their yields while conserving resources and improving the soil quality. “This is something that’s become pretty fundamental to our business,” he said, and a soil mapping system would support the design of some of those products. Soil is a tremendously complex system, and at the moment, a great deal remains unknown about how the products work under various soil and environmental conditions. To understand these mechanisms under dynamic conditions, much more research is needed. “So I’m happy to see these types of conferences taking place because the more input that we have, the more interest and more research going on, it’s going to make us much more effective in making these recommendations to growers that will meet both their needs and also the needs of society moving forward,” he said.

Question 2: What is the Holy Grail for a soil map?

Next, Chandra asked about the characteristics of an ideal soil map. Or, as he put it, what is the Holy Grail for a dynamic soil map?

Percy began by saying that he would only answer this question in general terms because he is not a soil scientist. The ultimate goal, he said, is an enhanced understanding of soils. Soil measurement systems should be standardized, scalable, and affordable so that they can provide comparable data across large geographic areas. This need is fundamental, he said, because researchers today are forced to work with data that may not be comparable from place to place. Also important will be integration of the various types of data (e.g., on the microbiome, on soil structure, on water retention) in order to develop a more holistic understanding of soils and then, against that background, investigate how different individual parameters influence the quality of the soil and the output that arises from it. Finally, all of the information in a soil map should be provided in an understandable format so that growers who choose to use that information can make decisions that will improve their farms.

Bekele agreed that a soil map needs to be actionable from the perspective of the farmer. However, the definition of actionable for farmers has been changing. Seven years ago, when he became deeply involved in this area, “we were just happy that farmers were even using a

soil map to make decisions,” he said. Even though the map was static, farmers were starting to use soil composition information. Now some farmers are not only using soil maps but also collecting soil samples every year, particularly in the fall after the harvest, to understand how the soil composition has changed, and then compare that information to a yield map to decide which crops and varieties to plant and how to manage the fields.

Therefore, Bekele said, it will be important to have maps with more current information. At present, soil samples are generally taken every 3 years or so. But soil constantly changes in terms of not only its composition, but also whether soil erosion or conservation practices have been put in place. He continued that a dynamic soil map should provide current details about soil composition and soil quality. “Have you changed your tillage practices? Have you put some terraces in? Are you putting in buffer strips? Do you understand where erosion is happening and how can you avoid that?” A dynamic understanding of what is happening with the soil at any given time, not just once per year, enhances companies’ ability to offer better recommendations to farmers, as well as farmers’ ability to make adjustments during a cropping season and adopt more sustainable practices overall. “I think the key part for me in what is the Holy Grail,” he said, “is the word ‘dynamic.’”

Temme built on Percy’s and Bekele’s answers by zooming out and talking about bridging length scales and time scales. He explained that the lens through which he views the question is customers, particularly the different needs of different types of customers of dynamic soil information. Growers focus on the actions that they can take to improve their crops’ yield. Thus, they are interested in how their plants respond to environmental conditions and stresses. This response depends on the local soil structure around the plants’ roots, which can change on an hourly or even minute-by-minute basis over the 100 growing days each year for most row crops. In contrast, the time scale for companies such as Pivot, which hope to bring new innovations to the market as quickly as possible, is one growing season because no company can practically bring new products to market more quickly than once per growing season. Other customers—for instance, those interested in carbon sequestration—might be focused on time scales of 1 year or longer and on length scales that are acre by acre. Thus, he concluded, the Holy Grail of soil maps must provide information on a variety of scales that are targeted to different potential users.

Question 3: What sorts of data are brought together to create soil maps?

For this question, Chandra asked the panelists to describe the datasets that their organizations assemble in creating soil maps for their purposes and for their customers. These datasets could capture information such as soil samples, farm management practices, geospatial data, or even genomic data in addition to the typical measurements of various soil characteristics.

Speaking first, Bekele said that Land O’Lakes began with some publicly available data from the U.S. Department of Agriculture, which were not as accurate as hoped but a good starting point. The company also asked farmers to sample soils, especially after harvest, in order to gain insights into the variable rates for things such as fertilizer application and seeding. Although a large component, soil is not the only component of determining variable rate maps. Information on yields can be collected and overlaid on the soil information to better inform variable rate maps. Because soil is constantly changing, Land O’Lakes would like to acquire the data in as close to real time as possible. “That would be the next frontier in my mind,” he said.

The ability to acquire near real time data to inform crop input maps is complicated by several factors, Bekele continued. First, data collection and analysis often do not occur in

timeframes short enough to create dynamic models that can provide insight on current conditions. Field staff are often stymied by nonexistent or insufficient broadband connectivity to quickly exchange information or connect with the Cloud, both for its data storage and its applications. Another challenge, Bekele said, is the availability of data. The relevant data are distributed in many different places, and many of the data are proprietary. He said that what a user does with soil information may be proprietary, but understanding the soil should not be so. He wondered whether a better way exists to collect information that everyone could access to understand at any given point or in every field what the soil composition is and what changes have happened over time. “Today, if you want to have that information, you’ve got to go get it somehow or you got to partner with somebody to go get it,” he said. Delays and difficulties in accessing data are slowing down many innovative applications.

Finally, Bekele said, a third issue is collaboration. He highlighted open-source software as a powerful product of collaboration, but people in the agricultural field tend to eschew that open-source mentality. Instead, they tend to protect their information. “These are the things we have to overcome over time,” he said.

Percy began his response to the question by describing UPL Limited’s approach to developing some of its products. A new biostimulant or a new soil health−enhancing product, for example, begins in the laboratory and then moves to the greenhouse, where it is tested under very controlled conditions to inform understanding of how different nutrient levels or different water regimes may affect the activity of the particular molecule under study. From there the product goes to a research farm, where the company has mapped soil properties in great detail. Next, the product is placed in the hands of growers for open field trials, which Percy described as “an enormous leap.” Ideally, the company would like to be able to predict with reasonable accuracy the outcomes of using a product under various conditions. Once a product is in the field, dynamic soil mapping is critical to the company’s ability to follow the outcomes from season to season to understand their variability and effect on the quality of the soil. “So that dynamic portion is really, really important,” he said.

The company must also consider various regulations regarding, for example, runoff and the dissipation of products in the soil, and soil maps can support this effort as well. From the environmental protection angle, the company must ensure that its products do not harm the soil or the environment in any way.

Percy said that today’s soil maps are useful for his company, but could be improved in various ways. Dynamic soil maps, as described during this workshop, would be tremendously valuable for the agricultural products industry as well as for growers.

Temme explained that Pivot aims to contribute to basic science on the smallest of time scales and the shortest of length scales possible. Providing insight into the inspiration for his company, he explained that during his time as a graduate researcher in the nascent field of synthetic biology, a significant innovation was the move from bulk fluorescent measurements on a population of cells to use of a cytometer to obtain a fluorescent measurement on each individual cell in a culture. This innovation presented a new perspective on how cells behave because researchers could see how different populations emerge. At the same time, the ability to write DNA meant that researchers could write thousands of combinations into DNA to achieve a new order of magnitude of information to test hypotheses. “So I think we want to take a similar approach when it comes to how Pivot understands the dynamic soil properties that influence products we design,” he said.

In particular, he added, Pivot tries to collaborate with as many research institutions as possible. For example, the company is working with teams at Iowa State University on mesocosms to create highly controlled conditions in a real-world environment in order to study interactions among microbes, crops, and the non-living components of soil. He also

mentioned work in which Pivot has collected tens of thousands of root samples from around the world and digitized the microbiome in that root structure to enable a plant-by-plant look at how the plant, the soil microbiome, the soil chemistry, and the soil structure interact to lead to a particular outcome. In short, he said, Pivot devotes a lot of time to pushing the boundaries of science and working with people who collect various types of data, from satellite data all the way down to sensors in the ground, and then figure out how to integrate all of them.

Question 4: How can more collaboration be encouraged in developing soil maps?

Chandra introduced the next question by noting that the data necessary for a dynamic soil map do not exist in one central place. Different organizations have the data, some of them in the public sector and some in the private sector. As a result, it is difficult to know even what data exist, so discoverability is a problem. Companies may not share data for a variety of reasons—to protect privacy regarding sensitive farm management data or to guard intellectual property. So, he asked, how could greater collaboration between the public and private sectors be encouraged, with different entities sharing their data?

Temme began by observing that increasing the incentives for such collaboration is one approach. “Sometimes it feels like there are more disincentives to making data available than there are incentives,” he said, “and that might just be a challenge of communication.” Pivot has benefited when stakeholders within the broader agriculture community have spoken about their difficulties with solving certain problems, which creates an opportunity for the company to find ways to work with these stakeholders. “One of the things that’s really helped any of the collaborations we do or any of the joint research we’ve done with academic partners,” he said, “is to begin by talking about where we have difficulty operating and then finding how we can complement each other because of different resources or skill sets.”

Percy believes that people in this field have an enormous appetite to work together. Because many issues must be solved in a relatively short time, it only makes sense for industry, academia, and government to collaborate in the precompetitive space to generate foundational knowledge for all to share and benefit from, whether the ultimate goal is to design a carbon market or develop a crop input. To do so, however, all players must work with similar standards and similar tools, he said. He is encouraged by some industry consortia that are already working to create such standards.

Agreeing with both Temme’s and Percy’s points, Bekele highlighted the barriers to establishing collaborations. One barrier is intellectual property, which companies are reluctant to share without clear prospects for a return on investment. For farmers, he said, privacy is a very real issue; they hesitate to share their information because “they could easily see where a decade later, all this information that’s been shared could be used against them.” For example, it could come to light that a farmer did not follow certain practices, which caused significant runoff, and was then assessed penalties. “That fear is real whether it’s going to happen or not,” Bekele said, “so we have to be conscious of that.”

For these reasons, the value of collaborating and sharing data needs to be understood by all parties involved, he said. What value do farmers derive from a dynamic soil map that everyone can use? What is the value for companies that collect the data? What is the value for technology companies that do not collect data but can create products that enable data users to make more informed decisions? Another involved party is academics, who use the data to advance science, but may be reluctant to share secondary data or new insights because of intellectual property constraints. All levels of the public sector possess a lot of

data that hold great value for farmers, companies, and academics. If all stakeholders benefit from that data, then they need to be advocates for the funding needed to collect, store, analyze, and share that data. The incentives for collaboration must be clear to all.

In response to Bekele’s point about technology companies, Chandra, who works for Microsoft, said that the computer science research community is working on multi-party compute—that is, performing artificial intelligence on encrypted data. In this scenario, one party could share data with another party that performs aggregate analytics on those data without ever seeing them. This type of tool could help assure growers and stakeholders that they can share their data without sacrificing their privacy or intellectual property. Homomorphic encryption, another new computer science tool, could also help to address this issue.

Question 5: What sorts of new businesses might be enabled by a dynamic soil map?

For his final question, Chandra asked the industry representatives to describe the new business models or scenarios that might be enabled by the data provided by dynamic soil maps. He asked them to assume that such maps have already been created and to project out to 10 years.

The basic question, Temme said, is who is the customer, and what makes that customer willing to part with cash in exchange for some benefit? He envisions three types of customers for a dynamic soil map: growers who want to be more productive, companies such as Pivot that wish to bring innovation to market faster, and stakeholders who care about the underlying value and sustainability of the land. “A more complete and powerful dynamic map is going to be useful to each of those types of potential customers for different reasons,” he said, “and maybe it leads to different types of businesses or scales of businesses that can grow out of solving challenges that we face.” But whatever businesses ultimately take advantage of such a map, he said, “there’s a lot of opportunity there.”

Percy stated that businesses that develop and sell soil health input products would benefit from a dynamic soil map because they could better research and then provide concrete recommendations on input use to ensure the full realization of their efficient use. Another beneficiary would be advisory services to growers. “And by the way,” he added, “10 years is far too long. We need to do this a lot quicker than that.” For instance, many companies consider the carbon market to be one way for farmers to achieve value from sustainability practices on their land. Businesses could also use a dynamic soil map to help farmers improve their precision planting, a current practice that could be improved with more accurate and up-to-the-minute data.

Bekele expanded on a dynamic soil map’s role in carbon sequestration. Although this topic has garnered a lot of interest, he said, many questions remain about the amounts of carbon sequestered and emitted and the factors that influence those numbers. Better estimates of those numbers—which depend on a complete as possible understanding of the soil—would inform pricing of the carbon sequestered in the soil. Farmers would realize a great revenue stream, he said, if they could be compensated for following particular practices that help sequester carbon in their fields.

Today, much of the advice to farmers is product-driven, with product companies trying to convince farmers that they can improve results by using their products. In the future, however, with the valuable information provided by a dynamic soil map, it could be that farmers pay for the insight or the advice rather than getting it from companies whose main interest is in selling products. A switch to an insight-driven subscription model could open up some interesting business avenues.

Finally, Bekele said, a dynamic soil map could help farmers move to different crops. There is significant interest in moving away from corn and soybeans and toward other crops such as peas and lentils, but such a switch requires a significant commitment, in terms of both buying new equipment and learning new practices. With a better understanding of the soil, farmers would be more comfortable with their decisions to make such a switch.

In summarizing the session, Chandra said, “A dynamic soil map helps build better soils. It’s better for the environment. It’s better for the farmers. It’s also better for the world because you can grow much better food if you know the soils better.”

This page intentionally left blank.