2
Keynote Presentations
The workshop opened with three keynote presentations focused on different aspects of dynamic soil information systems. Joe Cornelius from Gates Ag One spoke about new and emerging technologies for gathering data on and managing soils. Jerry Hatfield, a retiree from the U.S. Department of Agriculture’s (USDA’s) Agricultural Research Service, discussed soil in agricultural systems, including the functions of soil and ways to improve it. Alison Hoyt from the Max Planck Institute for Biogeochemistry and the Lawrence Berkeley National Laboratory (LBNL) talked about the importance of data archiving and data integration. A question-and-answer session followed the three keynotes.
INNOVATIVE TECHNOLOGIES FOR MANAGING SOILS
Cornelius began his presentation with a quote from a 2017 article in Nature: “Science is informed by what it is possible to measure, and it takes a great leap forward when we can measure something new.” That quote, Cornelius noted, echoes a statement made by Lord Kelvin in 1883: “When you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind.” These quotes are particularly inspirational today, he said, given the many significant opportunities for measurement that are emerging from the convergence among physical, biological, chemical, and mathematical sciences.
In particular, new technologies for measuring and monitoring soil offer unprecedented opportunities to improve agriculture and environmental health. Over agriculture’s 10,000-year history, Cornelius said, soil has been the proverbial black box, or “the hidden half of crop productivity.” The ability to explore that black box will open up new possibilities for agriculture and will help to avoid and even reverse much of the damage to the soil done by traditional agriculture. Sanderman et al. (2017) found that 12,000 years of human land use and agricultural practices have caused a global loss of more than 130 gigatons of soil carbon. At the end of the day, Cornelius said, soil maintenance and soil health are not only good business, but also important paths toward achieving the United Nations’ Sustainable Development Goals that are “fundamental to the survival of our planet and humanity.”
Cornelius provided a general overview of the range of new and emerging tools for soil characterization. These tools vary in scale from laboratory to greenhouse to field to landscape, and their resolution spans from millimeters to kilometers. There are clear trade-offs between resolution and scale, he noted. For example, high-resolution tools can provide high levels of detail and accuracy, but they tend to be labor-intensive, expensive, and not scalable, while low-resolution tools lack the detail and accuracy desired but are generally large-scale, rapid, and inexpensive. Thus, he said, a range of sensing tools is needed to enable the capture of subtle interactions in the soil that determine its chemical and biological properties and the scaling up of the insights gained over large environments.
Cornelius offered the EcoFAB platform as an example of a new tool that has the capacity to perform high-resolution measurements. Developed at LBNL, this platform offers a controlled laboratory environment to study plant–soil–microbe interaction. Plants can be grown in the EcoFAB platform on soil or controlled media, and samples can be extracted for analysis. Automated handling systems that can grow and analyze multiple EcoFABs in parallel provide a high-throughput capacity. Using these systems, Cornelius said, researchers have identified minimal microbial communities that are stable and that can be used to study changes in plant growth and soil micro-environments.
Next, he described a device for direct root imaging using low-field magnetic resonance imaging (MRI). When Cornelius and his team proposed this idea to Ernest Moniz, the Secretary of Energy during the second term of the Obama administration, Cornelius said “[Moniz] thought it was wildly crazy going from a medical MRI device, which weighs tons, to a field device.” However, the team working on this device is already developing version 3.0, which will just be a flat plate on the surface of the soil. The device makes it possible to visualize root–soil interaction, growth, and development, which is important for understanding both the performance of crops in the field and the potential impacts of various crops on soil health (Bagnall et al., 2020).
A root tracker developed by Hi-Fidelity Genetics uses electrical impedance to measure dynamic root growth and depth, and how those differ by genotype across various environments. In particular, Cornelius showed data on two maize varieties whose root development responded very differently under different moisture regimes. “So we’re actually developing soil tools that enable selective breeding for specific micro-environments,” he said.
Continual nitrogen detection, developed by Iowa State University and the University of Nebraska, uses a silicon-based microfabrication technique to produce micro needles that contain nitrate sensor electrodes, which will enable very low cost production at scale, Cornelius said. These sensors are being commercialized by EnGenious Ag. The devices can be used in a variety of environments for measuring nitrates, thus allowing growers to optimize nitrogen application rates or breeders to identify crop varieties with improved nitrogen use efficiency.
Researchers at LBNL and the Noble Research Institute have used electrical resistance tomography to measure soil and plant properties via electrodes inserted into the soil. The system is powered by solar cells and can collect data without human oversight. The researchers developed a new model that is sensitive enough to distinguish among 64 individual wheat lines based on soil moisture levels down to 6 feet, identifying which genotypes had deeper roots and increased water uptake from the soil. “Think about this,” Cornelius said. “We’ve never bred for crops based on root characteristics, such as depth and penetration, which allows us to actually start measuring carbon partitioning in ways never before ever imagined.”
In a different vein, Cornelius described a program called Advanced Plant Technologies that was funded by the Defense Advanced Research Projects Agency. The program’s goal was
to harness plants’ innate mechanisms for sensing and responding to environmental stimuli in order to create living sensors. Some of the resulting technologies have been commercialized by InnerPlant, such as plants that have been genetically engineered to fluoresce when exposed to certain stimuli. Such technologies, Cornelius said, offer “entire new mechanisms for detecting different soil properties and characteristics and correlating those with crop growth and development.” These sensors can be used in a multitude of ways that are not possible with conventional sensors.
At the landscape scale, these sensors can provide site-specific factors such as the landscape position and dynamic soil characteristics that correlate crop yield response to weather. This is the end game that researchers hope to achieve with these particular technologies, Cornelius said, integrating chemistry, biology, engineering, and computational science to improve the monitoring, assessing, and managing of soil, climatic, and genetic resources.
From the agricultural application of new sensing technologies, Cornelius then turned to climate stewardship and, in particular, soil carbon monitoring. “Measuring soil carbon has been a real challenge,” he said, “and it’s absolutely imperative that we create new scalable tools” for that purpose; he then described examples of three such tools. The first, Yard Stick, is a handheld rapid visual sensor that can assess soil carbon and soil bulk density using spectral analysis, resistance sensors, machine learning, and statistics to measure and calculate the amount of carbon in an area of soil.
Second, LBNL is working on a noninvasive, neutron-based system that will provide real-time soil carbon concentrations. The ultimate goal, he said, is to be able to obtain spatial–temporal observations of soil carbon that can inform management systems “so [that] rather than waiting for a decade before we can measure differences in management practices … [we are] able to do that on a time scale that is measured in years or less.”
Third, Impossible Sensing, a firm in St. Louis, is developing a new sensor that can resolve different forms of carbon in soil. One of the company’s sensors is on the Mars Rover, Cornelius noted. “How ironic that we’re actually taking a lot of these technologies to other planets,” he said, “yet we’re not fully deploying them here in the United States.”
Because these various types of high-resolution sensing technologies are not amenable to landscape-level assessment, it is essential to develop models that can establish correlations between the high-resolution data from these sensors and landscape-level insights from remote sensing. In particular, data from the different technologies are used to parameterize and validate models that can be scaled up or applied to different geographies and ecosystems. Several biogeochemical models for soils exist, Cornelius said; in particular, the Microbial Efficiency-Matrix Stabilization (MEMS) model simulates the transport and conversion of organic matter in the soil based on microbial and chemical processes. MEMS is building on existing biogeochemical models to better account for the impacts of plant growth above and below ground.
Ideally, he continued, models will be combined with remote sensing data to provide insight on soils across large acreages or even entire countries. Researchers at the University of Illinois are using satellite imagery from across the Midwest in combination with data from existing technologies to calculate soil organic carbon (SOC) concentrations. Knowing SOC values makes it possible to calculate carbon stocks in the soil, and SOC maps can also point to areas of uncertainty where focused sampling could be used to provide more accurate data. In the future, Cornelius said, these models will continue to improve as ground-based sensors provide additional data to inform them.
Another example of combining remote sensing data with ground-based data technologies is Innovative Solutions for Decision Agriculture (iSDA), funded by the Bill & Melinda Gates Foundation. iSDA’s goal is to combine low-cost, field-level agronomic diagnostics
with advanced geospatial analytics to provide agronomic advice for small-scale farmers in Africa at a reasonable cost. iSDA offers a digital soil map of Africa at 30-meter resolution based on soil samples from more than 100,000 locations, high-resolution satellite data, and cutting-edge machine learning approaches. It predicts soil properties, such as soil fertility, for more than 24 million individual locations, and it is open access.
Concluding his talk, Cornelius referred to his opening comments about scale and resolution, adding that the capacity to make measurements at scale makes it possible for research to inform policy, assess impacts, and influence market adoption (see Figure 2-1). “At the end of the day,” he said, “the sensor technologies and the research that many of you are doing inform these models, these policies, and actually translate into a significant impact and economic development.”
Successful innovation requires three ingredients, he continued: compelling technology, receptive markets, and enabling policies. The current climate-smart soil technology landscape has all three. “With the convergence of technologies and the hard work of the scientists and the funding coming from many of these public programs, we’re actually seeing a profusion of innovation,” he said. “This is truly a robust innovation landscape, and it needs to continue to grow and flourish.”
SOIL IN AGRICULTURAL SYSTEMS
Introducing his talk on soil in agricultural systems, Jerry Hatfield said that the basic concern is how to work with soils to enhance ecosystems and, ultimately, human existence. Learning how to do this requires an understanding of soil, its current state, and what options are available to move to a more desirable state.
Soil has many vital functions, Hatfield said. In addition to the provision of food, fiber, and fuel, it has roles in carbon sequestration, climate regulation, water purification, flood regulation, habitat for organisms, and many others (see Figure 2-2). However, regarding agriculture, soil has five basic functions: (1) soil serves as a water reservoir, and, indeed, much of the variation in crop performance from one place to another is due to how well the soil supplies water to the plants; (2) soil provides the nutrients that the plants need to
SOURCE: Cornelius, slide 19.
SOURCE: Food and Agricultural Organization of the United Nations; see http://www.fao.org/resources/infographics/infographics-details/en/c/284478 (accessed June 1, 2021).
grow and produce products such as grain, foliage, or fruit; (3) soil provides physical support for plants; (4) soil cycles carbon; and (5) soil helps decompose pesticides and antibiotics.
Given the importance of soil, the current state of soil around the United States and the world is concerning. Showing a graph of SOC in farmland in Illinois and Missouri from 1885 to the present, Hatfield commented that humans have had an impact on soils since they started to cultivate agricultural fields, but the effect has been quite dramatic over the past century. According to data and some estimated figures, fields in Illinois and Missouri with a corn–oats–hay rotation lost 35 percent of their SOC over a century’s time. Soils planted continuously with corn in Illinois and Missouri lost almost 60 percent and 70 percent, respectively, of SOC in less than a century.
Generally speaking, said Hatfield, agricultural systems have changed the country’s soils in several ways. First, “we’ve removed organic matter through tillage,” he said, and “as we till the soil, we oxidize a lot of that organic matter.” Second, various agricultural cropping practices have limited the return of carbon to the soil. The monocultures and limited
rotation systems that have become common in crop production only put carbon into the soil at particular times during the year, which has reduced the functionality of the soils and increased agriculture’s reliance on external inputs. “We have soils that have very limited infiltration rates; we have soils that have limited nutrient capacity,” Hatfield said, so nutrients and water have to be supplied to those fields. Finally, agriculture has increased erosion rates and soil degradation. Cultivation has eroded higher organic matter soils into the slope. Many slopes have lost their A horizons, and the B horizon is now what is being farmed.
The primary factor affecting agricultural systems is water, Hatfield said, and soil plays a major role in determining how water behaves in a field. Depending on the soil’s infiltration rate, water can be absorbed by the soil, stand on top of the soil, or run off the land to some other location. If it stands on top of the soil, the water drowns the crop. If it runs off the soil, the water is lost in terms of its use for agriculture and the topsoil goes downstream with it. Limited infiltration rates are due to the changes that agriculture has made to soil functionality.
How one views soil and agriculture is scale dependent, Hatfield noted. One can think in terms of soil profiles or individual horizons, one can think about what is happening at field scale, or one can move to the landscape scale. “A lot of our sampling, our interpretation, and our management is really scale-dependent,” he said, with one’s observations and decisions depending on the scale at which one views the system. He added that information needs to be taken from a soil profile and then translated into the variation that is present in a field. That translation at the field scale can then be used to understand activity at the landscape scale and subsequently determine the best way to enhance the functionality of the soil at all three scales.
One thing to keep in mind when thinking about soil dynamics, Hatfield said, is that soils are diverse, with their properties depending on their parent materials and on the particular history that led from those parent materials to the soil. Some soils are rocky, some are clay-like, some have high levels of organic material, and so on.
Another factor to consider is the interdependence among soil functions. It is not possible, for instance, to separate soil-water from nutrient availability or support for plants from soil-water availability. None of these areas are a silo, he said. “They are really interdependent when we start talking about soil functionality.”
As an illustration of this interdependence, Hatfield showed two figures that mapped average county yield in soybeans and corn against the average National Commodity Crop Productivity Index (NCCPI), which is a measure of various soils’ productivity (see Figure 2-3). In Kentucky and Iowa, the average yield was clearly linked to average soil productivity—the better the soil, the higher the yield. However, data from Nebraska told a different story, with high average yields for soils of every NCCPI. The reason for this was that the data from Nebraska came only from counties that used irrigation. “So if you can manage the water, the quality of the soil is not important,” Hatfield said, “but if you’re rain-fed agriculture, the quality of the soil and how it’s influencing this becomes very important.”
The variability in soil productivity raises the question of how to improve the quality and functionality of soils, he said. One key fact about soils is that different textures of soil have different water-holding capacities. Sand holds relatively little water, and there is little difference between the wilting point of sand—the minimum amount of moisture in soil required for a plant not to wilt—and the field capacity of sand. Silt loam, by contrast, holds a great deal of water and its field capacity is much greater than its wilting point, making it much easier to keep plants healthy. Furthermore, adding organic matter improves water storage capacity in all different soils. Thus, carbon is a major factor in the dynamics underlying soil quality.
SOURCES: Egli and Hatfield, 2014a,b.
Much of the varying productivity across fields, Hatfield said, is due to differences in water-holding capacity. Hatfield suggested the need for more research into how soils with high versus low water-holding capacity are created in terms of their parent material, and how that capacity relates to the functionality of supplying nutrients and water to crops. The knowledge gained from this research would inform management practices, including the changes needed to improve soil water-holding capacity and functionality. Hatfield said that the central questions are “what can we do to change soil-water availability?” and “what information is needed to evaluate the effect of these changes?”
Agricultural systems involve a carbon cycle, a water cycle, a nitrogen cycle, and various other nutrient cycles, as well as several different processes. Therefore, to better understand how to change these systems, it is necessary to take into account the overlap and inter-connectivity of the different cycles. This requires, according to Hatfield, thought about the types of research that help to clarify the dynamics and interdependencies of agricultural systems.
To illustrate, Hatfield presented a figure based on information in a study by Wiesmeier et al. (2019) that ranked the relative importance of the drivers of SOC storage (see Figure 2-4). The largest effects on SOC storage came from microorganisms and fauna, followed by texture–clay content, land use and management, vegetation, climate, and several others. Because of the interdependencies among the various systems, Hatfield said, separating the effects of one driver from the rest is not possible, but “we need to understand what may be causing the bigger change in our overall system and how do we measure that as part of that overall component.”
In northern Iowa, Hatfield’s group has followed the content of organic matter in three fields since 1984. At first the percentage of organic matter in the soil was quite low—an average of 2.7 percent for the three fields. However, the fields were switched to no-till/strip-till, and the percentage of organic matter steadily increased to an average of 5.3 percent in 2015, an increase of 2.6 percent organic matter in the soil. The major change has been reduced tillage, although cover crops have been added in recent years. The major message is that soils can be changed by altering management practices. To evaluate changes in soil, Hatfield said, researchers should understand the interactions among processes within the
NOTE: SOC = soil organic compound.
SOURCES: Hatfield, slide 22, adapted from Wiesmeier et al., 2019.
soil volume, the history of soil management, and the information they must gather about soil response to management.
To close, he offered several thoughts about the necessity of soil information. First, efficient production requires understanding of the functionality of soils. Second, the functionality of soils is linked with climate and management in order to produce crops and livestock. Finally, the overarching challenge is to identify ways to leverage information at different scales to simultaneously increase functionality and production efficiency.
THE IMPORTANCE OF DATA ARCHIVING AND DATA INTEGRATION
In the final keynote address, Alison Hoyt spoke about the importance of data archiving and data integration, including the programs that conduct such activities, with a focus on the role of soils in the global carbon cycle and implications for climate change.
A main question of interest to Hoyt centers on how land management and climate change affect soils. In particular, her research explores the effects on vulnerable soil carbon due to management practices and climate change, and whether that carbon is released into the atmosphere as a greenhouse gas. In addition, various feedback loops must be considered because soils are not only affected by management and climate change, but also could, in turn, influence climate change via, for example, carbon capture. “In all of these cases we’re really interested in understanding these changes and seeing how quickly they might occur,” she said.
To understand soil carbon, Hoyt said, three basic questions must be answered: What are the current stocks of soil carbon? How might they change in the future? And in what time scales can these changes be expected to take place? Soil data can be leveraged to answer these questions. Databases are currently being applied to address these questions in two major ways: first, through continental-scale sampling efforts and data organization efforts, and second, through grassroots efforts to organize data around central scientific questions and the understanding of soil processes. These efforts are essential, she said, and complement each other nicely.
Hoyt discussed continental-scale sampling and data organization efforts, which are typically top-down and systematically structured, and provide an excellent snapshot of the current state of soils. Examples of such efforts include the following:
- USDA’s Rapid Carbon Assessment, which performs systematic sampling of soils at high resolutions across the United States and provides information on soil carbon stocks.
- The National Ecological Observatory Network, operated by Battelle Memorial Institute, which performs systematic sampling at sites that represent different ecological regions.
- The European Union Soil Observatory platform, which provides soil knowledge and data flows to safeguard soils and works in combination with the European Soil Data Centre, a center for datasets of maps, documents, and ongoing projects and events related to soil.
- The African Soil Information System, which has around 20,000 samples that were collected systematically and then analyzed, enabling much better mapping of soil carbon as well as soil properties across the African continent.
These continental-scale efforts have major benefits, Hoyt said. They enable systematic sampling and organization of data that are made available to a wide range of players. They also use careful sampling design and consistent standards. In the case of analytical efforts, measurements are often made by the same laboratory or with shared standards, and in terms of sampling design, sampling schemes are repeated consistently, which simplifies data collection and analysis and enables cross comparisons. However, she continued, this approach provides information from a limited time period because many of these programs began operations relatively recently, and many questions—such as how soils change in response to management and climate change—require more depth in time to answer.
Fortunately, some of today’s grassroots efforts to organize data around central scientific questions can provide data that stretch far back into the past and add another dimension to the data in the continental-scale databases. A dimension is added because these databases include data from not only ongoing sampling efforts, but also tap into the published literature, archived data, and past samples, with the overall objective being to use the assembled knowledge to understand mechanisms and processes that can help predict soil’s response to various pressures and forces.
Because small grassroots efforts lack the resources to make new systematic continental-scale measurements, they have become very resourceful in repurposing past datasets and realizing the most from past investments, Hoyt said. Historically, most grant awards have been relatively small, and most studies have focused on one question at one place and one time. The creators of grassroots datasets repurpose the data from these small studies—combining them to answer centralized questions different from the studies’ original research questions.
One implication of this approach, Hoyt said, is that datasets must be archived so that they can be repurposed in the future. Unfortunately, she continued, data are being lost at a precipitous rate. In one study, for example, researchers learned that the data of researchers from previous studies were inaccessible for a myriad of reasons, including defunct email addresses and lost hard drives. This inability to retrieve previously collected data highlights not only the importance of researchers archiving their datasets, but also of the many grassroots efforts that are emerging to compile these different datasets to answer core scientific questions.
Hoyt offered the International Soil Radiocarbon Database (ISRaD) as a case study of some of the common challenges that these collective databases face and some of the outcomes that they have achieved. ISRaD, she said, is a large collaborative effort to compile soil radiocarbon and related data with the goal of better understanding rates of soil carbon cycling and mechanisms of stabilization in soils.
The different grassroots databases have come into existence in different ways, Hoyt said. ISRaD was a product of a dedicated community along with some support from the Powell Center of the U.S. Geological Survey to organize workshops to bring people together. ISRaD is focused on radiocarbon in soils because it strongly constrains the global rates of soil carbon cycling, which in turn allows researchers to address some of their key questions, such as how climate change might affect carbon stocks and on what time scales.
Radiocarbon naturally occurs in the atmosphere at very low levels, Hoyt explained, but the testing of thermonuclear weapons in the 1960s caused a sharp increase in global atmospheric concentrations of radiocarbon. This event effectively led to a “global tracer experiment” through which researchers can examine how that radiocarbon from bomb testing was incorporated into vegetation and then soils and then respired back to the atmosphere.
“We can use this [radiocarbon isotope] labeling, in effect, to see how long it takes carbon to cycle through soils,” Hoyt explained. By quantifying levels of radiocarbon in vegetation and soils, it is possible to determine how quickly carbon cycles through soils, which can in turn offer insights into how fast soils might respond to climate change in the future. Recent research has shown, for example, that carbon cycles through soil with a mean transit time of a few years in the tropics to decades in middle latitudes and even centuries at higher latitudes (see Figure 2-5).
Although ISRaD is just one database focused on radiocarbon, Hoyt said, researchers have used it to answer many different questions, including the age of soil carbon, how carbon is distributed in soil fractions, and whether earth system models can be benchmarked with radiocarbon. Although ISRaD focuses on a specific issue—using radiocarbon in soils to understand time scales and rates of cycling—it shares a feature with many other grassroots databases, that is, it pulls and uses data from multiple published studies intended to address other scientific questions to address a new scientific question.
SOURCE: Shi et al., 2020.
Other success stories beyond ISRaD, Hoyt said, include databases such as Soils Data Harmonization, Soil Organic Carbon Data Rescue and Harmonization Repository, and Continuous Soil Respiration (known commonly by the acronyms SoDaH, SOC-DRaHR, and COSORE, respectively). She pointed in particular to the Soil Respiration Database (SRDB), launched in 2010. SRDB is a global database of soil respiration data that generates hundreds of direct citations and thousands of indirect citations per year in research fields as diverse as agriculture, biodiversity conservation, ecology, and forestry in addition to soil science.
Hoyt asked whether the field should rely on grassroots efforts to collate understandings about mechanisms and past knowledge, noting, in particular, the failure rate for such efforts. “For every database that’s been successful, there are probably many more that have failed or not made it to fruition,” she said. Small databases and grassroots efforts face many challenges, including heavy reliance on individuals, so if funding priorities or jobs change, a database can be derailed overnight; lack of standardization, so that researchers are constantly reinventing the wheel; many inefficiencies related to unique focus of each database; and various logistical hurdles, such as how to build the database because most researchers in the space lack the requisite database experience.
Another limitation of these grassroots efforts, according to Hoyt, is their tendency to incorporate the biases inherent in the published literature on which they rely. For example, a large percentage of site-level studies are concentrated in North America, Europe, and, more recently, in China, which complicates research that requires global data for analysis, such as global climate change.
New measurements can be made from archived soil with new technologies and sampling approaches. To do so, however, archived soils must remain available, which requires funding and organization. “It is really important that we support retiring scientists and individuals who have built up large collections of valuable samples, and [that] these resources become a resource for our community going forward,” Hoyt said. Therefore, past knowledge and past soil samples must be retained because people cannot go back in time. However, even well-organized archives face challenges in terms of space and resources, she said, mentioning that the World Agroforestry Archives in Nairobi, Kenya, recently had to dispose of large amounts of soil samples. Many scientists in the United States or Europe likely would have been excited to analyze those samples, she said, but the archives did not have a well-established network to get those samples to the people who could use them.
Hoyt brought up another challenge: the lack of connectivity and integration in this field. “We can do much better [at] integrating both the people who work on these networks as well as the data themselves,” she said. Not every database needs to be connected with every other database, but in many cases, integration could result in synergy and the ability to answer previously unanswered questions. However, most of the world’s databases cannot talk to each other efficiently because of the lack of resources or standardization. The International Soil Carbon Network (ISCN) has served as a hub for many grassroots database efforts, she said, but new databases are constantly emerging, and many of them are not connected. “So there’s a much bigger challenge here than ISCN alone has been able to tackle,” she said.
Summarizing her talk, Hoyt reiterated that continental-scale and grassroots database efforts must work together to offer the best chances to understand the impacts of climate change and management on soil carbon. In particular, large-scale continental efforts are valuable in providing snapshots of current conditions or those in the recent past, as well as, one assumes, future conditions, while grassroots database efforts are better poised to understand the changes over time by leveraging past data.
To assure that this collaboration happens, Hoyt suggested support for both large-scale efforts and grassroots efforts as well as cross-sector centers to ensure integration of their data. Also important are long-term funding and systems for archiving soil samples “so that individuals who have really fantastic collections of samples also have a place to put them when they retire or to keep them organized over the course of their careers.” More generally, she continued, the mandate for archiving soil data must become stronger “so that we don’t have a continuous loss of data in the future and we’re actually making the most of all the samples that have been collected and analyses that have been run.” Hoyt concluded that grassroots efforts that leverage existing frameworks require additional support so that each new group interested in using the synthesized data to address a scientific question does not have to reinvent the wheel. Smaller grassroots efforts also need access to dynamic part-time resources, such as ad hoc guidance on database development and long-term maintenance.
DISCUSSION
Opening the discussion session, Charles Rice, an organizing committee member from Kansas State University, noted that Hatfield, in his presentation, listed microbial activity as a top driver of carbon. How, then, should microbial activity be measured, he asked, and what would be the appropriate temporal or spatial scale? Hatfield answered that quantifying microbial activity around the roots of plants is a major question in this area. Various parameters could play a role, from carbon dioxide levels to nutrient cycling, and even changes in the color of organic matter, but examining these parameters in the field is challenging. Microbial systems are one of the most dynamic aspects of soil systems, according to Hatfield, and finding ways to characterize them will likely require interdisciplinary science. Cornelius added that some new technologies might be useful. He stressed the utility of the EcoFAB platform to study plant–soil–microbe interaction. Results can then be used to inform models of real-world interactions. He also mentioned micro gas chromatographs, which are in development and can do chemical analyses at a very small scale and of very small samples.
Monica Farfan from the Global Soil Biodiversity Initiative commented via Slack that microbial diversity, abundance, and function are important, but soil fauna can inhibit or accelerate their activity. Collecting data on the animals that act as regulators of nutrient cycling should be considered, she noted.
Alfred Hartemink from the University of Wisconsin–Madison asked whether a dynamic soils information system should target only agricultural soils or all soils. “To me, it’s all soils,” Hatfield answered. Hartemink then asked Hoyt about grassroots efforts: Would it not make sense for these efforts to focus more on method development instead of data collection? Hoyt answered that bottom-up efforts are valuable for both tackling specific questions and developing methodology, “especially by providing comparisons between different particular approaches.” For example, with ISRaD, researchers are comparing different soil fractionation methods because much of the data have not been combined in the same place. Although challenging, systematic aggregation of the data enables comparisons of results from different methods, which in turn will generate recommendations to help the community better standardize practices going forward.
Michelle Wander from the University of Illinois at Urbana-Champaign asked Hoyt about the composition of the cross-sector centers that she had discussed. “Who are the sectors, and what will that look like? Or do we have any models?” Wander asked. Hoyt answered that many of the databases she referenced have focused on the role of soils in climate change. Both the soils community and the climate change community are generating important data, she added. Although the two communities overlap in understanding soil
processes and mechanisms, they have not done enough to share data and information about their activities and to support each other. Wander then commented on sharing data between government and academia on the one hand and industry on the other hand in efforts to create carbon maps and similar products. Engaging industry while maintaining transparency and openness, she said, will likely require the forging of new relationships. Hoyt agreed and expressed her hope that the workshop discussions might offer some creative ideas for researchers to make the most of proprietary data, even if not granted full access.
Basso then read a related question posed by Anna Cates from the University of Minnesota through Slack: What types of efforts are there to connect public scientific data with the data collections owned by private companies? Cornelius began his answer by saying that most of the interesting data—in terms of diversity and being on the cutting edge—are coming from the public sector. Data generated by industry tend to be narrow in scope because companies are trying to answer questions related to specific products. Although private data are valuable, the most creative work will likely be generated from public-sector data. Addressing the question directly, he explained that the private sector is very receptive to creating new open innovation models, and his organization has been working with industry on that issue. “I think that’s a significant opportunity for [the] public sector,” he said, because the public sector and private sector have different strengths, and collaborations to combine these strengths would benefit both sectors.
Phil Robertson from Michigan State University asked whether funders recognize the need for a set of calibration sites that can be used to compare and intercalibrate different instruments and methodological approaches. He explained his emphasis on sites rather than samples: the greatest source of variability for important properties such as carbon is often not at the archive sample scale, but at the field scale, “and I don’t know if this has been recognized by funders or developers yet or not.” Cornelius responded that creating such testbeds is imperative, particularly when various disciplines are integrated. When he worked for the Advanced Research Projects Agency–Energy, the U.S. Department of Energy (DOE) created a large robotic gantry system in Arizona to test the testbed concept. It has been largely successful, he said, and efforts to convince federal agencies to build testbeds are critical.
Hatfield agreed on the importance of testbeds and suggested that the National Ecological Observatory Network and USDA’s Long-Term Agroecosystem Research (LTAR) network could devote space at some of their sites to such calibration purposes. Jude Maul from USDA added via Slack that the LTAR working groups are currently developing data repositories and data collection guidelines that fit many of the soil testbeds’ attributes mentioned (e.g., long-term, historical data; defined treatments; consistent data collection). Hoyt said that, to a certain degree, such sites already exist. “The scientific community has certain sites where there’s large concentrations of measurements that have been made through time,” she said, for example, Harvard Forest. She added that different databases could be connected through the common measurements from these sites. There are, for example, databases focused on root traits, databases focused on soil carbon, and databases focused on radiocarbon and rates of turnover. All of these different databases have measurements from key sites of interest, often with long-term measurements. They could all be connected via those common sites.
Next, Mark Bradford from Yale University asked about the level of effort that should be devoted toward innovation for measurement versus toward understanding the measurements’ meanings, given that understanding of the implications of many measurements remains incomplete. “At the end of the day it is about knowledge,” Cornelius answered. “So, being able to measure something doesn’t guarantee anything.” For this reason, it is important to have as many eyes and brains as possible digesting the information coming out
of these systems, because it will be the public debate that teases out the important insights from that information. However, he added, he is unsure of appropriate ratio between innovating measurements and working to better understand existing ones.
Hoyt said that although some uncertainties cannot be resolved with more measurements, areas exist “where we definitely know that we can do better and where we’re missing fundamental knowledge.” For example, much of the world’s soil data are from North America and Europe, and many other places are under-sampled, which weakens the field’s ability to develop accurate global models, among other things. In this case, new measurement methods and technologies might not be as valuable as simply making better use of existing tools.
Avni Malhotra from Stanford University asked via Slack how much those who are developing new soil carbon measurement technologies were thinking about integration with existing data systems. Cornelius answered that integration largely depends on how programs are designed on the front end. Federal agencies such as USDA, DOE, and the National Science Foundation now require their programs to build on top of existing analytic pathways. Of course, there is always room for improvement, he said, and this evolving landscape is constrained by limited resources. Nevertheless, program design should anticipate how data will be imported into a system.
Vanessa Bailey from the Pacific Northwest National Laboratory referred to the discussion about measuring what is modeled. It is important to not diminish model-informed measurement guidance, she said. For example, if researchers had focused only on what could be measured, archived, and curated 20–30 years ago, the advent of new molecular microbial datasets would have been missed. Another example is high-resolution mass spectroscopy data, whose utility for understanding soils was uncertain only a short time ago. Now, within a few years of democratization of that instrument, researchers can understand nominal oxidation states of carbon and collect new datasets of reaction networks that transform the soil carbon cycle from simple photosynthesis-in/respiration-out models into metabolic models of how microbes are operating in different soils. Therefore, she said, researchers should always be on the lookout for the next bleeding-edge measurement, even if the value is, at present, uncertain. Hatfield agreed and reiterated his view that investigating soils is a transdisciplinary endeavor that requires a bigger tent to bring together more technologies to tackle the unknowns.
Connected with Bailey’s comment, Jason Sprung from the Bureau of Land Management noted on Slack that cutting-edge equipment is nice but cost prohibitive for data collection in remote field locations. At one time, USDA’s Natural Resources Conservation Service used soil quality kits with field-portable elements, and Sprung wondered whether any new technologies or techniques build on that type of simplicity to monitor indicators such as soil carbon. Erika Foster from Purdue University responded that the Nexus Institute (a collaboration between Purdue and the Universidad Nacional de San Agustín in Peru) has been updating a field soil health test kit, pairing some USDA soil quality tests with tests from the soil health kit created by Colorado State University researchers1 for use in remote sites in Peru. There could be an opportunity to collaborate and design a kit for wider use.
Also on Slack, participants discussed barriers to collaboration and digital data sharing. “How significant is a lack of embedded software/programmer/database expertise?” asked Cathy Chiba from Climate Data Hub. Kathe Todd-Brown, a member of the organizing committee, responded that the lack of programmer and database expertise is a substantial hurdle. Fenny van Egmond from the International Soil Reference and Information Centre added that demonstrating the added value for the data owner is another hurdle. For example, most databases are constructed using a template with manual data transcription.
___________________
Creating a robust template is challenging and often becomes an iterative process in which the initial work is redone. If a coder or database developer were on the team, then the template could be scripted to some extent and would better leverage metadata annotations. Todd-Brown added that data annotation is another significant issue; better semantic tools are needed to describe soil data to a level accurate enough to construct databases without manually pulling annotations from the original paper.
Irfan Ainuddin asked about the level of programming development expertise needed and the associated cost. Chiba wondered whether a lack of funding prevents the integration of developers into soil data teams or whether the relative invisibility of the work limits the field’s ability to attract this talent. Todd-Brown replied that a partial explanation is the need for a whole suite of experience and development skills: knowledge engineers, ontologists, database managers and developers, and others. In addition, some cross training is needed, for example, training the informatics experts in soils and training soil scientists in informatics.
This page intentionally left blank.
