8

Concluding Session

The final session summarized the workshop discussions and concepts presented, with a look to the future under the guise of keeping individuals in the field engaged with soil data and informatics after the workshop. Organizing committee members Jim Tiedje, Rodrigo Vargas, and Kathe Todd-Brown each presented a piece of this summary. Tiedje synthesized the workshop presentations, while Vargas described the contributions to the workshop that had been made via Slack. Todd-Brown spoke about continuing engagement opportunities. Then, Bruno Basso offered some concluding remarks.

SYNTHESIS SESSION

Jim Tiedje

Tiedje began his workshop synthesis with a look at the ways in which a dynamic soil information system is valuable. In agriculture it can lead to benefits such as increased productivity and profitability and maintained or even enhanced land quality. Environmentally, its benefits include improved ecosystem services, improved soil and water quality, and mitigation of the global climate effects of greenhouse gases. A dynamic soil information system also offers value to fundamental planetary science, Tiedje said, helping to test hypotheses about scale, trajectories, and the lithosphere and providing data to study biological, chemical, and physical interactions, as well as ecological and evolutionary changes and the mechanisms behind them.

Referring to the keynote presentation by Jerry Hatfield, Tiedje then spoke about the functions of soil. Soil supports crops, cities and roads, forests, tundra, and so on. Tiedje noted that researchers do not know the land uses of the future, and that the information we gather now will help inform those future uses. Soil also has various uses related to water—its storage and recycling as well as its distribution and safety, for instance—and to nutrient supply, reliable biogeochemical cycles, and planetary terrestrial clean-up. As Hatfield said in

his presentation, understanding the functions of soil is all about water and carbon, as well as dynamics and interdependencies.

Speaking next about Joe Cornelius’s keynote, Tiedje recalled that Cornelius spoke about measurement—what can be measured and what should be measured. Core chemical and physical measures have traditionally been at the heart of soil information systems, but Cornelius also spoke about plant roots serving as a window to what is happening below the surface. Today there is a variety of new types of measurement tools, from microsensors to drones and satellite-based technologies. Cornelius’s presentation reviewed how these tools can be applied at different scales and over time.

In her keynote, Alison Hoyt talked about soils at a more global scale, Tiedje said. She spoke about management, change over time, and projections for the future. She also emphasized the importance of historical data, time series, and archives as extremely valuable resources. Finally, she discussed the value of connectivity, the importance of data integration, and the benefits from using diverse approaches to collecting soil data, from continental-scale efforts to smaller-scale grassroots projects.

There is little disagreement about the importance of collecting large amounts of data, Tiedje said, and soil sciences are now in an era of “big data,” which is changing the ways that researchers collect, analyze, and share their data. Tiedje noted that two of the breakout session summaries dealt with data-related issues. He believes that at least one-third of soil-related work will occur in the domain of big data going forward.

Tiedje provided a quick overview of progress to date with soil information systems. A great deal of data has been collected, he said, with much of them archived, but the data are in varying states, in various locations around the world, and largely inaccessible. Many of the data are at risk, particularly those collected in an earlier era before modern data management began. Complementing the data are archived soils. Many of these soils are air-dried, which damages their usefulness for biological measurements, but it may be possible to correct for this in some way. The archived samples are particularly valuable because they provide time-series records of how soils changed over time. As an example of the value of such archived soils, Tiedje highlighted the well-known example of Dutch stored soil samples whose analysis first indicated the problem of antibiotic resistance in the environment.

Next, Tiedje noted, a great deal had been said in the workshop about current applications of soil information systems—and, importantly, what lessons have been learned. That is, what have researchers learned about the value of these systems, and how are they communicating that value? What has been learned about the different levels of effort and the resources needed to operate and sustain these systems? What is known about the users of these systems? How can one effectively communicate with users and work to extend the user base? What has been learned about nimbleness? Sometimes systems are very structured, which can be good but may limit future adaptability.

Some of the most important learnings from the workshop presentations, he said, relate to the challenges associated with dynamic soil information systems. Those challenges include finding continuous funding for monitoring, getting sufficient staffing for soil science and data analytics, solving various issues related to data privacy and security, harmonizing data and developing common methods for sampling and analysis, and capturing spatial and temporal data at varying scale. Recommendations for meeting these challenges, which were heard both in the listening sessions and at the workshop, included performing more repeated measurements, improving communication among agencies, increasing the amount of metadata that describe samples and methods, and archiving soils for future analysis.

Is the vision of a dynamic soil information system big science? If so, Tiedje said, then the system should be planned for as big science. This planning involves organizing at several

levels and with multiple components, making sure that the appropriate leadership and community building are in place, developing major guiding science questions, building a support base in multiple sectors, and involving people from multiple disciplines.

Pursuing a dynamic soil information system as big science does have potential downsides, Tiedje said. It can, for instance, fragment the community because not everyone may be on board with this approach. The system could end up being overdesigned and sink from its own massive weight. “There’s always a tendency to measure everything everywhere all the time,” he said. Previous efforts have experienced difficulty with getting started because of overdesign. Furthermore, it is difficult to obtain the long-term commitment necessary to sustain such a big science project, particularly in a field with historically short-term funding. It can also be challenging for a project to remain dynamic and visionary and not lose focus, perhaps by becoming too diverse.

What if the worse happens and the big science effort fails? Is it all wasted? No, Tiedje said. The planning that went into the effort will provide some valuable pieces that can be pursued at a smaller scale. Even if the major effort does not receive all of the funding needed for success, it will probably attract more funding than would a collection of less ambitious, smaller-scale projects.

To have the greatest chance for success, Tiedje said, the field must learn from other models. The soil sciences community could learn from the work of other communities, such as the marine science and oceanography community, as they are the most similar. Soil scientists may have easier access to their sites, and marine scientists may not have as much small-scale heterogeneity, but otherwise the questions and challenges that they face are very similar. The soil sciences community could learn from atmospheric sciences, astronomy, biology, and the geosciences, most of which have large-scale, long-term research efforts, and some of which are planned out for a decade at a time. By studying these other areas as exemplars, the soil sciences community could learn valuable lessons about infrastructure needs, organizational strategies, and science drivers.

Soil scientists can also learn a lesson about funding large programs from physicists, Tiedje said. In neutrino physics, for example, the entire community comes together to develop one large proposal. Together, the physicists decide on what is important and then maintain unity around agreed-upon major goals. As part of this effort, the community develops decadal plans with goals set in 10-year increments. In this way, the neutrino physics field is able to obtain large-scale funding, he said, and the principle of organizing around goals that have been agreed upon by the community could be applied in the soil sciences.

Tiedje offered a different lesson from GenBank, the U.S. repository for genetic data. At GenBank’s inception, there was a lot of discussion about data quality. Should genetic sequences be carefully checked before release to the public, or should they be released upon receipt? Some people thought it was a mistake to release data whose quality could not be assured. Ultimately, though, the decision was made to release sequence data publicly before it was checked, which turned out to be the right decision, he said, because checking sequences before their release would have caused a huge backlog and delayed release by many years. It is better to have the information out there, Tiedje said, where many people can review and correct it. The soil sciences community can learn from this lesson as well, he said.

Tiedje also relayed a story about the importance of audiences beyond the scientific community to building momentum toward a dynamic soil information system. Some people want to learn how scientific discovery can make money, while others want to learn about fundamental advances in scientific knowledge. Another audience is interested in the importance of scientific discovery for the environmental sustainability of the planet. A dynamic

soil information system can provide utility to all of these audiences, but scientists must be able to tell that story.

Tiedje concluded by sharing his experience in a National Academies study to scope and explore the field of metagenomics when it was emerging in the 2000s. At the time, it was important to Tiedje to focus on not only genetic sequencing but also data analysis and computational science. Metagenomics eventually became the field of microbiome science and grew in terms of resources and effort. Tiedje expected that the field would change over time, starting with observational outcomes, then hypothesis-driven outcomes, and finally more experimental hypothesis-driven science (see Figure 8-1), which turned out to be the case. So what is next? The field of metagenomics highlights the need to have a vision, and Tiedje advised the participants to adopt a vision to help guide their efforts to develop a dynamic soil information system. “And so I would challenge people,” he said, “just to draw your vision for the next 10 years for this topic area we’re discussing. What would be the component parts, how it would develop over time, and how it would grow? … It’s important to have that vision for the next decade.”

Rodrigo Vargas

Following Tiedje, Vargas summarized communications that arrived via Slack during the workshop. He began with challenges. A lot of information is available, he said, so how can it be brought together? It will be important to increase connectivity and incentivize different parts of the community to share data. Archiving physical samples will be another challenge. Who will pay for it, and how will it be done? Data fidelity is important, Vargas said, and not only the fidelity of data from point measurements and the laboratory, but also the fidelity of data from value-added products such as maps and machine learning products. Finally, work to unify the soil sciences community will be particularly challenging, especially

because development of a dynamic soil information system will require big science, as Tiedje discussed.

Vargas then shared two related opportunities that were discussed on Slack. First, an opportunity exists to link large amounts of spatially distributed data and create models that bridge various areas, from physics to biology. This work would lead to another opportunity, that is, the ability to provide various stakeholders with different types of information that they will find valuable.

Another important point made on Slack, Vargas said, concerns the availability of new computer science tools such as confidential computing and encrypted data. “We want to have data available,” he said, “but we also have to think about national security issues, policy issues, commercial issues, etc.” Participants also discussed the use of artificial intelligence, which could help users to easily find the data and tools they need.

To reward researchers and others for their contributions to the data information system, Vargas said, tracing information as it moves through the system must be straightforward. This could be achieved through the use of persistent identifiers, which enable linkages between data and their authors and/or owners. Such information is important for agencies that wish to track the impact of data produced by their employees or grantees, as well as for independent researchers who wish to know how their data have been used.

Vargas also noted that conversation on Slack emphasized the importance of defining ontologies for soil and communicating the meanings of different variables and the process to define and quantify uncertainty. It is important that users of a soil information system clearly understand what the uncertainties are.

The applicability of a dynamic soil information system to grand challenges also needs to be kept in mind, he said. How can it be applied to helping solve problems such as food insecurity and climate change? Ultimately, this effort will require a subset of teams from the community working on different questions related to these grand challenges.

Important considerations for moving forward, Vargas said, include education, extension, and outreach efforts. Collecting data is critical, but those data need to be interpreted and applied to answering questions, which requires training the next generation of students to access that information, analyze and interpret it, and then gather more information. Such education is an effort that will require many sectors of the community to work together—federal agencies, the private sector, the research community, and even the citizen science community.

Data sharing is crucial, yet requirements attached to most federal grants to share resultant data are not well enforced, Vargas said. Therefore, discussion is needed to identify how to best enforce data sharing, whether from the funders’ side or the journals’ side.

Finally, he said, there is the question of where the funding for this effort will come from. There is much to be funded—data collection, data storage, data mining and analysis, and also education. Without funding, nothing else is possible, so who will pay for all of this?

CONTINUING ENGAGEMENT OPPORTUNITIES

For the workshop participants who wished to remain engaged with soil data and soil informatics, Todd-Brown listed a number of opportunities and resources that were described during the workshop. She began with a list of U.S.-based resources, which was also posted on Slack:

• U.S. Department of Agriculture’s Natural Resources Conservation Service soil database interface¹
• National Microbiome Data Collaborative²
• Agricultural Collaborative Research Outcomes System³
• Long-Term Agroecosystem Research Network⁴
• National Aeronautics and Space Administration’s (NASA’s) Soil Moisture Active and Passive⁵
• National Geochemistry Survey database⁶
• Long-Term Ecological Research Network⁷
• Consortium of Universities for the Advancement of Hydrologic Science, Inc.⁸
• National Ecological Observatory Network⁹

Participants from Slack also added:

• Northeastern Soil Monitoring Cooperative¹⁰
• NASA’s Carbon Monitoring System¹¹

Next, Todd-Brown offered some data resources, beginning with websites with soil data:

• Soils Revealed¹²
• International Soil Reference and Information Centre Soil Data Hub¹³
• SoilGrids web portal¹⁴
• European Space Agency’s WORLDSOILS¹⁵
• Soils 4 Africa¹⁶
• OpenGeoHub—OpenLandMap¹⁷

Participants from Slack also suggested:

• iSDASoils¹⁸
• Australian Microbiome Initiative¹⁹
• Sistema de Información de Suelos de Latinoamérica y el Caribe²⁰

___________________

¹ See http://ncss-tech.github.io/soilDB/docs.

² See https://microbiomedata.org.

³ See https://agcros-usdaars.opendata.arcgis.com.

⁴ See https://ltar.ars.usda.gov.

⁵ See https://smap.jpl.nasa.gov.

⁶ See https://mrdata.usgs.gov/geochem.

⁷ See https://lternet.edu.

⁸ See https://criticalzone.org/learn-more.html.

⁹ See https://www.neonscience.org.

¹⁰ See http://www.uvm.edu/~nesmc/index.html.

¹¹ See https://carbon.nasa.gov.

¹² See https://soilsrevealed.org.

¹³ See https://www.isric.org/explore.

¹⁴ See https://soilgrids.org.

¹⁵ See https://www.isric.org/projects/esa-worldsoils.

¹⁶ See https://www.soils4africa-h2020.eu.

¹⁷ See https://gitlab.com/openlandmap/compiled-ess-point-data-sets and http://www.openlandmap.org.

¹⁸ See https://www.isda-africa.com/isdasoil.

¹⁹ See https://www.australianmicrobiome.com.

²⁰ See http://54.229.242.119/sislac/es.

• European soil data, including Land Use and Coverage Area frame Survey²¹
• Soil Moisture Active Passive/Global Learning and Observations to Benefit the Environment partnership²²

Next, Todd-Brown offered what she referred to as a “grab bag” of different types of resources. MESONET²³ offers climate data to help contextualize soils. Colorado State University researchers have created a soil health kit.²⁴ Open Data Science²⁵ is more generally focused. Soil Spectroscopy 4 Global Good is focused on building an open-source soil spectroscopy library for machine learning and calibration validations.²⁶ The Global Soil Biodiversity Initiative may be of interest to those interested in omics data and soil biology.²⁷ Finally, the International Soil Moisture Network²⁸ and the International Soil Carbon Network²⁹ are involved in data collection and integration. Participants on Slack also mentioned the tropical soil spectral library,³⁰ the Global Long-term Agricultural Experiments Network,³¹ the Open Geospatial Consortium (which includes an agriculture domain working group),³² the World Wide Web Consortium for work on ontologies,³³ and the International Union of Soil Sciences working groups.³⁴

Furthermore, a number of non-soil repository and information systems might be of interest to participants, she said, mentioning in particular the World Resources Institute, Resource Watch, the Global Biodiversity Informatics Facility, the Avian Knowledge Network, and Consortium of Universities for the Advancement of Hydrologic Science, Inc.’s Hydrologic Information System. In addition, the breakout group on collection and curation offered some suggestions: Two were from the U.S. Department of Energy’s (DOE’s) Joint Genome Institute, specifically its Functional Genomics Database for Plant Microbiome Studies³⁵ and its National Microbiome Data Collaborative,³⁶ which has a trellis interface, a data visualization tool for exploring data collection. A participant on Slack also noted that the Argonne National Laboratory has many decades of archived soils and that DOE’s Environmental System Science Data Infrastructure for Virtual Ecosystem database³⁷ could hold soil data. The breakout group discussion also generated the suggestion of Zooniverse,³⁸ which is dedicated to citizen science.

Todd-Brown mentioned two funding opportunities in soil sciences, Signals in the Soils and the new Center for Advancement and Synthesis of Open Environmental Data and Sciences, both of which are National Science Foundation programs. “My personal hope is that when the center gets funded, there is a large soils component,” she said.

___________________

²¹ See https://esdac.jrc.ec.europa.eu.

²² See https://www.globe.gov/web/smap/overview.

²³ See https://climate.umt.edu/mesonet/default.php.

²⁴ See https://smallholder-sha.org/protocol-1.

²⁵ See https://opendatascience.eu.

²⁶ See https://soilspectroscopy.org.

²⁷ See https://www.globalsoilbiodiversity.org/objectives.

²⁸ See https://ismn.geo.tuwien.ac.at/en.

²⁹ See https://iscn.fluxdata.org.

³⁰ See https://www.worldagroforestry.org/sd/landhealth/soil-plant-spectral-diagnostics-laboratory/soil-spectra-library.

³¹ See https://glten.org.

³² See https://www.ogc.org.

³³ See https://www.w3.org.

³⁴ See https://www.iuss.org/organisation-people/organisation/working-groups.

³⁵ See https://jgi.doe.gov/functional-genomics-database-for-plant-microbiome-studies.

³⁶ See https://jgi.doe.gov/join-national-microbiome-data-collaborative-trellis.

³⁷ See https://ess-dive.lbl.gov.

³⁸ See http://zooniverse.org.

Several soil data organizations are focused on answering specific questions. The Smithsonian Institution’s Coastal Carbon Research Collaboration Network³⁹ is a data harmonization and integration effort specifically focused on coastal soil. The Ecological Forecasting Initiative⁴⁰ is simulating entire ecosystems and calibrating biogeochemistry and hydrology in an integrated platform. The International Soil Modeling Consortium’s⁴¹ DO-Link science panel is focused on data and observations and linkages to soil modeling. The International Soil Radiocarbon Database,⁴² mentioned by Alison Hoyt in her keynote talk, is gathering soil radiocarbon datasets to create a single collective database. A participant on Slack also suggested the Permafrost Carbon Network.⁴³

For people who would like to attend conferences where soil sciences are discussed, Todd-Brown offered a number of suggestions, beginning with the American Geophysical Union and the European Geophysical Union, both of which have soil informatics groups that meet regularly. The Soil Science Society of America, the Ecological Society of America, the Soil Ecology Society, and the Geological Society of America are slightly smaller but host small groups and conferences to discuss soil informatics. Two semiannual meetings hosted by the Earth Science Information Partnership and the Research Data Alliance⁴⁴ are somewhat more focused on soil data and soil components specifically. International Data Week, which occurs every other year, is sponsored by the International Science Council’s Committee on Data and by the Research Data Alliance and is “sort of a conference call … that just talks about data in general, and there’s frequently soils components,” she said. The National Cooperative Soil Survey biannual national conference was also mentioned by a participant on Slack.

For people interested in finding working groups devoted to collaborative science, Todd-Brown offered several suggestions. The Earth Science Information Partnership has a number of specialty interest groups—“clusters”—that meet regularly to talk about data issues in a particular area.⁴⁵ In particular, she said, she co-chairs a soil ontology and informatics group that workshop participants would be welcome to join, and there is also an agriculture and climate cluster. The Research Data Alliance has several agriculture-focused working groups. Finally, she said, two Global Soil Partnership working groups are focused on lab harmonization and lab quality improvement⁴⁶ and on soil spectra library development.⁴⁷

CONCLUDING REMARKS

In his closing remarks, Basso spoke about what will be needed to create a dynamic soil information system. It will require a coordinated interdisciplinary approach with multiple objectives, from monitoring minimum datasets at benchmark sites using different techniques (sampling to remote sensing) to modeling (machine learning to process-based) to develop insight that can be used to guide decisions. The scale should be clearly identified in each sub-objective (e.g., soil biology), and the resulting data and knowledge need to be made widely available so as to help ensure the continuance of healthy soils, said Basso.

___________________

³⁹ See https://serc.si.edu/coastalcarbon.

⁴⁰ See https://ecoforecast.org.

⁴¹ See https://soil-modeling.org/science-panels/DO-Link.

⁴² See https://soilradiocarbon.org.

⁴³ See http://www.permafrostcarbon.org.

⁴⁴ See https://www.rd-alliance.org.

⁴⁵ See https://www.esipfed.org/get-involved/collaborate.

⁴⁶ See http://www.fao.org/global-soil-partnership/glosolan/en.

⁴⁷ See http://www.fao.org/global-soil-partnership/glosolan/soil-analysis/dry-chemistry-spectroscopy/en.

It will be important for those in the field to come together and decide what is important to measure, Basso said. “Like Jim said, we can’t measure everything, everywhere, any time.” The particular questions and objectives behind the system should direct choices about what to measure, and models can be used to get a clear picture of systems beyond what is possible to measure.

The best option for funding a dynamic soil information system may be a public–private partnership, Basso said, where funding is divided up according to interests and benefits. The initial funding might need to come from public sources, but private groups may join in later.

Basso closed by thanking the various individuals and organizations that made the workshop possible—the members of the workshop organizing committee; National Academies staff; workshop speakers, panelists, and audience members; and the workshop’s funders, including the National Science Foundation, The Nature Conservancy, USDA’s National Institute of Food and Agriculture, USDA’s Natural Resources Conservation Service, and DOE’s Advanced Research Projects Agency–Energy, with additional support from the National Academy of Sciences’ Arthur L. Day Fund and the National Corn Growers Association.

This page intentionally left blank.