FACT: AN INNOVATIVE CYBER-FRAMEWORK INTEGRATING PUBLIC/PRIVATE DATA FOR EVIDENCE-BASED RECOMMENDATIONS

• Sponsoring Institution: National Institute of Food and Agriculture
• Project Status: ACTIVE
• Funding Source: AFRI COMPETITIVE GRANT
• Reporting Frequency: Annual
• Accession No.: 1019790
• Grant No.: 2019-68017-29939
• Cumulative Award Amt.: $1,000,000.00
• Proposal No.: 2018-09189
• Project Start Date: Aug 1, 2019
• Project End Date: Jul 31, 2025
• Grant Year: 2019
• Program Code: [A1541]- Food and Agriculture Cyberinformatics and Tools

Recipient Organization
PURDUE UNIVERSITY
(N/A)
WEST LAFAYETTE,IN 47907

Performing Department
Agronomy

Non Technical Summary
In the 1990s, access to geo-referencing tools launched precision agriculture, a farm management concept to improve farmer profits and environmental quality by understanding and managing naturally occurring variability in soils within a field. Since then, there has been an explosion of data generated on the farm to create layers of maps such as yields and soil test results to try to determine exactly what amount of fertilizer is needed for very small areas within fields. However, many of the best management practices, recommendation frameworks, and associated tools for on-farm decision making have not been updated. Indeed, many fertilizer guides have proven ineffective for precision agriculture, which is unsurprising given that these historic recommendations were developed for average conditions across fields and farms and are not finely tuned enough to be meaningfully applied to specific sub-environments within fields. Further, despite extensive public research investments in studies to optimize fertilizer management, the rigorous synthesis of these studies has not been done; statistically sound synthesis across studies is a necessary first step to making recommendations that can be tailored to sites, soils, and farm enterprises as envisioned for precision agriculture. Reasons why studies have not been synthesized include (i) a lack of appreciation, knowledge and culture for "synthesis" science, (ii) publication and communication systems that prioritize novel results over confirmatory or null studies, thereby skewing the official scientific record as a foundation for synthesis; and (iii) lack of infrastructure and tools to facilitate use and reuse of accruing data relevant to management objectives.Long-term Goal and Supporting Objectives: We hypothesize that creating better infrastructure for science synthesis, coupled with mechanisms for using on-farm data, will deliver recommendations with more localized relevance, on-farm value and compliance with emerging farm-to-fork sustainability metrics. Our overarching goal is to provide proof-of-concept to a cyber-enabled process linking public and private data and create a model ecosystem and community for public-private research and clinical trials for continuous improvement of the evidence for management practices. We will focus on phosphorus (P) and potassium (K) management recommendations as case studies for the following Main and Facilitating objectives:Main Obj. 1: Create a Cochrane-style collaborative (https://www.cochrane.org/) for research synthesis that employs open-access cyber tools for systematic reviews with statistical meta-analyses as the foundation for evidence-based practice and incentivizes open data in public agronomic research;Main Obj. 2: Create a cyber-framework and community ecosystem for profitable, evidence-based decision making and resource stewardship that ingests on-farm data, integrates it with research data, and recommends locally relevant optimization vs generic strategies;Facilitating Obj. 3: Develop a 10-credit certificate program for graduate students and practicing professionals emphasizing data / computational literacy and rigorous synthesis of disparate data. Main objectives will develop and implement best practices, policies and tools for (i) public-private partnerships for data use in evidence-based practice inclusive of privacy, security, cost and intellectual property concerns (ii) triage and recovery of high-value legacy data and (iii) conversion and sharing of grey / dark data from null or confirmatory studies in FAIR (findable, accessible, interoperable re-usable) formats. The facilitating objective will prepare graduate students for conducting data-intensive syntheses of agronomic research and on-farm data and will equip practitioners for site-specific management using their data. This proposal will create key elements for our ultimate vision: a cyber-framework for customization that ingests and anonymizes user data, combines ingested with existing research and on-farm data, and uses artificial intelligence and machine learning to fine-tune recommendations with accruing new data to minimize human resource requirements.

Animal Health Component

45%

Research Effort Categories

Basic

10%

Applied

45%

Developmental

45%

Classification

Knowledge Area (KA)	Subject of Investigation (SOI)	Field of Science (FOS)	Percent
102	1510	1070	12%
102	0110	1070	26%
102	1820	1070	12%
102	7299	2080	30%
102	5210	3010	20%

Knowledge Area
102 - Soil, Plant, Water, Nutrient Relationships;

Subject Of Investigation
1510 - Corn; 0110 - Soil; 1820 - Soybean; 7299 - Research equipment and methods, general/other; 5210 - Fertilizers;

Field Of Science
1070 - Ecology; 2080 - Mathematics and computer sciences; 3010 - Economics;

Keywords

fiar

open data

evidence-based recommendations

legacy data

public-private partnerships

Goals / Objectives
Long-term Goal and Supporting Objectives: We hypothesize that creating better infrastructure for science synthesis, coupled with mechanisms for using on-farm data, will deliver recommendations with more contextual relevance, on-farm value and compliance with emerging farm-to-fork sustainability metrics. Our overarching goal is to provide proof-of-concept to a cyber-enabled process linking public and private data and create a model ecosystem and community for public-private research and clinical trials for continuous improvement of the evidence for practice. We will focus on phosphorus (P) and potassium (K) management recommendations as case studies.To achieve our goal of providing proof-of-concept to a cyber-enabled process facilitating a model community ecosystem for a public-private research and clinical trials, 6 tasks will be pursued under our two main objectives with an additional task associated with the facilitating objective of curriculum development. Objectives and tasks are:Main Obj. 1. Create a Cochrane-style collaborative (https://www.cochrane.org/) for research synthesis that employs open-access cyber tools for systematic reviews with statistical meta-analyses as the foundation for evidence-based practice and incentivizes open data in public agronomic research by:Task 1 (T1). Create a prototype model for agricultural data including data and metadata standards; surface legacy data and transform to FAIR formats and develop best practices for data recovery and sharing, andT2. Synthesize FAIR data from T1 with peer review literature/data with Systematic Review (SR) to test hypotheses about existing P/K recommendation algorithms and potential alternatives.Main Obj. 2. Create a cyber-framework and community ecosystem for profitable, evidence-based decision making and resource stewardship that ingests on-farm data, integrates it with research data, and recommends locally relevant optimization vs generic strategies by:T3. Evaluate T2 outputs (recommendation algorithms and key parameters) with high-quality on-farm data provided by Advanced Agrilytics (AdvAgry);T4. Apply machine learning (ML) to AdvAgry on-farm data to search for important yield drivers and low cost proxy measures currently not included or deemed theoretically important in P/K recommendations;T5. Deploy improved P/K recommendations via iOS and Android apps and apply lessons learned in T1 - T4 to develop and implement a roadmap for rapid and efficient updating WI P/K recommendations, andT6. On-farm testing of new recommendations in collaboration with farmer networks; data profitability analysis to make the business case for on-farm data value and focus groups to understand needs for policies and tools for public-private partnerships for data use in evidence-based practice, inclusive of privacy, security, cost and intellectual property concerns.Facilitating Obj. (T7). Develop a 10-credit certificate program for graduate students and practicing professionals emphasizing data / computational literacy and rigorous synthesis of disparate data.

Project Methods
T1. Ag Data Model: Although, at present there are no universally accepted data and meta-data standards for soil fertility/plant nutrition (SF/PN) research, agricultural standards are under active development; we will conduct an environmental scan for these resources. The Dublin Core Metadata Schema will be referenced for developing controlled vocabularies and used together with other specialized vocabularies to meet implementation requirements for specific domains. We anticipate using the NAL Thesaurus and Glossary (https://agclass.nal.usda.gov/) and other resources to create the foundation for the pruning across ontologies to select candidate standards. We will examine legacy datasets to identify and understand commonalities in data organization and description and to determine the layers of elements for an integrated data model for full lifecycles. A primary resource will be the raw datasets already accessible to CoPIs as well as data from state experiment stations, Extension reports, conference proceedings and other grey literature reports available through NAL. Focus groups will solicit feedback from SF/PN colleagues on prioritization criteria and triage strategies for legacy data; we will create a webhosted listing of at-risk datasets. Our decisions will be informed by the Reference Model for Open Archival Information Systems. The AdvAgry on-farm database will provide an independent, private data model for comparison. Our ongoing collaboration with NAL's AgDataCommons will ensure that our data will be interoperable with data in NAL from other AFRI-NIFA projects.T2. Systematic Review: Proof-of-concept of the data model and the value of making all primary data FAIR will be in using harmonized, small study data from T1 with additional data extracted from the peer-review literature in a synthesis analysis. In general, we will follow the methodological guidelines for SRs developed by the Collaboration for Environmental Evidence (http://cebc.bangor.ac.uk/); specific activities include 1) characterizing the need for evidence and initial logistics, 2) planning the review, 3) conducting the review, 4) reporting results, and 5) disseminating outcomes. We apply an array of on-line, open-access resources (e.g. tools provided via systematicrevietools.com). If possible we will conduct a quantitative synthesis using open source software (R metafor) to summarize effect sizes with appropriate weighting. We will also examine the impact of grey literature and dark data on synthesis outcomes.T3. Rec. Eval. with On-Farm Data: We will use the high quality, on-farm dataset available from AdvAgry to test the relevance of the existing recommendation algorithms as well as any new parameters identified from the T2 SRs with potential to improve P/K recommendations in conjunction with or separately from soil testing. For example, the current recommendations for K include cation exchange capacity as a covariate, but the underlying scientific evidence for its inclusion is not understood. To test this and other hypotheses we will use a variety of statistical tools (regression, binary logistic regression, principle component and cluster analyses, etc.) to interrogate and explore the on-farm dataset looking for confirmatory associations. We anticipate this analysis to validate the T2 SR results but also expect additional attributes, especially those not in the original meta-analysis database, to emerge as informing to P and K recommendations. We will also evaluate two important and persistent questions with economic implications: 1) at what intensity (cores/ha) should farmers sample soils for P-K analysis and 2) what soil, land, and environmental characteristics influence or interact with optimal sampling intensity? Using the AdvAgry data and a Bayesian decision theoretic framework, the economically optimal soil sampling density will be determined using an objective function to balance the expected costs and benefits of sampling using a Bayesian prior distribution for the unknown parameters.T4. Machine Learning w/ On-Farm Data: To uncover variables that are drivers of highest yields and could improve recommendation site specificity, we will use ML/AI to mine/explore associations between yield levels and all other factors to identify important but overlooked covariates and efficienct proxy measures. For each data item in the AdvAgry dataset, we will treat crop yield as the response variable yand all available measurements as the covariates x (a feature vector). We will construct a data set with n training items {(x1, y1) ... (xn, yn)}. Note, the covariates capture specific information for individual subfield, which allows our eventual recommendations to be adaptive and optimized at subfield scales. Subsequent activities are: (1) fitting of models aimed at interpretability and at predictive accuracy, and (2) exploratory qualitative and quantitative analyses to gain human insight into the effect of different covariates on yield and to give numerical recommendations on covariate adjustments that have the potential to improve yield.T5. Rec. Deployment: Anticipated changes to existing recommendations include classification and decision trees based on novel covariates (e.g. soil series, texture, organic matter content, within field elevation) and willingness to assume various risk associated with input costs in context of probability of yield response, etc. We will include both agronomic outcomes (i.e., yield maximization) and economic outcomes (i.e., profit maximization) such that farmers can incorporate individual constraints on input allocations. Obtaining economic optimals is a relatively easy extension once the production functions are known, needing only to find the point where the input-output price ratio is tangent to the production function. Deployment of the new framework will be facilitated by modifying existing Purdue Extension apps and lessons learned from T1-4 will be incorporated into a roadmap for cost-effective updating of WI P/K recommendations.T6. On-farm Testing/Business Case: We will couple on-farm evaluations with focus groups to understand famer concerns about cyber-tools that are founded on meshing public research data with on-farm monitoring, education needs, and other clinical issues surrounding evidence-based, data-driven practices. Two on-farm networks will evaluate new recommendations by comparing the yields and other agronomic metrics with current farmer practice; we will include fields with a range in soil types in order to expand the inference space. A partial budgeting framework will evaluate the changes in revenues and costs associated with applying the new recommendation as compared to FP. Additionally, we will provide a simple economic evaluation of the data collection and storage aspects of this project by examining costs and benefits of tracking additional field/weather characteristics and assessing which variables provide a marginal value greater than the marginal cost of collecting/archiving data.T7: Certificate: We will leverage recent investments in content and distance education delivery by Purdue's Libraries and the College of Agriculture. The certificate will focus on educating graduate students to manage data for FAIR, and to access and synthesize large datasets for decision making. Required courses will be: (1) Data Management at the Bench (3 cr.), (2) Data Sharing and Publication (3 cr., new), (3) Introduction to Information Strategies and Systematic Review for Agronomic Sciences (3 cr.; new) and (4) Emerging Topics in Digital Agriculture (1 cr., may be repeated). We will use our on-farm datasets to create modules that guide students as they work with a variety of raw and processed data streams. The Agronomy e-Learning Academy will modify content from the certificate for asynchronous delivery to on-farm decision makers (https://ag.purdue.edu/agry/ADE/Pages/default.aspx).

Progress 08/01/23 to 07/31/24

Outputs
Target Audience: We have continued to reach soil fertility researchers and Extension specialists through our collaboration with the Fertilizer Recommendation Support Tool project (FRST; https://frst.scinet.usda.gov/). In addition to monthly meetings, PI Brouder co-organized a special panel on data stewardship and sharing and presented on behalf of the FRST projectat the NCEAR013-SERA6-NECC1012 Joint Meeting in Minnesota; the intent of the panel was to encourage all soil fertility researchers to share their data by highlighting the value of sharing and raising awareness of federal funding agency requirements and of institutional resources to facilitate stewardship. PI Brouder has now been invited to give the same presentation to the ESCOP Science and Technology Committee at their September meeting. This project's work on data stewardship has also been recognized by the Efficient Fertilizer Consortium (EFC) of the Foundation for Food and Agriculture Research and other NIFA-funded projects. Our lessons learned on grey literature, data and data sharing were shared with key EFC stakeholders (researchers, funding agencies and policy makers) during a 2-d workshopin Washinton, D.C. and have now been incorporated into a manuscript intended to guide soil fertility researchers on how to conduct field research that is inclusive of environmental metrics on water quality and greenhouse gas emissions. We anticipate our audience growing in the coming year as we expand our international efforts. Recently PD Brouder and co-PI Volenec have been asked to participate in the Digital Decision Agriculture (D2Ag) consortium. This is a component of the US-AID funded Climate Resilient Sustainable Intensification Innovation Lab awarded to Kansas State University. The D2Ag consortium includes partners from Cornell University, the University of Florida, the African Plant Nutrition Institute, the International Rice Research Institue, along with collaborators from industry (Syngenta, Corteva Agriscience). This program will train agricultural scientists and leaders in Cambodia, Bangladesh, Ethiopia, Tanzania, Senegal, Ghana, Guatemala and Honduras on best management practices for developing workflows, processes, and repositories that will enable curation of high-priority agricultural datasets in these countries. Partnering with Co-PI Murrell at Mohammad VI University in Morrocco is leading to another opportunity to introduce concepts of systematic reviews, data curation and meta-analysis to stakeholders and students in Africa and beyond. OCP Group, a large fertilizer manufacturer headquartered in Morrocco is collaborating with Mohammad VI University and the African Plant Nutrition Institute on a project aimed at understanding the impacts of blended versus mono-nutrient fertilizers on agronomic and environmental performance. Launching in the coming months, this related project will broaden the audience impacted by the knowledge acquired under this project. Co-PI Volenec helps coordinate the creation and developmentof the Forage Data Hub, a platform sharing datasets for agricultural system resilience. This hub currently contains 42 MB of data compiled from 108 unique locations across the United States, with harvest dates ranging from 1958 to 2022 [Ashworth, Amanda; Berti, Marisol; Foster, Jamie; Brewer, Linda; Casler, Michael; Bacon, Tyler; et al. (2023). Data from: Framework to Develop an Open-Source Forage Data Network to Improve Primary Productivity and Enhance System Resiliency. Ag Data Commons. Dataset. https://doi.org/10.15482/USDA.ADC/1529174)]. It is a component of the Resilience CAP project (AFRI Sustainable Agricultural Systems Coordinated Agricultural Program (SAS-CAP) grant no. 2021-68012-35917) led by the University of Wisconsin-Madison. It involves a diverse, transdisciplinary team of more than 50 researchers and stakeholders from 23 universities, two USDA-Agricultural Research Service centers, as well as 12 farmer organizations, industry groups, non-governmental organizations and government agencies. (see https://ag-resilience.org/). Key members of this group are planning to leverage this hub's datasets to submit a data-related proposal to USDA AFRI-NIFA this autumn. Changes/Problems:In July we requested and received a 1 yr NCE. However, we have encountered challenges in hiring post-doctoral expertise for this project and because of this challenge and others, we are currently planning a rebudget andwith arequest of a 2nd year NCE. This project was initiated in August 2019 and initial phases included systematic review / meta-analysis and acquisition of a large on farm data set (40 GBs) for an AI component demonstrating the value of on-farm data for fine-tuning soil management recommendations. We were able to start the initial phases of the Systematic Review as planned in our proposal timeline, however, hiring and on-boarding of graduate students and postdoctoral researchers necessary for an array of project tasks was disrupted and delayed by the COVID 19 pandemic. Campus COVID policies at Purdue and at our collaborating institution, Univ. of WI Madison, also greatly curtailed other on-campus project activities. While some of the literature searching and data extraction could be done remotely, targeted materials included retrieving gray literature and Extension reports, many of which have not been digitized or linked to search engines and data catalogs and exist only as paper copies in physical repositories that were closed during COVID. Further, a co-PI in information sciences responsible for designing and implementing search strategies for gray literature and data was stricken with long COVID and eventually resigned from the project and Purdue. Acquisition of the large, on-farm dataset was also delayed. Although the data exist as characterized in the proposal, the metadata were incomplete, and the magnitude of the deficit only became apparent when our private collaborator directed an initial transfer of data. Overall, the data (15.2 M lines of yield data; 2.8 M lines of soil test data; 1.8 M lines of landscape data) were neither "research ready" nor appropriately structured for exploration with AI and machine learning tools, and required an immense investment of human resources on the part of our private, unfunded partner to prepare the data for project use. However, once campus COVID policies were relaxed, we began making significant progress in objectives related to the systematic review including the access and use of gray, undigitized resources, and a new information specialist joined our team. We also now have a strategy to access and explore a portion of the on-farm data sufficient to achieve project objectives and, therefore, we request a NCE to complete this COVID-delayed work. What opportunities for training and professional development has the project provided? The main opportunity for professional development related to this project continues to be the graduate course "Systemic Reviews in Agriculture in the Environment," which is co-taught by our team. The course was reviewed by the Purdue administration and assigned a permanent number as a graduate student-specific, 3 credit hour course (AGRY 602). In Spring 2024, the course was offered to students at Purdue, the UW-Madison, Ohio State University and was expanded to include faculty and students at Iowa State University. Faculty from Mohammad VI University also attended, but did not enroll students. We again used piazza as the learning management system to deliver course materials (https://piazza.com). Twenty-five graduate students participated in the course and completed all assignments. Students were from several departments including soil science, agronomy, agricultural and biological engineering, agricultural science, education and communications, horticulture and landscape architecture, and botany and plant pathology. A current Ph.D. student of PD Brouder, Mandana Mirbakhsh, is completing the systematic review/meta-analysis she started in this class and will be submitting it for publication in the near future. CoPIs Cai and Murrell continue to adapt curricular materials for Mohammed VI Polytechnic in Morocco, and, more generally, for use by personnel within or collaborating with the African Plant Nutrition Institute. Building on the success of the Systematic Review/Meta-analysis course, PD Brouder, Co-PIs Volenec, Cai, Barford and Craig have organized a workshop focused on methods in systematic review and meta-analysis to be convened at the joint meeting of the American Society of Agronomy, Crop Science Society of America, and the Soil Science Society of America in Nov. 2024. We will include high-level overviews along with hands-on exercises in this 8 hr session. We will again use Piazza as a platform to deliver the workshop content and will provide access to workshop materials on the site for an extended period after the workshop concludes. We will report on the outcomes and impact of this workshop in future reports, but the workshop is oversubscribed (limited to 50 participants, currently a wait list of an additional 10 persons). How have the results been disseminated to communities of interest? ?CoPIs Brouder and Volenec continue to participate in monthly calls with the Fertilizer Recommendation Support Tool (FRST) project where the challenges of achieving FAIR data are an ongoing objective and continuing subject of discussion. Co-PI Volenec also participates in monthly calls with the FRST-sulfur subcommittee aimed at acquiring the data and meta-data needed to improve S recommendations in the US. Both Brouder and Volenec also participated in monthly meetings convened by the Foundation for Food and Agricultural Research (FFAR) aimed at developing minimum data sets, data standards, and corresponding meta-data to support/guide research on enhanced efficiency/novel fertilizers, including biological products. This consortium included approximately 20 individuals from around the US and several others from Europe and Canada, from both the public and private sectors. See "Target Audiences" for additional information. All project CoPIs continue to make presentations at their professional meetings and conferences on lessons learned through our project on the current status of data stewardship and systematic reviews and meta-analysis in agriculture. What do you plan to do during the next reporting period to accomplish the goals? We are continuing teaching the Systematic Review-Meta-analysis course at multiple institutions and have received a permanent number for the course (AGRY 602) at Purdue after review by the university and the College of Agriculture. We will conduct a workshop at our professional meetings (ASA/CSSA/SSSA) in November 2024. This will cover the basics of how to conduct a systematic review, extract data from published literature, and using this data, use meta-analytical statistics to analyze the results. We will launch two complementary data initiatives. The first will be the D2Ag effort with the US-AID Sustainable Intensification Innovation Lab awarded to Kansas State University. This effort will focus on data recovery, interpretation and curation in six select US-AID partner countries. The other effort will be initiated jointly with OCP Group, Mohammad VI University, the African Plant Nutrition Institute and Purdue using SR-MA to identify best management practices for fertilizer use aimed at optimizing agronomic and environmental performance. We will advance our work exploring how the large farmer data sets from Advanced Agrilytics can be used to inform P and K recommendations to optimize yield of corn and soybean, and to understand key covariates that interact with soil P and K levels and alter agronomic performance, and the extent to which on farm data can improve interpretation of soil test diagnostics.

Impacts
What was accomplished under these goals? ?T1(1): We compiled, annotated, and provided a large legacy dataset on alfalfa potassium and phosphate nutrition to the Fertilizer Recommendation Support Tool (FRST, https://frst.scinet.usda.gov/tool). This contribution nearly doubled the amount of data available for making P and K recommendations for alfalfa (30 of the current 68 trials currently in the tool database). These data also are unique as they provide soil fertility data in 5-cm increments to a 20 cm depth, monitor within-season changes in soil test P and K, and do so for 5 consecutive years. These data were also provided to a team developing a manuscript on quantifying potassium requirement and removal across crop species. T1(2)-T6: In our meta-analytical assessment of bias and and synthesis of data extracted from the literature to assess several measures of diagnostic accuracy (sensitivity, specificity, positive and negative predictive value, likelihood ratio, area under the sensitivity-specificity curve, diagnostic odds ratio, etc.),we used several benchmark soil testing datasets extracted from the peer and grey literature to demonstrate the overall diagnostic accuracy of routine, commercially available P and K soil tests. Historically, the soil tests for P and K have been considered "good" and "fair," respectively, but this rating was based on professional opinion and not linked to any formal statistical evaluation. Adiagnostic test should be considered good only if it correctly identifies both the responsive and unresponsive soils 70 to 80% of the time. Diagnostic accuracy should be considered sufficient or bad with correct identification of soil fertility conditions 60 to 70 and 50 to 60% of the time, respectively. Our analysis of data from field experiments in the 70's and 80's found P and K diagnostic accuracies of 59 and 81%, respectively, at the theoretically, evidence-based critical levels (levels above which no response to fertilizer is expected). A recent publication has suggested increasing these critical levelsfor the region of our study, and our preliminary analysis of more recent field studies suggests diagnostic accuracy is now increasedto 66% or sufficient for Pbut reduced to46% for K (i.e., of no diagnostic value) at these new critical levels. For both nutrients, these diagnostic accuracies associated with published critical levels indicate that large numbers of fields may be receiving unnecessary fertilizer applications. The paired nature of diagnostic test accuracy data (i.e., sensitivity and specificity to weigh accurate diagnosis of responsive and non-responsive soils against false positives and negatives) makes meta-analysis complicatedespecially when risks of economic loss to farmers can be considered in the context of environmental degradation as is the case for P. This work is ongoing and will inform the analysis of the on-farm data where economic loss to farmers and costs to society of excess nutrients can be assessed at scale. Building on ourwork in collaboration with the Science Assessment Team for Indiana's Nutrient Reduction Strategy where network meta-analysis was used to synthesize across N rate categories, we are now using a similar approach to analysis of data extracted from the peer literature for this FACT project. Here, datais beingsynthesized by the ranking categories used to indicate the expected magnitudes of response to nutrient additions. For example, an interpretation of a soil test value as Very Low, is accompanied by a recommendation for a higher rate of P or K fertilizer than a soil test value interpreted as Low. Current commercial lab and Land Grant University categories differ and will be compared, and the on-farm datasets will be used to assess the magnitudes of over- and under-applications at scale. Finally, PD Brouder was invited to contribute to the Regional Assessment for the upcoming decadienal FAO report on The Status of the World's soils. She used data discovered in our FACT project literature and data searches to draft an analysis of the current state of nutrient mismanagement in North America with a focus on P and N management. T7: Our on-going development of graduate teaching materials and delivery via graduate curricula and workshops is discussed under Professional Development Opportunities.

Publications

• Type: Journal Articles Status: Under Review Year Published: 2024 Citation: Lyons, S.E., D.B. Arnall, D. Ashford-Kornburger, S.M. Brouder, E. Christian, A. Dobermann, S.M. Haefele, J. Haegele, M.J. Helmers, V.L. Jin, A.J. Margenot, J.M. McGrath, K.T. Morgan, T.S. Murrell, D.L. Osmond, D.E. Pelster, N.A. Slaton, P.A. Vadas, R.T. Venterea, J.J. Volenec, and C. Wagner-Riddle. 2024. Field trial guidelines for evaluating enhanced efficiency plant nutrition products. Soil Sci. Soc. Amer. J.
• Type: Conference Papers and Presentations Status: Published Year Published: 2023 Citation: Brouder, S. M. (2023) The Current Status of Soil Nutrient Management and Mismanagement in North America [Abstract]. ASA, CSSA, SSSA International Annual Meeting, St. Louis, MO. https://scisoc.confex.com/scisoc/2023am/meetingapp.cgi/Paper/152542
• Type: Conference Papers and Presentations Status: Published Year Published: 2024 Citation: Brouder, S.M. 2024. Data Stewardship & FAIR Practices. From: Developing Common Evaluation Protocols for Enhanced Efficiency & Novel Fertilizers. Sponsored by the Foundation for Food and Agricultural Research. Washington DC. January 23-24.
• Type: Conference Papers and Presentations Status: Published Year Published: 2023 Citation: Ashworth, A.J.; Marshall, L.; Volenec, J. J.; Berti, M.; van Santen, E.; Williams, C.; Gopakumar, V.; Foster, J.; Picasso, V.; and Su, J. 2023. Forage Data Hub  A Platform for Sharing Valuable Datasets for Resilience. International Grassland Congress, Lexington KY. May 14.
• Type: Conference Papers and Presentations Status: Published Year Published: 2024 Citation: Brouder, S.M. 2024. Can you share (please)? Data Stewardship Panel, NCEAR013-SERA6-NECC1012 Joint Meeting, Minneapolis, MN, June 3.
• Type: Conference Papers and Presentations Status: Published Year Published: 2023 Citation: Mirbakhsh, M., S.M. Brouder, J.J. Volenec, B.A. Craig, and J.K. Yatcilla. 2023. Crop responses to free-air CO2 enrichment (FACE) and nitrogen fertilization: A systematic review and meta-analysis. ASA, CSSA, SSSA International Annual Meeting, St. Louis, MO. https://scisoc.confex.com/scisoc/2023am/meetingapp.cgi/Paper/149234
• Type: Conference Papers and Presentations Status: Published Year Published: 2023 Citation: Brouder, S. M., Mazer, K., Camberato, J. J., Craig, B. A., & Frankenberger, J. (2023) Network Meta-Analysis Quantifies Water Quality Benefits of 4R N Management [Abstract]. ASA, CSSA, SSSA International Annual Meeting, St. Louis, MO. https://scisoc.confex.com/scisoc/2023am/meetingapp.cgi/Paper/154349
• Type: Other Status: Published Year Published: 2024 Citation: Ashworth, Amanda J.; Marshall, L.; Volenec, J. J.; Berti, M.; van Santen, E.; Williams, C.; Gopakumar, V.; Foster, J.; Picasso, V.; and Su, J., "Forage Data Hub  A Platform for Sharing Valuable Datasets for Resilience" (2024). IGC Proceedings (1993-2023). 63. https://uknowledge.uky.edu/igc/XXV_IGC_2023/Sustainability/63
• Type: Conference Papers and Presentations Status: Published Year Published: 2024 Citation: Brouder, S.M. 2024. Does that BMP work on a real farm? Lessons in realism from 3 decades of monitoring nutrient fluxes. Larson-Allmaras Lecture, Univ. MN, Minneapolis, MN, April 26.

Progress 08/01/22 to 07/31/23

Outputs
Target Audience:A primary target audience remains soil fertility researchers, Extension specialists and emerging professionals (graduate students) interested in improving fertilizer recommendations for P and K. This group has been expanded and codified in 2021 through our recurring interactions with Fertilizer Recommendation Support Tool (FRST; https://frst.scinet.usda.gov/). The FRST collaboration involves 47 universities, 3 USDA agencies (ARS, FSA and NRCS), 4 NGOs (FFAR, TFI, TNC and Plant Nutrition Canada) and nearly 100 scientists and specialists from every US state. The FRST effort is connected with the Agricultural Collaborative Research Outcomes System (AgCROS) of the USDA-ARS that has the long-term goal is to curate all agricultural data. It also complements efforts at the Ag Data Commons of the National Agricultural Library of USDA where Director Wester has recruited several new positions focused on systematic reviews in agriculture. PI Brouder and co-PI Volenec met with Director Wester and his team data team at NAL during the most recent project year, to continue ongoing coordination and leveraging of NAL efforts with this NIFA-funded project and others described in this report. In 2023, we continued to expand the reach of our systematic review and meta-analysis curriculum by presenting on it at the ASA, CSSA, SSSA International Annual Meeting, Baltimore, MD in Nov. 22, which resulted in adding a third cohort of participants at The Ohio State University for spring semester 2023 and plans to add a 4th cohort from Iowa State in 2024. Data scientists and informaticians also remain important target audiences for this project. Our efforts are revealing challenges, some of them extraordinary, to agricultural data (e.g., large grey literature volume, non-standardized data workflows linked to disparate equipment suites with proprietary software), that may help this group improve workflows, meta-data standards, and curation processes. We continue to support professional societies and other organizations creating tools for and promoting broader adoption of data curation and open science. In his role as Co-PI Murrell extends our projects efforts to international colleagues through his work with OCP Africa and their Al Moutmir initiative https://www.almoutmir.ma/en/home-page) to create data quality standards for on-farm fertilizer product and rate studies conducted by fertilizer companies that are being used to develop local recommendations. Finally, in year 4 of our project, our team continued to work with various agencies and organizations interested in developing data policies for the projects they support including preparing written responses to FFAR's Request for Information on Data Solutions for Accelerating Climate Smart Agriculture Research and Action. We also expanded the scope of our project to support the needs of in-state conservation groups tasked with implementing Indiana's Nutrient Reduction Strategy. Changes/Problems:Given the disruptions in original workplans caused by COVID, in fall 2023 we will revise project plans and timelines in conjunction with the request for a no-cost extension. What opportunities for training and professional development has the project provided?The main opportunity for professional development related to this project continues to be the graduate course "Systemic Reviews in Agriculture in the Environment," which is co-taught by our team. As a result of presentations made at professional meetings in 2022, in Spring 2023, we expanded the 3-credit offering to a cohort at The Ohio State University and anticipate expanding to a fourth cohort at Iowa State in Spring 2024. In Spring 2023, we migrated the learning management from a box account to Piazza, a leading social learning platform in education (https://piazza.com). Twenty-five graduate students and post-doctoral students participated in the course and completed all assignments. A graduate teaching assistant from statistics assisted with the class and a statistic research assistant is currently helping develop data extraction exercises. A student from Spring 2022 published her class assignment as a chapter of her dissertation under the supervision of CoPI Ruark while another Spring 2022 participant has received a $15K dollar award plus any At Publication Charges from Purdue to complete and publish her class project on crop N requirements under ambient and elevated CO2 levels; she is advised by the project team. CoPI Murrell supervised two MSc students at Mohammed VI Polytechnic in Morocco in the application of systematic mapping techniques to improve nutrient management of olive for oil production and C sequestration. How have the results been disseminated to communities of interest?CoPIs Brouder and Volenec continue to participate in monthly calls with the Fertilizer Recommendation Support Tool (FRST) project where the challenges of achieving FAIR data are an ongoing objective and continuing subject of discussion. Co-PI Volenec regular participates in monthly calls focused on development of the Forage Data Hub mentioned above. Project PD Brouder has participated as a guest where her expertise was requested. Additionally, lessons learned are provided in response to ad hoc requests including the detailed written response to FFAR's Request for Information on Data Solutions for Accelerating Climate Smart Agriculture Research and Action. The package of curricular materials for the graduate-level class Systematic Reviews in Agriculture and the Environment (discussed under Opportunities for Professional Development) are available upon request. This includes videos of all class discussions. In 2023, all class material were made available to Purdue staff tasked with developing the Science Assessment to underpin Indiana's Nutrient Reduction Strategy and PI Brouder developed and delivered an application case study via a series of webinars. What do you plan to do during the next reporting period to accomplish the goals?Many activities will continue as originally planned albeit delayed during the first couple of years of the project by COVID restrictions. We will continue to extract data and run preliminary analyses on the project's systematic review questions. In fall 2022, we initiated conversations with National Agricultural Libraries regarding a framework and infrastructure to support SRs in agriculture, a goal aligned with NAL's strategic objectives. As feasible, we will expand our original objectives to support the development of Indiana's nutrient reduction strategy. The SR course will be taught again in Spring 2024 and it will be expanded to Iowa State University. Lead instructor at IA State is in the Agric. Biol Engineering knowledge domain, further broadening the impact of the course beyond crop and soil science, animal science, and environmental sciences originally envisioned. We will continue to enrich course materials with a focus on materials for class exercises in data extraction and the application of an array of meta-analytical techniques. We are exploring avenues to grow and socialize the course and its offerings via conferences and professional meetings as we view it as a foundational skill for agricultural research and Extension professionals.

Impacts
What was accomplished under these goals? T1: Prototype for ag data (data & metadata standards), surface legacy data & transform to FAIR formats, best practices for data recovery: We are moving project effort from legacy data recovery into an analysis of the limitations of the grey literature in comparison to peer-review limitations. A manuscript is being prepared describing recurring deficiencies and associated biases that may be introduced when including grey literature. Project personnel continue to lead, advise and/or participate in numerous, ongoing, and seemingly parallel and/or competing efforts to develop data and meta-data standards for agriculture. As examples, PI Brouder used project experiences to formulate a detailed response to the Foundation for Food and Agriculture's Request for Information "Data Solutions for Accelerating Climate-Smart Agriculture Research and Action" (prepared comments available on request). These comments specifically addressed gaps in and barriers for using existing ontologies, the need for additional meta-data standards, barriers to awareness and adoption by agricultural researchers, APIs, the status and utility of tools for automatic data annotation, the need for minimum datasets, and the challenges associated with legacy data. CoPI Murrell has contributed to CGIAR's Excellence in Agronomy Initiative to produce standard operating procedures to harmonize measurements across institutions and countries for major food crops. Numerous private entities are developing their own ontologies that they intend for universal use and many are largely unaware of each other's efforts; PI Brouder advises Agmatix in the development of their GUARDS (Growing Universal Agronomic Data Standard) protocol which "aims to translate the unique way each researcher preserves raw agronomic data into one common standardized language that can be understood by any researcher in the world" (https://www.agmatix.com/agronomic-modeling-technology/). Similarly, co-PI Volenec advises a USDA AFRI-NIFA Sustainable Ag Systems group as they develop an open-source data network focused on productivity of forages (https://data.nal.usda.gov/dataset/data-framework-develop-open-source-forage-data-network-improve-primary-productivity-and-enhance-system-resiliency). This includes data ontologies and dictionary development, meta-data development, and data identification and recruitment. This project currently includes over 40MB of data from 1956 to 2022. T2: Synthesize FAIR data with SR & MA: Although we will continue to assess / include legacy grey literature as we find it, we have shifted the effort from legacy data recovery to the analysis and assessment of value of the literature uncovered so far. Key ongoing efforts include the synthesis of the grey literature in parallel with the peer literature currently being extracted and synthesized. We continue to screen peer-review papers and extract data for our 2 main systematic review questions addressing the diagnostic accuracy of soil P and K tests in diagnosing deficiencies. The meta-analytical approach is following the STARD (Standards for Reporting Diagnostic Accuracy) statement (https://pubmed.ncbi.nlm.nih.gov/28137831/) developed for assessment of diagnostics in medicine in the assessment of bias and extracting data to assess several measures of accuracy (sensitivity, specificity, positive and negative predictive value, likelihood ratio, area under the sensitivity-specificity curve, diagnostic odds ratio, etc.), a novel approach in soil testing. Extraction of a database for an SR more narrowly focused on a question relevant only to K in the context of climate change (Does super-optimal soil K improve drought tolerance of major crops?), is complete and a data publication via the Purdue University Research Repository is in preparation. To accommodate, emerging state-level concerns and needs, we have expanded efforts in SR to improve management to nitrogen. Colleagues at Purdue have been tasked with developing a rigorous science assessment to underpin Indiana's Nutrient Reduction Strategy focused on reducing N and P pollution of state waters. As a demonstration of best practices, our team developed a search strategy for the question: How much improvement in nitrate pollution can be attributed to reducing fertilizer rates from farmer practice to an economic optimum rate. We subsequently extracted a regional dataset and used network meta-analysis to document a 25% reduction in nitrate concentration in drainflows. A manuscript summarizing this effort is in preparation. T3 and 4: Evaluating recommendations with on-farm data and applying machine learning to on-farm data. Although we have secured on-farm data from Advanced Agriclytics, use of the data to assess recommendation factors will begin in earnest once the results of the T2 diagnostic accuracy assessment are available. Additionally, we recently obtained the results from a large number of on-farm P and K "omission" plot studies conducted by colleagues in Ohio (2014 - 2018). A total of 118 P and 96 K trails were conducted for corn (57 P, 35 K), soybean (47 P, 47 K), and wheat (14 P, 14 K). In related work leveraging this project, CoPI Murrell at the African Plant Nutrition Institute has developed a new relationship with Statistics for Sustainable Development (Stats4SD, https://stats4sd.org/) to develop an open-source, research data management platform that accommodates structured and unstructured biophysical and socio-economic data. Work on profitability of current soil sampling strategies is getting underway. T7: Our related graduate education curricula is reported under "Opportunities for Training and Professional Development" (below).

Publications

• Type: Book Chapters Status: Published Year Published: 2023 Citation: Brouder, S.M., and J.J. Volenec. 2023. Nutrition of Plants in a Changing Climate In: Z. Rengel, I. Cakmak, P. White (eds.) Marschners Mineral Nutrition of Higher Plants, 4th Edition. Pp. 723-750. Academic Press/Elsevier, Amsterdam, Netherlands
• Type: Journal Articles Status: Published Year Published: 2023 Citation: Ashworth, A.J., L. Marshall, J.J. Volenec, M.D. Casler, M.T. Berti, E. van Santen, C.L. Williams, V. Gopakumar, J.L. Foster, T. Propst, V. Picasso, and J. Su. 2023. Framework to develop an open-source forage data network for improving primary productivity and enhancing system resiliency. Agron. J. DOI: 10.1002/agj2.21441.
• Type: Conference Papers and Presentations Status: Published Year Published: 2022 Citation: Brouder, S.M. 2022. Realizing global ecosystem and biodiversity benefits with local improvements in management of intensive agriculture: Strategies and challenges. In special session on Global-Local-Global Analysis of Land and Water Systems Sustainability. AGU Fall Meeting 2022. Chicago IL, 12  16 December. https://agu.confex.com/agu/fm22/webprogrampreliminary/Session161589.html
• Type: Conference Papers and Presentations Status: Published Year Published: 2022 Citation: Brouder, S. M., J.J. Volenec, and R. Owen. 2022. Using agricultural research to drive climate solutions. Special Symposium and Panel on Policy Strategies to Support Climate and Ecosystem Solutions in Agriculture. Nov. 9, 2022. ASA, CSSA, SSSA International Annual Meeting, Baltimore, MD. https://scisoc.confex.com/scisoc/2022am/meetingapp.cgi/Paper/147605.
• Type: Journal Articles Status: Published Year Published: 2023 Citation: Cisneros-Pineda, A., Dukes, J. S., Johnson, J., Brouder, S., Ramankutty, N., Corong, E., & Chaudhary, A. (2023). The missing markets link in global-to-local-to-global analyses of biodiversity and ecosystem services. Environmental Research Letters, 18(4), 041003.
• Type: Conference Papers and Presentations Status: Published Year Published: 2022 Citation: Brouder, S.M. 2022. What is Agronomys Role in Supporting Climate Science Research? Symposium on the future of Agronomy. ASA, CSSA, SSSA International Annual Meeting, Baltimore, MD. https://scisoc.confex.com/scisoc/2022am/meetingapp.cgi/Paper/141649.
• Type: Conference Papers and Presentations Status: Published Year Published: 2022 Citation: Brouder, S.M., C. Barford, C. Cai, B.A. Craig, K. Dunn, T.S. Murrell, M.D. Ruark, B.B. Sisolak, J.J. Volenec, and J.K. Yatcilla. 2022. A graduate course on systematic review and meta-analysis in agriculture and the environment. ASA, CSSA, SSSA International Annual Meeting, Baltimore, MD. https://scisoc.confex.com/scisoc/2022am/meetingapp.cgi/Paper/142577.
• Type: Conference Papers and Presentations Status: Published Year Published: 2022 Citation: Lyons, S., N.A. Slaton, S.M. Brouder, D.L. Osmond, J.T. Spargo. The importance of data sharing and publication. ASA, CSSA, SSSA International Annual Meeting, Baltimore, MD. https://scisoc.confex.com/scisoc/2022am/meetingapp.cgi/Paper/141543
• Type: Theses/Dissertations Status: Published Year Published: 2023 Citation: Ashmita Rawal, Department of Soil Science. Nitrogen update and nitrogen demand of potatoes on sandy soil. Chapter 2: Optimum nitrogen rates for yield and agronomic nitrogen use efficiencies for potato grown on sandy soils in the Midwestern agroecological region of the USA: a meta-analysis.

Progress 08/01/21 to 07/31/22

Outputs
Target Audience:A primary target audience remains soil fertility researchers, Extension specialists and emerging professionals (graduate students) interested in improving fertilizer recommendations for P and K. This group has been expanded and codified in 2021 through our recurring interactions with Fertilizer Recommendation Support Tool (FRST) group that now encompasses nearly 100 scientists and specialists from every US state. The FRST effort is connected with the Agricultural Collaborative Research Outcomes System (AgCROS) of the USDA-ARS that has the long-term goal is to curate all agricultural data. It also complements efforts at the Ag Data Commons of the National Agricultural Library of USDA where Director Wester has recruited several new positions focused on systematic reviews in agriculture. PI Brouder and co-PI Volenec have met with Director Wester during the most recent project year, and are scheduled to do so again in the near future. The goals of the meeting include coordination and leverage of NAL efforts with this NIFA-funded project and others described in this report. Data scientists and informaticians also remain important target audiences for this project. Our efforts are revealing challenges, some of them extraordinary, to agricultural data (e.g., large grey literature volume), that may help this group improve workflows, meta-data standards, and curation processes. Improved tools are critical to broader adoption of data curation and open science. Through our professional societies we have supported efforts to deliver evidence-based knowledge to the 14,000 Certified Crop Advisors and other practitioners. Several groups have interest in data underpinning specific issues like C sequestration in the context of climate and the C markets. This has led to development to "Decode 6" (https://decode6.org/) by the education department of the American Society of Agronomy to deploy both summary knowledge on these key topics along with the datasets used in summary development. Finally, in year 3 of our project, our target audiences expanded to include policy makers. In June, PI Brouder was invited to testify at a House Agriculture hearing on "The Role of Climate Research in Supporting Agricultural Resiliency." As requested, both her written and oral testimony focused on the data infrastructure needs to support agriculture's ability to mitigate and adapt to climate change. Written and oral testimony can be accessed at the US House of Representatives Committee Repository (https://docs.house.gov/Committee/Calendar/ByEvent.aspx?EventID=114888). Changes/Problems:Given the challenges experienced to date from COVID, we anticipate needing a no-cost extension to achieve project goals. Early in year 4, we will convene all CoPIs for a review/audit of progress to date and to revise timelines and any work elements that COVID has disrupted or altered conditions such that work elements are no longer doable given the post-COVID landscape. Some modifications have been already been communicated in previous Progress Reports; any modifications will be communicated to NIFA on or before next year's Progress Report deadline. What opportunities for training and professional development has the project provided?Spring Semester 2022, Co-PI's team-taught a new graduate course on Systematic Reviews in Agriculture and the Environment. The course was listed in both Purdue (AGRY 598) and Univ. of Wisc. (Soils 875) course catalogs and participants included 8 fully enrolled PhD students, 1 PhD visiting scholar (auditing but fully participatory) and 4 post-doctoral researchers from Univ. of IL and Guelph (2 completed all class exercises). The mode was a hybrid of face-to-face and Zoom with designated classrooms at each university. All Zoom recordings were posted to a class Box account along with all other course materials; instructors-of-record at each institution were responsible for moving content into their respective learning management systems and recording grades. However, all participants were invited to the Box folder and most chose to access resources from these folders. The class met twice per week (1 hr 15 min sessions) with 40 minutes of each class devoted to a lecture/presentation format followed by collaborative work activities. Content and weekly assignments spanned all stages of a systematic review from literature scoping and question development to quantitative analyses with visualization and manuscript and data publication strategies. Learning outcomes included: 1) assessing the need for a systematic review (SR) and review typologies, 2) explaining the critical steps of a SR, 3) searching the peer and grey literature, 4) planning for and implementation of eligibility screening and data extraction and 5) synthesis and presentation of results. The course emphasized "open" resources including manuals, primary literature and tools for reference management, data extraction and statistical synthesis. Pre- and post-assessments of 11 learning outcomes were completed by 8 participants and showed dramatic increases in knowledge levels. Key points of student self-assessment were the following: 1) they thought they knew the literature around their questions of interest but learned they did not know how to develop/deploy an unbiased search strategy with relevant databases, 2) data extraction was much more challenging then any had anticipated as most peer-reviewed papers did not present the all the data required by meta analyses, and 3) they were largely unaware of the wealth of human and electronic resources available to assist them in rigorously synthesizing existing science. While students could work together on a question as the foundation for class exercises most chose to work on their own project directly aligned with their current PhD or post-doctoral research. During the semester students performed the steps of an SR but not at the level of a peer-review publication. However, the majority planned to continue working on their SR with the goal a preparing a manuscript, some directly for use as a component of their dissertation. We plan to teach the course again Spring Semester 2023 with an additional location at The Ohio State University. In addition to the SR course, CoPI Murrell worked with colleagues at the African Plant Nutrition Institute to accept two MSc interns from University of Mohammed the 6th Polytechnic (UM6P) to begin developing systematic maps on topics related to mineral nutrition of olive. These interns completed and defended their work during presentations/oral exams conducted by UM6P faculty. CoPI Murrell is currently developing educational modules in Google Classroom for delivery of SR course content in Africa. CoPI Cai completed the Institute for Museum and Library Services' 3-d Evidence Synthesis Institute (https://www.lib.umn.edu/about/evidence-synthesis-institute#about) aimed at library staff supporting evidence syntheses in topics outside of the health sciences. This workshop covers: overview of systematic reviews and similar methodologies, guidelines and standards, search strategy development, software tools, quality assessment, and systematic review services. How have the results been disseminated to communities of interest?In the 3rd year of this project with COVID 19 still restricting some face-to-face scientific and Extension meetings, we reached our targeted audiences in the following ways: 1. Co-PIs Brouder and Volenec have participated in monthly webinars with the Fertilizer Recommendation Support Tool (FRST) project. This group has more than doubled in size in the last 12 month and now includes 100 scientists and specialists from virtually all US states and Porto Rico. While the original goal was to develop a comprehensive database of P and K correlation-calibration results that can be accessed through an online tool for use in research and fertilizer recommendation development, this effort has expanded to include discussions on soil sampling procedures, liming, and strategies for estimating maximum yield. The PIs surface ideas and have also contributed several large, multi-year soil P/K - yield datasets (DOI:10.4231/PPKB-VK18; . https://doi.org/10.4231/C1W0-8F75; https://doi.org/10.4231/WWSK-RA65). This FRST collaboration resulted in a peer-review journal article "Minimum dataset and metadata guidelines for soil-test correlation and calibration research" (Slaton et al. 2022. Minimum dataset and metadata guidelines for soil-test correlation and calibration research. Soil Sci. Soc. Amer. J. 86:19-33. https://doi.org/10.1002/saj2.20338). The Soil Science Society of America is the professional home for 6,000+ members and 800+ certified professionals dedicated to advancing the field of soil science. This includes two divisions whose members align closely with this project (Nutrient Management and Soil and Plant Analysis; Soil Fertility and Plant Nutrition). 2. Sustained interest in the open access publication of "Improving Potassium Recommendations for Agricultural Crops" (2021, 455 pp.,https://link.springer.com/book/10.1007/978-3-030-59197-7). This compendium was released in early 2021. The book puts forward a significantly expanded characterization of the K cycle in agricultural soils, critically examines previously used approaches to K recommendations and explores their relevance to further improvements and management objectives beyond crop productivity such as sustainability. FACT project CoPIs are co-authors on five chapters and CoPI Murrell was the lead editor. The book's target audiences are graduate students, educators, industry scientists, data scientists and advanced practicing agronomists. To date, the publisher has recorded >64,000 downloads of the book and over 30 citation in a little over 1 yr. The lead chapter "The Potassium Cycle and Its Relationship to Recommendation Development" co-authored by Co-PIs Brouder, Volenec and Murrell has had nearly 6000 downloads; this chapter emphasizesthe importance of open access and data infrastructure to current and future efforts to improve potassium recommendations. 3. Miscellaneous Activities: We participated in numerous courses, workshops and webinars in order to reach additional audiences and expand impact of this project. Examples include: 1. The course on systematic reviews in agriculture and the environment reached graduate students and their mentors at both Purdue and the Univ. of WI-Madison. However, students and post-docs located at the Univ. if IL, Univ. of Guelph, and materials are currently being adapted for use at Mohammad VI University in Morocco. Presentation on this class at a statistics conference in May (see Products) generated interest at other institutions; currently plans are to expand this course to include students and faculty at the Ohio State University and Iowa State University in Spring 2023. 2. PI Brouder participates in the Science Assessment for Indiana Nutrient Reduction Strategy. This effort will use data synthesis to develop programming aimed to reduce nutrient losses to surface waters. Brouder has made the Box folder with all SR class resources available to the Nutrient Reduction Strategy lead for use on this project. 3. PI Brouder participated in a US-AID Feed the Future seminar on Global Food Security Research Strategy: New Evidence and Opportunities that explored the evidence underpinning food insecurity. 4. co-PIs Brouder and Volenec participated in a year-long seminar series on open data/open science sponsored by the Council for Scientific Society Presidents (CSSP) and included several other professional societies including the National Academy of Sciences, American Geophysical Union, ASA, SSSA, and CSSA. 5. co-PIs Brouder and Volenec participated in an NSF-sponsored, multidisciplinary workshop focused on development of data communities. This workshop brought data communities at various stages of development together for sharing successes, failures/challenges and resulted in an extensive report with numerous recommendations to improve data curation and re-use (Ruediger et al. 2022. Leveraging Data Communities to Advance Open Science: Findings from an Incubation Workshop Series. Ithaka S+R Research Report). What do you plan to do during the next reporting period to accomplish the goals? Continue work as planned on data infrastructure development and on the two systematic review questions related to P and K management and recommendations. This includes continued screening and data extraction from the peer-review literature identified with our search strategy (977 unique journal articles). CoPI Barford plans to focus any additional searching of grey literature on a more confined geographic area with a strong history of annual reporting on Extension and applied research (e.g. Ohio). A strategy for using the grey literature to assess publication bias will be developed and an assessment initiated. Publications of the grey literature database and a manuscript for a peer-review journal will be prepared; the focus of this manuscript will be potential, pitfalls and limitations of assessing and using the grey literature in management recommendation development. CoPIs Yatcilla and Cai will continue their assessment of the quality of systematic reviews in agriculture and the development of the associated manuscript. A collaboration with National Agricultural Libraries for development of a framework and infrastructure to foster Systematic Reviews in Agriculture will be initiated Fall 2022. Discussions may be extended to encompass entities such as the USDA Climate Hubs and Professional Societies (e.g. American Society of Agronomy, Crop Science Society of American and Soil Science Society of America) and their platforms. On-farm data and machine learning: This activity, intended to begin in Project Year 2, will begin in the coming year. CoPIs Volenec and Berg with continue to collaborate to transfer Advanced Agrilytic's data and CoPIs Craig and Zhu will develop the initial machine learning strategies using the subsets of data that have already been transferred to Purdue and currently reside in deep storage. A survey of farmer perspectives and costs of soil sampling strategies will be initiated by CoPI Thompson to frame the current return-on-investment farmers realize given the precision and accuracy of existing recommendations. The SR class will be taught again and expanded to The Ohio State University. Discussions are underway to also include a class cohort at Iowa State University. We will explore opportunities to enhance the existing array of class resources (reading materials, featured tools, videos and exercises/assignments) into an open access curriculum that can be used by any instructor seeking to independently develop/deliver a class on system review methodology. Resources currently sought by many include a Data Extraction Manual; we will explore developing such a manual in conjunction with teaching the class.

Impacts
What was accomplished under these goals? T1 (1): Our data-related work with the Fertilizer Recommendation Support Tool (FRST) group (https://soiltestfrst.org/) continues. PI Brouder and co-PI Volenec collaborated with others from FRST to codify minimum dataset standards for soil test P and K sampling and analysis. This effort culminated in a publication (see Slaton et al. in the products listing) whose "Research Interest Score" is higher than 92% of all research items published in 2021 according to Research Gate (https://www.researchgate.net/publication/354684748_Minimum_dataset_and_metadata_guidelines_for_soil-test_correlation_and_calibration_research/stats). Participation in this effort has expanded to 99 scientists from nearly every state in the US. Research questions being addressed have also expanded to include key questions about quantifying relative yield, soil sampling depth impacts on soil test P and K results, and soil test lime requirements. PI Brouder and co-PI Volenec have contributed data and ideas to these emerging topics. Additionally, project PIs continue to advocate for data infrastructure and policy to facilitate reuse of agricultural research data and integration of research with on-farm/private data for improved recommendations with site- and soil-specificity. See Progress Report Sections on Target Audiences and How Results Have Been Disseminated for additional details. As interest in agricultural science synthesis grows, CoPIs Cai and Yatcilla are also exploring the need for guidance to improve quality and ensure outcomes meet integrity standards sought in rigorous and comprehensive syntheses. They have been coding a sample of agriculture-related systematic reviews and meta-analyses to get an overview of the authors' reporting practices. They have found that less than a quarter of the articles (~23%) referred to following specific guidelines for their SR/MA methodology, and even fewer articles (~5%) referred to a written protocol. While systematic reviews were based on literature searches in multiple databases, many of the meta-analyses were based on convenience samples of data sets, rather than data extracted from studies discovered through comprehensive searches. Only around 10% of articles included full search strategies, and only 1% referred to working with an information professional. Based on these preliminary results, it is clear there is a lot of room for agriculture systematic reviews or meta-analyses to improve in terms of transparency and reproducibility, two hallmarks of evidence synthesis projects. A manuscript is under development as the foundation for development of practice standards for the agricultural sciences. T1(2): Screening of and data extraction from peer-review literature for our two systematic review questions (detailed in last year's Progress Report) are ongoing. Initial screening (Screen 1) of literature on soybean has been completed in Rayyan. Papers passing this screen one of abstracts, titles and keywords are now being screened for data that can be extracted and used in a meta-analysis. Extraction from graphs is being performed with WebPlotDigitizer. When variance measures (e.g. standard error of the mean difference) for targeted effects are not directly reported, they are being back-calculated from the reported statistics using best practices. Additionally, CoPI Barford extracted another 1000+ lines of data from the grey literature bringing the number of to over 2000. Preliminary assessment of the grey literature observations found poor relationships between soil test P and K and yields and little apparent responses of yields to fertilizer additions; an evaluation of publication bias in the peer-review literature is strongly indicated. Our SR protocol is largely complete and we are considering options for its publication. T4: We have acquired four large sets of on-farm data from Advanced Agri-Lytics. The data includes site descriptors (hydrology, elevation, slope, acreage, soil wetness, ...), standard soil test information, management details (irrigation cover crop use, tillage, row spacing,...), along with yield. The data has a high level of spatial resolution; the smaller of these datasets has ~10,000 rows of data describing a 124 acre field in a single year. We are developing workflows for managing the data as some of these datasets will characterize much larger fields in multiple years. Additional datasets are forthcoming from our cooperators in the coming year. T6: A new MSc student has been identified and will join the project (Fall 2022) to explore profitability of current soil sampling strategies for P and K recommendations. T7: The main activity for this project's educational and professional development activities was the delivery of a 3-credit graduate course in "Systematic Reviews in Agriculture in the Environment." Details of on this course including its modality, learning objectives, and outcomes are reported under "Opportunities for Training and Professional Development" (below).

Publications

• Type: Conference Papers and Presentations Status: Published Year Published: 2022 Citation: Craig, B.A., S.M. Brouder, J.J. Volenec, C. Cai, J.K. Yatcilla, T.S. Murrell, M. Ruark, C. Barford, B.B. Sisolak, and K. Dunn. 2022. A seminar course on systematic reviews in agriculture and the environment. Annual meeting of NCCC170: Research Advances in Agricultural Statistics. Utah State University, Logan UT. May 16, 2022.
• Type: Journal Articles Status: Published Year Published: 2022 Citation: Slaton, N.A., S.E. Lyons, D.L. Osmond, S.M. Brouder, S. Culman, L.C. Gatiboni, J. Hoben, P.J.A. Kleinmann, J.M. McGrath, R. Miller, A. Pearce, A. Shober, J.T. Spargo, and J.J. Volenec. 2022. Minimum dataset and metadata guidelines for soil-test correlation and calibration research. Soil Sci. Soc. Amer. J. 86:19-33. https://doi.org/10.1002/saj2.20338.
• Type: Conference Papers and Presentations Status: Published Year Published: 2021 Citation: Murrell, T.S., S.M. Brouder, and J.J. Volenec. 2021. Big data for smart agriculture: What are we hoping for? In: Agri-innovations to Combat Food and Nutritional Challenges. Proc. 5th International Agronomy Congress. Nov. 23 to 27. Indian Society of Agronomy, PJTSUA, Hyderbad India.
• Type: Conference Papers and Presentations Status: Published Year Published: 2021 Citation: Volenec, J.J. 2021. A Confluence of Opportunities: Agronomic Partnerships for a Better Tomorrow. In: Agri-innovations to Combat Food and Nutritional Challenges. Proc. 5th International Agronomy Congress. Nov. 23 to 27. Indian Society of Agronomy, PJTSUA, Hyderbad India.
• Type: Conference Papers and Presentations Status: Published Year Published: 2022 Citation: Brouder, S.M., and J.J. Volenec. 2022. Co-Production of Bioenergy, Food, Feed (BFF) and Ecosystem Services (ES) in the Midwest Cornbelt: Focus on cover cropping. Biomass Research and Development Operations Committee (cross-agency in DC; USDA-NIFA, EPA, DOE, OSTP, USDA-ARS, &&). March 24, 2022.
• Type: Conference Papers and Presentations Status: Other Year Published: 2021 Citation: Slaton, N.A., S.E. Lyons, D.L. Osmond, S.M. Brouder, S. Culman, L. Gatiboni, J. Hoben, P.J.A. Kleinman, J. McGrath, R.O. Miller, A. Pearce, A.L. Shober, J.T. Spargo, and J.J. Volenec. 2021. Development of minimum dataset for soil test correlation and calibration. ASA, CSSA, SSSA International Annual Meeting, Salt Lake City, UT. https://scisoc.confex.com/scisoc/2021am/meetingapp.cgi/Paper/133770
• Type: Conference Papers and Presentations Status: Published Year Published: 2021 Citation: Brouder, S.M., and J.J. Volenec. 2021. Publishing raw data: how it helps you, others and science and how to do it. ASA, CSSA, SSSA International Annual Meeting, Salt Lake City, UT. https://scisoc.confex.com/scisoc/2021am/meetingapp.cgi/Paper/137821.
• Type: Other Status: Published Year Published: 2022 Citation: Brouder. S.M. 2022, Invited Testimony. House Ag committee Hearing on The Role of Climate Research in Supporting Agricultural Resiliency, June 15, 2022. Washington, D.C (https://docs.house.gov/Committee/Calendar/ByEvent.aspx?EventID=114888)

Progress 08/01/20 to 07/31/21

Outputs
Target Audience:The target audience still includes other soil fertility researchers interested in improving recommendations for P and K. This group has been expanded and codified in 2021 through our recurring interactions with Fertilizer Recommendation Support Tool group that now encompasses over 30 states. This effort is connected with the Agricultural Collaborative Research Outcomes System (AgCROS) of the USDA-ARS, where the long-term goal is to curate all agricultural data. Data scientists and informaticians also remain important target audiences for this project. Our efforts are revealing challenges, some of them extraordinary to agricultural data (e.g., large grey literature volume), that may help this group improve workflows, meta-data standards, and curation processes. Improved tools are critical to broader adoption of data curation and open science. Through our professional societies we have expanded discussion about data to Certified Crop Advisors and other practitioners. Several groups have interest in data underpinning specific issues like C sequestration in the context of climate and the C markets. Plans are being developed with the education department of the American Society of Agronomy to deploy both summary knowledge on these key topics along with the datasets used in summary development. Changes/Problems:Our Advanced Agrilytics collaborator has encountered challenges in reconciling data layers we want to use for the machine learning research (e.g., soil test data, yield monitor data, wetness indicators, slope/aspect,...). We plan to have an in-person meeting with him in Dec. 2021 to discuss ways to advance the development of these data sets. Once data are available, they will be shared with Dr. Jerry Zhu at the UW-Madison and the machine learning/artificial intelligence research will commence. As discussed under project accomplishments (Task 4), we plan to publish a synopsis of the key challenges we encountered preparing these data for use in research. Data extraction is slow, especially from the grey literature where the research methods are often poorly described. In addition, titles and abstracts (when present) of the grey literature often do not contain accurate wording that reflects the report's content. As a result, reviewing these articles and extracting their data that are underrepresented in most syntheses is progressing at a slower rate than desired. We are continuing to sort through grey literature, and expect to include problems associated with this effort in a "lessons learned" manuscript to assist others planning to extract data from grey literature. While we have made progress in developing the graduate course on systematic reviews and knowledge synthesis, logistic and institutional issues have arisen. Semester and class start timings differ at the UW-Madison and Purdue, and we have reconciled these as best we can by working around institution-specific calendars and class scheduling. While the course is intended for graduate students, we were expecting interest from advanced undergraduates who are interested in graduate study. Their participation in the course may be limited because of the misalignment of semesters and class timing. However, we plan to capture lectures digitally and will make these available to interested students. The COVID 19 pandemic has limited in-person interactions and slowed overall progress on the plan of work. Part of this is indirect, as most other PI responsibilities, and, most notably teaching, take far more time with COVID 19 restrictions, and redundancies like both in-person (masked) and remote instruction modalities. The surge in the delta variant of the virus returned Purdue and our collaborating institutions to high levels of restriction in August 2021 after only one month of lower restrictions. Additionally, key team members have had to be replaced on the project for health and/or early retirement reasons related directly and indirectly to COVID; the process of updating our team expertise has also slowed progress. We anticipate improved interactions/progress will be possible in early 2022 as more people are vaccinated. What opportunities for training and professional development has the project provided?a. Ten team members participated in a 2-hour training to introduce them to Rayyan (https://www.rayyan.ai/about-us), an open access tool for screening articles for inclusion, exclusion in a systematic review. The training was conducted by CoPI Yatcilla, who will replace CoPI Walker on the project. b. Co-PIs Brouder and Volenec participate in a year-long "Data Sharing Seminar Series for Societies" sponsored by the Council for Scientific Society Presidents (CSSP). Topics covered many issues relevant to the FACT project including: Data Sharing and Citation: How Societies can Make a Difference; Public Access to Research Data: Guidance, Culture, and Practice; Open Science Incentives for Researchers: The Role of Societies and Organizations; Accelerating Science: Data Discovery and Interoperability; Fostering Data Reuse: Increasing Impact and Ease in Sharing and Reusing Data; Democratization of Data among others. Details, including the presentations, can be found here: https://wesharedata.org/ c. PI Brouder participated in a workshop "Leveraging Data Communities to Advance Open Science". This is a new NSF-funded collaboration between Ithaka S+R and the Data Curation Network. This effort seeks to support the development of infrastructures for data sharing within data communities in collaboration with the Data Curation Network. Co-PIs Brouder and Volenec will continue to participate in future workshops and webinars sponsored by this group. d. CoPI Murrell completed the following courses and trainings: (i) UCLA Computer Science X 450.1 (Introduction to Data Science, 3.6 CEUs), (ii) UCLA Computer Science X 450.3 (Hadoop and Managing Big Data, 3.6 CEUs) and (iii) CGIAR Big Data AgroFIMS. Murrell also train colleagues in evidence-based approaches, open science, and reference class forecasting. How have the results been disseminated to communities of interest?In the 2nd year of this project with COVID 19 still restricting face-to-face scientific and Extension meetings, we reached our targeted audiences in the following ways: Open access publication of "Improving Potassium Recommendations for Agricultural Crops" (2021, 455 pp., https://link.springer.com/book/10.1007/978-3-030-59197-7). This compendium was released in early 2021. The book puts forward a significantly expanded characterization of the K cycle in agricultural soils, critically examines previously used approaches to K recommendations and explores their relevance to further improvements and management objectives beyond crop productivity such as sustainability. FACT project CoPIs are co-authors on five chapters and CoPI Murrell was the lead editor. The book's target audiences are graduate students, educators, industry scientists, data scientists and advanced practicing agronomists. To date, the publisher has recorded 37,000 downloads of the book with 3300 downloads of the lead chapter "The Potassium Cycle and Its Relationship to Recommendation Development" co-authored by CoPIs Brouder, Volenec and Murrell. FRST collaboration on peer-review journal article "Minimum dataset and metadata guidelines for soil-test correlation and calibration research" (in press). Convened by the FRST project in which CoPIs Brouder and Volenec participate, a group of 14 Extension specialists met bimonthly to develop a consensus statement on standardizing research approaches and reporting requirements for research focused future soil testing research. This allowed us to share our FACT experiences to date in trying to extract data for Systematic Reviews and the limitations we are encountering with existing literature including experimental design issues and result reporting conventions that pose challenges in formal syntheses and meta-analysis. Miscellaneous Activities: We participated in numerous workshops and webinars with varying compositions of our targeted audiences where we described the goals and lessons-learned to date from this project. Co-PIs Brouder and Volenec participated in an on-line webinar "Advances in Agricultural Evidence" April 13, 2021 sponsored by the Collaborative for Environmental Evidence. This session aimed to bring together some of those working in evidence-based agriculture in order to highlight challenges, showcase work that is taking place and to stimulate discussion and future collaboration. Co-PIs Brouder and Volenec have participated in monthly webinars with the Fertilizer Recommendation Support Tool (FRST) project. This group is developing a comprehensive database of P and K correlation-calibration results that can be accessed through an online tool for use in research and fertilizer recommendation development. The PIs surfaced ideas and suggestions to better focus efforts to improve protocols for P and K soil sampling depth research that is being initiated. They also provided large, multi-year soil P/K - yield datasets that are being used to advance efforts to reconcile procedures used to determine maximum yield is nutrition calibration/correlation research. What do you plan to do during the next reporting period to accomplish the goals?Planned activities for the coming year include the following: Systematic Review: We plant to complete and publish or register our SR protocol following the guidelines provided by the Collaboration for Environmental Evidence and PRISM (Preferred Reporting Items for Systematic Reviews; doi:10.7326/M18-0850) in keeping with best practices for SRs as developed by other research communities. We will continue to extract data for our SR questions that are currently underway and expect to initiate meta analyses on questions where articles are sufficient. We anticipate finishing the alfalfa scoping review and a case study focused on our existing recommendations where we directly compare the recommendation with outcomes reported in the peer and grey literature. Data Standards: CoPI Murrell will take the lead in continuing to collaborate with technical staff working on the CGIAR Big Data Platform, specifically the AgroFIMs and FAIRscribe projects with the goal of leveraging these standards into ones that serve a broader community of interest around crop nutrition and soil fertility research. Additionally, he will work directly with a Moroccan Extension organization to conform their on-farm fertilizer demonstration studies in preparation for study syntheses and develop and test a smallholder engagement approach using in-field demonstrations to collect standardized quantitative and qualitative data across several African countries. On-Farm data: Machine Learning will be initiated on some portion of on-farm and or farmer Cooperative data. Co-PIs Volenec and Berg will continue to work with the Advanced Agrilytic data, releasing portions of it to the rest of the team as segments of the data are converted to structures necessary for our project. Additionally, CoPI Ruark will complete an assessment of the "big" data available through major farmer cooperatives in Wisconsin including characterizing the nature, extent and completeness of these data. Graduate Level Class: The Systematic review class (described under T7 Accomplishments) will be team taught in Spring Semester 2021.

Impacts
What was accomplished under these goals? T1: Key activities focused on understanding and adapting emerging data standards to the specific purposes of this project and achieving consensus across a network of soil fertility and plant nutrition researchers on minimum data sets, and meta data standards for future soil fertility research. Specifically, ICASA 2.0 and Agronomy Ontologies: In project year 1, some combination of the ICASA 2.0 https://dssat.net/data/standards_v2/) and AgrO (https://bigdata.cgiar.org/resources/agronomy-ontology/) emerged as the developing data standards most suited to our research objectives. Year 2 activities focused on adopting these standards for use in describing previous maize experiments conducted by the African Plant Nutrition Institute and partners in several countries and developing protocols for use in new experiments. Likewise, the standards were considered and adapted where possible for use in the data extraction templates for Systematic Reviews (T2). We are continuing to annotate and aggregate terms for data not yet encompassed by existing standards to inform the continued growth of these existing standards/ontologies. Participation in the FRST data standards project: CoPIs Brouder and Volenec collaborated with 12 CoPIs from the FRST project to develop and publish (in press) guidelines for minimum data collection and metadata standards for research intended to be synthesized into evidenced-based P and K recommendations based on soil testing. T2: Following best practices, the project team developed a draft protocol for addressing two question variants structured on the PICO (population, intervention, comparator, outcome) model. The questions are: (1) To what extent does nutrient amendment guided by soil P or K testing increase crop (corn, soybean, wheat, alfalfa) yield? And/or 2) Does a P or K soil test diagnose yield-liming P or K deficiencies in a crop (corn, soybean, wheat, alfalfa)? This protocol (and eventual systematic review) follows the guidelines of the Collaboration for Environmental Evidence (https://www.environmentalevidence.org/guidelines/section-9) and is in keeping with the ROSES reporting standard (https://www.roses-reporting.com/). The initial search focused on the primary peer-review literature and searched five bibliographic databases. Search returns were de-duplicated in Endnote; a total of 977 unique journal articles were identified along with 222 dissertations and theses. An initial inclusion/exclusion screening (Screen 1) is currently ongoing using Rayyan open access software ((https://www.rayyan.ai/about-us ) and is based on examination of titles, key words, and the abstracts. The Screen 1 criteria appear robust and sufficient for our questions. A data extraction template has been developed and template fields are being iteratively modified and/or expanded as dictated by the content of the included articles. The volume of literature returned varies across targeted crop species the only 72 references for alfalfa versus 200 or more for soybean, wheat, and maize. The alfalfa references are predominantly for outside of upper US Midwest and have a heavy emphasis on irrigated systems. Given this limited number of references, we have parsed out alfalfa for a formal "Scoping Review" anticipating data gaps will be significant with respect to our SR questions. This effort will determine several things including: 1) the overall effectiveness of the search strategy used to retrieve relevant literature; 2) the adequacy of and gaps in the retrieved literature for conducting a quantitative analysis; and 3) number of references that are not identified when only screening titles abstracts and key words. A separate review protocol is under development for this scoping activity. Lastly, CoPI Barford has extracted approximately 1000 observations of yield and related P/K and management data from the grey literature and has developed a case study to test the reproducibility of existing recommendations by comparing the theoretically recommended rate from Extension guides with observations in both the grey and peer literature, and CoPI Murrell has developed draft guidance in R markdown on using the "litsearchr" package in the R metaverse to develop search terms for citation databases. T4: We are in the process of acquiring on-farm data from our collaborators at Advanced Agrilytics. Several challenges have emerged as their staff began to merge data layers with different temporal and spatial sampling intensities. While this has delayed the Machine Learning research, it serves as an excellent case study for others interested in using on-farm data for research purposes. We will be documenting the nature of these issues and the changes in workflows required to resolve them. These findings will be included in future discussions of "lessons learned" and published so others can more easily acquire and use farm data in research and plan for the challenges we have encountered when preparing to integrate research and farmer-data. T7: This year's activities focused on developing materials for a new course, "Systematic Reviews in Agriculture and the Environment" that has been listed in the University of Wisconsin and Purdue University course catalogues for Spring Semester 2022. The intended audience is beginning graduate students, especially Ph.D. candidates reviewing the literature as they initiate their dissertation research. It is a 3 credit hour class co-taught primarily by FACT team members at both institutions including Brouder, Volenec, Craig, Ruark, Cai, Murrell, and others. This course will be in-person for students at the respective campuses (students will enroll at their respective institutions) with synchronous virtual connectivity between classrooms. The course goal is for students to understand the distinguishing features of review typologies (i.e. narrative, mapping/scoping, quantitative with meta-analysis, etc.) and to be able to apply their understanding to selecting and conducting the systematic reviews that are appropriate to their own research questions. Topics to be covered include: writing the review protocol; literature search strategies; selecting studies and collecting data; analysis of risk and bias; statistical analysis; interpreting the results and reporting the findings. The course will expose students to existing tools to facilitate all steps of the review process from searching to visualization and will emphasize use of open access resources. We will record all lectures so that they can be used by African Plant Nutrition Institute colleagues in their education program. Additionally, CoPI Murrell piloted development of interactive educational materials using Google Classroom, Formative, EdPuzzle, and Kahoot and tested the materials via online training of M.S. students in the Fertilizer Science and Technology program at the University of Mohammed VI Polytechnique, Morocco.

Publications

• Type: Journal Articles Status: Published Year Published: 2021 Citation: Slaton, N.A., S.E. Lyons, D.L. Osmond, S.M. Brouder, S. Culman, L.C. Gatiboni, J. Hoben, P.J.A. Kleinmann, J.M. McGrath, R. Miller, A. Pearce, A. Shober, J.T. Spargo, and J.J. Volenec. 2021. Minimum dataset and metadata guidelines for soil-test correlation and calibration research. Soil Sci. Soc. Amer. J. https://doi.org/10.1002/saj2.20338.
• Type: Journal Articles Status: Published Year Published: 2021 Citation: Berg, W.K., S.M. Brouder, S.M. Cunningham, and J.J Volenec. 2021. Potassium and phosphorus fertilizer impacts on alfalfa taproot carbon and nitrogen reserve accumulation and use during fall acclimation and initial growth in spring. Front. Plant Sci. 12:1698. https://doi.org/10.3389/fpls.2021.715936
• Type: Journal Articles Status: Awaiting Publication Year Published: 2021 Citation: Welikhe, P., S.M. Brouder, J.J Volenec, M. Gitau, and R.F. Turco. 2021. Dynamics of dissolved reactive phosphorus loss from phosphorus source and sink soils in tile-drained systems. J Soil Water Conserv. (in press).
• Type: Journal Articles Status: Published Year Published: 2021 Citation: Welikhe, P., S.M. Brouder, J.J Volenec, M. Gitau, and R.F. Turco. 2021. Using artificial neural networks to improve phosphorus indices. J. Soil Water Conserv. pp. 1-14. doi:10.2489/jswc.2021.00153.
• Type: Book Chapters Status: Published Year Published: 2020 Citation: Brouder, S.M. and J.J. Volenec. 2020. Mineral nutrient acquisition and metabolism. Chapter 5. In: M. Collins, C.J. Nelson. K.J. Moore and D.D. Redfearn (eds.). Forages-The Science of Grassland Agriculture. Volume II. 7th Edition. John Wiley & Sons, Inc., New York, NY. doi.org/10.1002/9781119436669.ch5.
• Type: Book Chapters Status: Awaiting Publication Year Published: 2022 Citation: Brouder, S.M., and J.J. Volenec. Nutrition of Plants in a Changing Climate In: Z. Rengel, I. Cakmak, P. White (eds.) Marschners Mineral Nutrition of Higher Plants, 4th Edition. Academic Press/Elsevier, Amsterdam, Netherlands.
• Type: Journal Articles Status: Published Year Published: 2021 Citation: Moore, Eli K; Kriesberg, Adam; Schroeder, Steven; Geil, Kerrie; Haugen, Inga; Barford, Carol; Johns, Erica M; Arthur, Dan; Sheffield, Megan; Ritchie, Stephanie M; Jackson, Carolyn; Parr, Cynthia: Agricultural data management and sharing: Best practices and case study. Agronomy Journal. 2021; 1- 11. https://doi.org/10.1002/agj2.20639

Progress 08/01/19 to 07/31/20

Outputs
Target Audience:We reached our various target audiences in the following ways: Other soil fertility researchers, extension scientists and graduate students. Discussions were had among this group to build collaboration and understand the availability of datasets. Collectively, project Co-PIs completed five book chapters in the new book "Improving Potassium Recommendations for Agricultural Crops," which reviews concepts traditionally used to make K recommendations and proposes a framework for making future improvements. It is scheduled to be published at the beginning of next year (Jan 2021). An additional Co-PI book chapter discusses the need to better link K management in the field to human nutritional needs for K. It makes the case that K recommendations need to consider more than crop production, and reinforces and expands upon certain concepts presented in "Improving Potassium Recommendations for Agricultural Crops. These materials are open source and intended for use by practicing professionals and in graduate education and will be used in developing T7 educational materials. Data Scientist and Informaticians: Discussions have started with collaborating datascientiststo help them understand the nature ofthe problem we are trying to solve, including the chemistry of P and K soil tests, data required to build response curves, and the philosophical approach developed to create guidelines. Also, as described under T7, project Co-PIs devoted over 80 hours of face-to-face videoconferencing for meetings and working session in efforts to coordinate activities among similar, emerging data initiatives. Activities focused on convening researchers with data scientists to discuss data workflow tools, strategies for legacy data recovery, and reconciliation of emerging controlled vocabularies and ontologies for agriculture. Changes/Problems:An over-arching issue for this reporting period was the COVID19 pandemic. This effectively closed Purdue University and the University of Wisconsin-Madison to most in-person work, including instruction and normal graduate student and post-doc mentoring. Transfer of these activities to the virtual environment on very short notice (most Co-PIs were engaged in the teaching mission when COVID hit) consumed a tremendous amount of time and were identfied as a very high priority by our respective administrations. As a result, there was a significant pause in research as we pivoted almost exclusively to fulfill the teaching mission. Commitments to transforming existing courses, an effort that is still ongoing, preculded offering any new courses including those associated with this project. Opportunities to advance specific tasks were suspended by others external to this grant because of COVID19. For example, Co-PIs Brouder and Volenec were invited to participate in the annual meeting of the Collaboration for Environmental Evidence (CEE; https://www.environmental evidence.org) meeting scheduled for early June in Ottawa Canada. This group provides information and tools to assist in development of systematic reviews including registration of reviews that are in-process. Our plan is to use these tools provided by CEE as we develop our own systematic review for this project. In addition, Co-PI Brouder had been asked to give an invited presentation at this meeting describing this FACT project. Unfortunately, COVID19 canceled this meeting. Nevertheless, CEE has created an on-line seminar series as a vehicle for distributing the knowledge and engaging stakeholders in this "new normal". PIs Brouder, Volenec, and Murrell have participated in several of these seminars to-date and will continue to participate in future seminars in the future. Similarly, the in-person Tri-societies (American Soc. Agron.; Crop Sci. Soc. Amer.; Soil Sci. Soc. Amer.) meeting scheduled for mid-November in Pheonix AZ was moved to a virtual platform. Planned face-to-face interactions to advanced project goals, including presentations by team members, have been shifted to virtual sessions and separate, stand-alone Zoom meetings. Adjustments for the coming reporting period include enhanced virtual interaction using Zoom and similar platforms. Currently we are having bi-weekly team meetings using Zoom to discuss issues and advance project goals. Between these formal meetings there is daily use of email by the team to fine-tune timelines and priorities day-to-day work activities as necessary. At this writing, COVID19 infection rates are increasing across the US. We anticipate continued virtual platform use well into 2021. However, the virual platform skills we have acquired the past 8 months of the pandemic will be beneficial going forward so we can maintain progress towards project goals. What opportunities for training and professional development has the project provided?In order to learn more about workflows for the entire process of legacy data recovery, PIs Brouder and Volenec worked with the data curation specialist (Stanislav Pejša) from Purdue Libraries who specializes in data recovery, curation, publication, and conversion to FAIR formats. Because the data for this project are in various states of recovery, cleaning and organization, Brouder and Volenec used a complete, but un-curated alfalfa variety trial data set to serve as a case study. These alfalfa data share many of the same attributes we anticipate eventually accessing for this project, including being large (>200,000 rows of observations), collected over a 15 yr time frame with multiple observations within each year, and sourced from numerous independent investigators located across the Midwest US and beyond. Lessons learned included triaging data into manageable subsets, creation and inclusion of meta-data with all subsets to ensure inter-operability, file structure and syntax, quality control and assurance, publication workflows, and issues pertaining to copyright/ownership when principle investigators can no longer be consulted; all things relevant to water and crop production data in the NutriNet project. The resultant alfalfa datasets were published with DOIs in the Purdue University Research Repository (see citations below). These publications can be accessed at: doi:10.4231/PHKH-4334, doi:10.4231/FMY9-6966, doi:10.4231/02PR-9H36, doi:10.4231/KK4K-QD96, doi:10.4231/Y31N-5R10. In order to develop skills for T1 and T4, CoPI Murrell is currently pursuing the UCLA Data Science Certificate and has completed UCLA Extension course 450.00 - Data Science Fundamentals (3.3 CEUs, 33 Contact Hours) to fulfill one of the requirements for certificate. How have the results been disseminated to communities of interest? Nothing Reported What do you plan to do during the next reporting period to accomplish the goals?We will continue to focus on and advance progress in Task 1 - 4 in reporting period two, which is in keeping with our original timeline. In T1 we will continue to participated in the ontology community by using terms and definitions already developed, revising terms, and developing/proposing new terms. We will continue our facilitation and leverage activities by coordinating with NAL, the CGIAR, the FRST group, ICL and other emerging data initiatives. In T2, we will focus on extracting legacy data, establishing and refining best policies and procedures and creating the database and establishing our protocol for our systematic review. We will continue our participation in CEE seminars focused on systematic reviews. In T3, we will have acquired the Advanced Agrilytics data through the 2020 cropping season and this will be foundational dataset for both T3 and T4. We will analyze data to determine factors other than soil tests that may be predictive of yield responses both in traditional supervised and machine learning/unsupervised approaches to identification of important variables and/or their proxies. We will explore publishing Python and R functions on GitHub that automate aspects of data engineering, data analysis, and data visualization. Project CoPIs will continue their professional development activities including formal participation in workshops and certificate programs to acquire new skill needed for our project. In T7, we will conduct a scan of the new offerings to make any adjustments to our original plan. We will focus on developing course content for the course on systematic reviews in agronomic sciences and the graduate participatory seminar. We are currently exploring Insight Scope (http://insightscope.ca/) as a model and tool to engage students in Systematic Reviews in the new era of routinized virtual learning.

Impacts
What was accomplished under these goals? T1: We conducted and extensive environmental scan for related, ongoing activities to uncover data workflows and data and metadata schema to leverage and/or coordinate with as a key goal of this task is to advance interoperability at local to global scales. Several large efforts were identified as critical to advancing FAIR data practices and workflows in agricultural research. Project Co-PIs devoted over 80 hours of video conferencing to three of the larger efforts which include: CGIAR Big Data Platform (https://bigdata.cgiar.org/) and USDA National Agricultural Library (NAL): The CGIAR has a major initiative and platform dedicated to democratizing access to agricultural data and has a prototype Agronomy Ontology (AgrO) that provides terms from the agronomy knowledge domain that are semantically organized and can facilitate the collection, storage and use of agronomic data, enabling easy interpretation and reuse of the data by humans and machines alike. While the crop science ontology is fairly well developed, the AgroO leadership are seeking assistance from Project PIs with soils-related terminology. Project PIs have attended e-workshops with Platform leaders and have convened several video conferences aimed at coordinating efforts among our projects, and Co-PI Murrell is currently contributing to the following ontologies that participate in the OBO Foundry: AGRO, ENVO, ChEBI. The Fertilizer Recommendation Support Tool (FRST, https://doi.org/10.1002/ael2.20008): This newly initiated 30+ state effort envisions a soil-test and crop-response-to-fertilization searchable database and web-based tool that enables consistent, transparent, and science-based decision support for nutrient recommendations (including P and K) for major crops across the USA. PIs Brouder and Volenec participate in this effort, including biweekly calls focused on development of minimum data sets, meta-data standards, and ontologies. They have also contributed data to this effort. Ultimately these data will reside in the USDA's AgCROS data repository (https://agcros-usdaars.opendata.arcgis.com/). Israel Chemical Ltd (ICL):PIs Brouder and Volenec hosted scientists from ICL in late 2019 to coordinate data set development, ontologies, and workflows that will be used to develop a set of agronomic e-tools, including several focused on soil fertility and plant nutrition. The platform called Growers Tech (https://www.growerstech.com/) currently is creating a data fusion center and is facilitating and standardizing agronomic data. Their long-term goal is to aggregate data and meta-data that enable use of artificial intelligence to generate novel insights for growers and others in agriculture. PIs Brouder and Volenec continue to advise the group on data and meta-data standards, have provided data, and serve in a coordinating capacity among the groups (CGIAR, FRST, ICL) working on data in the agronomy/plant nutrition space. T2: Key activies were: Establishing a search strategy:Project CoPIs conducted a preliminary review of current University of Wisconsin-Madison, Purdue and other regional fertilizer guidelines including assessing he history and evolution of the guidelines, and any available data used to develop the guidelines. Project CoPIs refined several key questions for the SR process and initiated Boolean-based keyword scans of the peer-reviewed literature in order refine search strategies. This effort has been iterative, with several sets of representative peer-reviewed papers (~300 per set) used as key word sources. Initial scans produced approximately 100 unique keywords (or keyword combinations) that can be used to identify relevant peer-reviewed papers. We are preparing to run final literature scans and begin extraction of data for the systematic review. Legacy Data from the Gray Literature:Hundreds of Experiment Station reports and similar reports from regional research committee (e.g., NC and NCR committees) exist that contain data and results never published in peer-reviewed literature. Project PI Barford has obtained print copies of a large subset of these reports and is systematically reviewing and cataloging studies whose findings pertain to P and K nutrition of maize, soybean, alfalfa, and wheat. To date, Barford has scanned about 300 volumes, starting in the 1960's, and has already identified 1000 data records from the Univ. of Illinois reports alone. Other volumes awaiting review include those from major soil fertility programs located in Land Grant Universities from IN, IA, KS, MI, MN, MO, SD, WI, OH, and NE including those residing in the North Central Fertilizer Fertility Conference (https://northcentralfertility.com/proceedings/?&action=years). This effort is being coordinated with the gray literature recovery effort led by Dr. S. Lyons of the FRST project described above, including sharing of data dictionaries, templates, meta-data standards and keywords. Since many of these studies report negative or neutral findings (e.g., no response to fertilizer application) that is often judged as unpublishable in the peer-review process, this effort is critical to assessing bias in the peer reviewed literature that may influence soil fertility recommendations. We are coordinating this effort with NAL and professor Katrina Fenlon at the University of Maryland, College of Information Studies who is formalizing processes for assessing legacy data values (e.g. Data Rescue Processing Guide, https://drum.lib.umd.edu/handle/1903/26473). Collaboration for Environmental Evidence (CEE's). CoPI Walker conducted an environmental scan for systematic review protocols and confirmed CEE protocol as the best foundation for our project. We initiated conversations with CEE regarding strategies for adopting CEE SR tools for broad availability to US agricultural researches but summer workshop plans were interrupted by COVID (see Changes Problems). T4: We are waiting until data from the 2020 growing season are available before acquiring the complete multi-year dataset from our corporate partner Advanced Agrilytics for the artificial intelligence (AI)/machine learning (ML). In the meantime, they have provided a subset of data so we can familiarize ourselves with the data structure and verify/clarify the metadata needed for the AI/ML analysis. Preliminary review of this dataset indicates that is spatially intense with 9983 rows of observations made within a 125 acre (51 ha) field. Each observation includes details on soil chemical (soil test pH, P, K, Ca, Mg, S, organic matter, exchanges site saturation by Ca, H, and Mg,), physical (slope, wetness, elevation), and crop management (irrigation, tillage, crop, cover crop use, row spacing, yield) attributes. We are working with IT staff at the UW-Madison, Purdue, and Advance Agrilytics on best management practices for transfer and security of this very large database (>50 TB) that includes data from approximately 400,000 acres collected for as many as six growing seasons. T7:Since writing this proposal, there has been tremendous progress in creating data-related curricula and programs although our own plans for new curriculum development in this proposal were disrupted. At Purdue course/curricula/programs have emerged in the last 12 months. Likewise, Univ. of WI has a new Information School (iSchool, https://ischool.wisc.edu/programs/) that includes an M.S. in Information Science These emerging efforts will provide important context and support as we develop the graduate data certificate program originally outlined in our proposal and include a Data-Driven Agriculture Minor and a spatial data science graduate certificate in agriculture targeted to learners with a B.A. CoPIs Walker and Chao initiated new courses in Critical Data Studiesand Data Management At The Bench, respectively.

Publications

• Type: Book Chapters Status: Awaiting Publication Year Published: 2021 Citation: Brouder SM, Volenec JJ, Murrell TS (2021) The potassium cycle and its relationship to recommendation development. In Murrell et al. (eds) Improving potassium recommendations for agricultural crops, https://doi.org/10.1007/978-3-030-59197-7 (in press)
• Type: Book Chapters Status: Awaiting Publication Year Published: 2021 Citation: Goulding K, Murrell TS, Mikkelsen RL, Rosolem C, Johnston J, Wang H, Alfaro MA (2021) Outputs: potassium losses from agricultural systems. In Murrell et al. (eds) Improving potassium recommendations for agricultural crops, https://doi.org/10.1007/978-3-030-59197-7 (in press)
• Type: Book Chapters Status: Awaiting Publication Year Published: 2021 Citation: Majumdar K, Murrell TS, Dutta S, Mikkelsen RL, Kotnis A, Zingore S, Sulewski G, Oberth�r T (2021) Illustrating a disjoint in the soil-plant-human health nexus with potassium. Lal R (ed) The soil-human health nexus. Advances in Soil Science. https://www.routledge.com/The-Soil-Human-Health-Nexus/Lal/p/book/9780367422134 (in press)
• Type: Other Status: Published Year Published: 2020 Citation: Berg, W. K.; Lissbrant, S.; Volenec, J. J.; Brouder, S. M.; Joern, B. C.; Johnson, K. D.; Cunningham, S. M. (2020), "Phosphorus and Potassium Influence on Alfalfa Nutrition." https://doi.org/10.4231/PPKB-VK18.
• Type: Other Status: Published Year Published: 2019 Citation: Volenec, J.J., and S.M. Brouder. 2019. Maize Response to P and K in 2006. Experiment 1 at Throckmorton Purdue Agricultural Center: Yield, Soil and Tissue P and K, and Seed and Stover Composition. Purdue University Research Repository. https://doi.org/10.4231/C1W0-8F75.
• Type: Other Status: Published Year Published: 2019 Citation: Volenec, J.J., and S.M. Brouder. 2019. Maize Response to P and K in 2006. Experiment 2 at Throckmorton Purdue Agricultural Center: Yield, Soil and Tissue P and K, and Stover and Grain Composition. Purdue University Research Repository. https://doi.org/10.4231/WWSK-RA65.
• Type: Book Chapters Status: Awaiting Publication Year Published: 2021 Citation: Murrell TS, Pitchay D (2021) Evaluating plant potassium status. In Murrell et al. (eds) Improving potassium recommendations for agricultural crops, https://doi.org/10.1007/978-3-030-59197-7 (in press)
• Type: Book Chapters Status: Awaiting Publication Year Published: 2021 Citation: Majumdar K, Norton RM, Murrell TS, Garc�a F, Zongore S, Prochnow LI, Pampolino M, Boulal H, Dutta S, Francisco E, Tan MS, He P, Singh VK, Oberth�r T (2021) Assessing potassium mass balances in different countries and scales. In Murrell et al. (eds) Improving potassium recommendations for agricultural crops, https://doi.org/10.1007/978-3-030-59197-7 (in press)
• Type: Book Chapters Status: Awaiting Publication Year Published: 2021 Citation: Volenec JJ, Brouder SM, Murrell TS (2021) Broadening the objectives of future potassium recommendations. In Murrell et al. (eds) Improving potassium recommendations for agricultural crops, https://doi.org/10.1007/978-3-030-59197-7 (in press)