Enabling Open-source Data Networks in Public Agricultural Research

Abstract

The next generation of agricultural problem solving will require big science and linkages forged across data sets and disciplines. Currently, a lack of data sharing and data accessibility is a major barrier for making better decisions in agriculture. Business cases for data-sharing infrastructure include that pooling datasets and computational power efficiently extends sparse data resources, facilitates new discovery, derives better answers and decision making, lowers the barrier of entry, and ensures scientific reproducibility so that U.S. production agriculture can compete sustainably. Immediate imperatives for facilitating data sharing to fully realize open access to public agricultural research are the following: (1) development and implementation of best practices for data—workflows and standards—in all future federally funded projects; (2) incentives and mechanisms for making available data not represented in the peer-review literature (grey and dark data); (3) coordination among existing and emerging data initiatives, networks, and repositories; and (4) dedicated and sustainable infrastructure—hardware, software, and human resources—to curate, preserve, and add value to data beyond the primary use for which they were collected. Agriculture’s pathway forward requires dedicated partnering among domain researchers, data scientists, science administrators and agencies, professional societies, and private publishing entities. To simultaneously achieve sustained and equitable data access, the authors suggest the most promise lies with a novel business model in which funding agencies pay directly for stewardship in proportion to grant volume. Further, they propose four major institutional strategies to advance data-driven research in agriculture: bridging gaps, reorienting institutions, leveraging assets, and connecting feedbacks. Teams must bridge expertise gaps through meaningful collaborations between agricultural researchers and data scientists. Institutions will need to reorient to prioritize team science and data sharing over smaller scale, individual efforts and to infuse an understanding of data sciences into curricula and learning outcomes. Initiatives to leverage assets should focus on surfacing grey/dark data not represented by peer-review publication, including high-value legacy datasets for which time and cost prohibit replication. Finally, for research data to achieve and maintain public value, it must connect feedbacks to ensure data are useful and useable for informing the end-user “apps” designed to enhance and secure our current food supply and address environmental and social challenges.