Agricultural data management and sharing: Best practices and case study

Abstract

Agricultural data are crucial to many aspects of production, commerce, and research involved in feeding the global community. However, in most agricultural research disciplines standard best practices for data management and publication do not exist. Here we propose a set of best practices in the areas of peer review, minimal dataset development, data repositories, citizen science initiatives, and support for best data management. We illustrate some of these best practices with a case study in dairy agroecosystems research. While many common, and increasingly disparate data management and publication practices are entrenched in agricultural disciplines, opportunities are readily available for promoting and adopting best practices that better enable and enhance data-intensive agricultural research and production.

The problem with open geospatial data for on‐farm research

“…Curation complements the idea of a “minimally acceptable dataset” proposed by Moore et al. (2021). Streamlined datasets have increased potential for reuse (Charvat et al., 2014) and probability for sharing (Sayogo & Pardo, 2013).…”

“…Poorquality metadata presents challenges for reuse and, accordingly, compliance with FAIR principles (Volk et al, 2014). Ideally, a well-planned data collection should include data management and metadata creation from the onset; this could be better facilitated by more exposure of students (Volk et al, 2014) and faculty (Moore et al, 2021) to good data management practices. Upskilling of researchers addresses two noted issues related to sharing -lack of skill in annotating data for release and lack of time to do so effectively (Sayogo & Pardo, 2013;Tenopir et al, 2011Tenopir et al, , 2020Williams et al, 2019).…”

“…This includes data that are actively collected by the collaborator, such as financial records, but also that which is passively collected by farm machinery. Moore et al (2021) presented a picture of farmers who give their data without reservation, but that model of open sharing is not evident in practice (Turland & Slade, 2020). While many would be comfortable sharing data with researchers, the unspecified third parties granted access through open data reduce willingness precipitously even in countries where farms are obliged to provide data to agencies (Jakku et al, 2019;Wiseman et al, 2019).…”

“…In this model, a researcher might only need to secure permissions from a few collaborators on a limited number of farms and/or from selected model fields. Curation complements the idea of a "minimally acceptable dataset" proposed by Moore et al (2021). Streamlined datasets have increased potential for reuse (Charvat et al, 2014) and probability for sharing (Sayogo & Pardo, 2013).…”

“…Researchers might expect increased citations, among other benefits, from sharing their data (Colavizza et al., 2020; Piwowar et al., 2007; Thelwall & Kousha, 2017). Nevertheless, many scientists remain hesitant to share their data (Dorta‐González et al., 2021; Janssen et al., 2012; Moore et al., 2021; Piwowar & Chapman, 2008; Savage & Vickers, 2009; Thelwall et al., 2020; Williams et al., 2019), despite asserting that sharing is a necessity of science (Tenopir et al., 2020). Additional documented challenges to this open science system have been discussed in other works (e.g., Grant & Hrynaszkiewicz, 2018; Keßler & McKenzie, 2018; Schmidt et al., 2016; Sturges et al., 2015; Williams et al., 2019).…”