3C dataverse: Community capitals, cover crops, & conservation agriculture in the U.S. corn-soybean belt

What?

A dataset containing 313 total variables from 33 secondary sources. There are 261 unique variables, and 52 variables that have the same measurement but are reported for a different year; e.g. average farm size in 2017 (CapitalID: N27a) and 2022 (N27b). Variables were grouped by the community capital framework's seven capitals—Natural (96 total variables), Cultural (38), Human (39), Social (40), Political (18), Financial (67), & Built (15)—and temporally and thematically ordered. The geographic boundary is NOAA NCEI's corn and soybean belt (figure below), which stretches across 18 states and includes N=860 counties/observations. Cover crop data for the 80 Crop Reporting Districts in the boundary are also included for 2015-2021.

Why?

Comprehensively assessing how community capital clustered variables, for both farmers and nonfarmers, impact conservation practices (and perennial groundcover) over time helps to examine county-level farm conservation agriculture practices in the context of community development. We contribute to the robust U.S. cover crop literature a better understanding of how overarching cultural, social, and human factors influence conservation agriculture practices to encourage better farm management practices. Analyses of this Dataverse will be presented as recomendations for farmers, nonfarmers, ag-adjacent stakeholders, and community leaders.

How?

Variables used in this dataset range 20 years, from 2004-2023, though primary analyses focus on data collected between 2017-2024, primarily 2017 and 2022 (NASS Ag Census years). First, JAM-K requested, accessed, and downloaded data, most of which was already publically available. Next, JAM-K cleaned the data and aggregated into one dataset, and made it publically available on Google Drive and Zenodo. 

What is 'new' or corrected in version 2? 

Edited/amended: Carroll, KY is now spelled correctly (two 'l's, not one); variable names, full and abbreviated, were updated to include the data year; Pike County's (IL) FIPS has been corrected from its wrong 17153 (same as Pulaski County) to 17149 (correct fips), and all Pike County (IL) data has been correctly amended; Farming dependent (ERS) updated for all variables; Data for built capital variables irrCorn17, irrSoy17, irrHcrp17, tractor17, and combine17 were incorrect for v.1, but were corrected for v.2; Several variable labels aggregated by Wisconsin University's Population Health Institute's County Health Rankings and Roadmaps were corrected to have the data's original source and years included, rather than citing CHR&R as the source (except for CHR&R's originally-produced values such as quartiles or rank scores); variables were reorganized by hypothesized community capital clusters (Natural -> Built), and temporally within each cluster. 

Added: 55 variables, mostly from the 2022 Ag Census.

Omitted: Four variables deemed irrelevant to the study; V1 codebook's "years internally available" column.

CRediT: conceptualization, CBF, JAM-K; methodology, JAM-K; data aggregation and curation, JAM-K; formal analysis, JAM-K; visualization, JAM-K; supervision, CBF; funding acquisition, CBF; project administration, CBF; resources, CBF, JAM-K

Acknowledgements: This research was funded by the Agriculture and Food Research Initiative Competitive Grant No. 2021-68012-35923 from the United States Department of Agriculture National Institute for Food and Agriculture. Any opinions, findings, conclusions, or recommendations expressed in this presentation are those of the authors and do not necessarily reflect the view of the U.S. Department of Agriculture. Much thanks to Corteva for granting data access of OpTIS 2.0 (2005-2019), and Austin Landini for STATA code and visualization assistance.