Land use change ranges in each panel are in acres per thousand county acres. The white colored counties represent missing yields for at least one crop in all years.

Historical (1981-2005) vs. Projected (2031-’55) Yields. Each year’s crop yields are calculated as an average of all counties in North and South Dakota. Hashed representations of projected yields are from RCP 4.5 emissions scenario from seven GCMs, namely CESM (Community Earth System Model), CNRM (Center National de Recherches Météorologiques (France)), GFDL (Geophysical Fluid Dynamics Laboratory), GISS (Goddard Institute of Space Studies), HADGEM (Hadley Global Environment Model), IPSL (Institut Pierre-Simon Laplace (France)) and MIROC (Model for Interdisciplinary Research on Climate). Median projection in a given year is calculated by taking the median yield value of the yield projections from each of seven climate model outputs in each county and then taking the average across counties. We restrict spring wheat and alfalfa yield forecasts to zero for years in which these are projected to be negative values.

These are the soil quality data for each county (listed by fips code) for each scenario. This dataset is associated with the following publication: Zhang, X., T. Lark, C. Clark, Y. Yuan, and S. LeDuc. Grassland-to-cropland conversion increased soil, nutrient, and carbon losses in the US Midwest between 2008 and 2016. Environmental Research Letters. IOP Publishing LIMITED, Bristol, UK, 16: 1-14, (2021).

These are the soil quality data for each county (listed by fips code) for each scenario. This dataset is associated with the following publication: Zhang, X., T. Lark, C. Clark, Y. Yuan, and S. LeDuc. Grassland-to-cropland conversion increased soil, nutrient, and carbon losses in the US Midwest between 2008 and 2016. Environmental Research Letters. IOP Publishing LIMITED, Bristol, UK, 16: 1-14, (2021).

The impact of climate change on land conversion was assessed by projecting the land-use model under GCM hindcast and forecast climatic conditions. For each GCM, we projected future row crop proportions under 40 years of average hindcast conditions (i.e., assuming historical climate prevails into the future), and under 20 year of average hindcast conditions followed by 20 years of average forecast conditions. We used twenty years of forecast conditions to avoid assuming that climate changes projected for mid-century would have occurred immediately. The 20-20 assumption approximates a linear transition from the historic to future climate and is not limiting since the probabilities will converge given sufficient years under a given climate regime. Projected climate change impacts differed in direction and spatial pattern between GCMs (Fig. S3). Projected changes in the proportion of the landscape planted in row crops were generally small (mean change = 0.01), but included increases of over 0.25 of the landscape and declines of over 0.3 of the landscape (Fig. S3). Conversion probabilities responded to precipitation as expected, with an increase in precipitation associated with higher probabilities that grass converts to crops and lower probabilities that crops convert to grass.

This data set contains a single data file (.csv format) that provides gridded values of net primary productivity (NPP) for cropland in eight counties in the central United States for the year 1992 and estimates of interannual cropland NPP in Iowa for years from 1982 through 1996. The data file also includes climate, soil texture, and land cover data for each 0.5 degree grid cell. The magnitude and interannual variation in NPP was estimated using crop area and yield data from the U.S. Department of Agriculture, National Agricultural Statistics Service (NASS). The major harvested commodities were corn, soybean, sorghum, sunflower, oats, barley, wheat, and hay. Total NPP estimates include both above- and below-ground components. County-level NPP in 1992 ranged from 195 to 760 gC/m2/year. The area of highest NPP, ranging from 650 to 760 gC/m2/year, was found in a band extending across Iowa, through northern Illinois, Indiana, and southwestern Ohio. Areas of moderate NPP, from 550 to 650 gC/m2/year, occurred mostly in Michigan and Wisconsin, while large areas of low NPP, from 200 to 550 gC/m2/year, occurred in North Dakota, southern Illinois, and Minnesota. The area of highest production was also the area with the largest proportion of land sown with corn and soybean. NPP for counties in Iowa varied among years (1982-1996) by a factor of 2, with the lowest NPP in 1983 (which had an unusually wet spring), in 1988 (which was a drought year), and in 1993 (which experienced floods). Revision Notes: The documentation for this data set has been modified, and the data files have been reformatted. The data files have been checked for accuracy and the contents are identical to those originally published in 2001.

This dataset includes herbaria specimen records that report collection location and date (from 1895 – 2013) for grasses from the Southwest Environmental Information Network (http://swbiodiversity.org/seinet, Accessed 3 March 2014; 79% of records) and Global Biodiversity Information Facility (http://www.gbif.org, [accessed 3 March 2014]; 21% of records) for 12 states in the western United States. Associated climate data include monthly mean temperature, maximum temperature, minimum temperature, and precipitation from the herbaria record locations from climate rasters provided by the PRISM Climate Group (http://www.prism.oregonstate.edu, [accessed 1 May 2014]). We derived climate variables expected to be biologically meaningful for plant performance based on annual trends, seasonality, and extreme conditions (Bioclim, http://www.worldclim.org/bioclim, [accessed 14 May 2014]) using the PRISM monthly temperature and precipitation values.

This metadata record describes a computed ratio of county acres of farmland reported in agricultural conservation programs to the total acres of farmland reported for the county. Acres in conservation programs are for the 2012 time period, and reported by survey from the U.S. Department of Agriculture, Farm Field Survey. Acres of total farmland were used from the U.S. Department of Agriculture, 2012 Census of Agriculture. The ratio is intended to provide an indication of the intensity of agricultural management practices.