Who: USDA ARS and NDSU range and wildlife researchers, graduate students, and undergraduate techniciansWhat: Structural characteristics and community composition collected from southwestern North Dakota rangelands from 2017 through 2020Where: Hettinger Research Extension Center in Hettinger, North Dakota USA6, 65 ha patch-burn grazing pastures were the primary data collection locationsWhy: These two files come from a patch-burn grazing study in southwestern North Dakota that were comparing an iteration of patch-burn grazing with cattle to a version of patch-burn grazing with sheep for the grazing component. Feel free to contact me at jonathan.spiess@usda.gov or jwspiess@gmail.com.How: We used 0.5m x 0.5m quadrats to measure vegetation structure characteristics and community composition along 100m transects in patches (subsections) of larger pastures or management units. We measured 1 quadrat spaced every 10 m starting at 0 on both sides of the transect for 22 total quadrats per transect in patch-burn grazing pastures. Transects were distributed amongst patches of each pasture and management unit.Data were analyzed using a combination of mixed-effect models and ordinations to compare time since fire (TSF) and grazer type (cattle or sheep).17_18_19_20vegFG.csv is the primary dataset for this paper and repository here. We collected vegetation structure and community composition data in 2017, 2018, 2019, and 2020.Columns Year through PastPatch are various grouping variables used throughout the analysis.Pasture is the primary ID for a given unitBlock is the assigned set of pastures the pasture matchesTSF is the time since fire for a given locationUse is whether the pasture or management unit was managed for heterogeneity or homogeneityManagement is the grazer type for pbg pastures and hay or idle for management unitsPatch is a subsection of the pasture or management unitPastPatch is a combination of the pasture name with the patch numberVOR: Visual Obstruction Reading was measured using a Robel pole marked and recorded in 0.25 dm increments. We took four readings per quadrat and calculated an average score from these.MaxLive and MaxDead: these were the tallest living and tallest standing dead plant material within the quadrat measured in 0.25 dm increments using the Robel pole.LitMean: We measured litter depth using a ruler to the nearest cm in the four corners of each quadrat. After 2017, we started recording all four measurements instead of just recording the average of the four measurements.BGCover: bare ground cover is any exposed soil surface than can be seen when looking down on the quadrat. We expected this to be higher in recently burned patches.GCover: ground litter cover is any visible horizontal ground litter than can be seen when looking down on the quadrat. We expected this to be higher in recently burned patches.LitCover: vertical litter cover is any visible standing or vertical litter than can be seen when looking down on the quadrat. We expected this to be lower in recently burned patches.Columns ACMI through VIAM are the 4 letter species codes used during data collection on a tablet to record cover by cover class. The tablet was programmed to autorecord a '0' for species that were not present in the quadrat.Columns NatForb through NatShrub are the calculated cover values for finer scale groupings based on native and introduced status.Columns Forb through Litter are additional calculated cover values.RadGraph.csv was used to expedite making a community composition figure that is now in the supplemental materials for the paper.