This experiment took place within the Phosphorus Plots experiment in watershed 2C. In 2010 we built fully covered rainout roofs to exclude the majority of rainfall. Roofs were placed in the control, 10 g N, 10 g P, and 10 g N+P treatment plots. Shelters were 2.3 m × 2.3 m, which effectively droughted a 1 m2 area in the center of the plot. The shelters went up in April and were taken off in October for each year of the three-year imposed drought, removing rain during that period. We then tracked recovery for three years (2013-2015).
Species composition: Every year (2010-2015) community composition was recorded by estimating the cover of each species to the nearest 1% in late May or early June and in August. Between these two time points the maximum cover of each species is used in subsequent analyses. Species cover data was collected in a 1 m2 plot in the center of the shelter, where there was less blow-in, and was divided into four 0.25 m2 quadrats. Data from the four quadrats were averaged.
Standing plant biomass: In the years that the site was burned, 2011, 2013, 2015, we used a disc pasture meter, a weighted plate (50 cm diameter) that is dropped onto a plot, and its height is a proxy for standing biomass. In burn years, all aboveground biomass was removed by the fire, and standing biomass is a proxy of how much plants grew over the growing season, or biomass production. We measured standing biomass in each of the subplots within the species composition plot taking four measurements in the four quadrants. The disc is calibrated to Konza uplands (where the experiment took place) and equations were used to convert the height of the disc to standing biomass.