Both strips had biomass below half their own medians less often at higher planted richness at Cedar Creek in 2010–2021

We tested severe low biomass in 154 Cedar Creek grassland plots assigned 1, 2, 4, 8 or 16 species. Each plot had two harvest strips in each year from 2010 to 2021. An event required both strips to have live biomass below half their own medians for this period. There were 32 events in 1,848 plot-years. Monocultures had 17 events in 384 plot-years (4.43%). Plots assigned 16 species had one event in 420 plot-years (0.24%). The fitted frequency slope was -1.04 percentage points per richness doubling. Its 95% plot bootstrap interval was -1.67 to -0.49 points. With plots and common three-year blocks sampled, the interval was -1.57 to -0.10 points. The direct 16-minus-1 contrast had a time-block interval that reached zero. Milder thresholds and an early-period check had slope intervals across zero. Independent source sums and event checks passed. A clean offline repeat matched all tracked results and figures. This is a new analysis of shared data, with a complete event dataset. It does not identify climate causes or estimate future event risk. The result is limited to two samples per plot, retrospective medians, the severe threshold, and the stated site and years.

biodiversity · grassland · biomass · species richness · severe low biomass · Cedar Creek · paired harvest samples · reproducibility