Relative biomass variance was lower in single 16-species plots than in monocultures, but the 16-plot mean contrast was uncertain at Cedar Creek, 1997–2021

We tested whether lower single-plot biomass variance at high planted richness also gave lower variance when plot records were averaged. We used 154 fixed Cedar Creek plots and 25 annual harvests from 1997 to 2021. Each plot index is live biomass divided by its own time mean. The mean individual temporal variance was 0.1606 in monocultures and 0.0757 in 16-species plantings. The ratio was 0.471, with a 95% plot and common three-year block resampling interval of 0.306 to 0.676. For means of 16 distinct plot indices, the expected variance ratio was 0.808, with interval 0.358 to 1.666. The pair covariance estimates were similar, but their difference was uncertain. Removing linear trends or using 2010–2021 reversed the pooled point estimate. The data support lower single-plot relative variance more consistently than lower pooled variance. This is a retrospective index comparison at one local field. It does not test separate landscapes or identify the cause of shared fluctuations. The package preserves unchanged source data, provenance, code, all numerical resamples, figures and repeat checks.

biodiversity · grassland · biomass · temporal variance · covariance · Cedar Creek · spatial insurance · reproducibility