Temporal log biomass covariance between strips decreased with planted richness at Cedar Creek in 2010–2021

We tested whether a plant richness pattern in temporal biomass variance remained when we used separate spatial samples. We used two clip strips in each of 154 Cedar Creek grassland plots from 2010 to 2021. Planted richness was 1, 2, 4, 8 or 16 species. We split the temporal variance of mean strip logs into covariance between strips and strip disagreement. The split is exact. The total variance slope was -0.01976 per doubling of planted species. The covariance slope was -0.01443, with a 95% plot interval of -0.02630 to -0.00384. Strip disagreement supplied 27.0% of the total slope, with an interval of 14.1% to 60.2%. The covariance slope remained negative in checks of year blocks, linear trends, sorting roles, mixtures alone, and an early four-strip window. This result concerns temporal variation shared by the sampled strips. Covariance between strips estimates a common signal variance only under an error model. Shared errors and real spatial variation remain unresolved. An added model sensitivity check shows that assumed shared errors can change the inferred signal slope. The data do not estimate that error correlation. The package preserves unchanged source files, provenance, code, all results, figures, independent checks and offline reproduction steps.

biodiversity · grassland · temporal stability · biomass · spatial sampling · observation error · variance decomposition · Cedar Creek