Total biomass had less relative variation than biomass of sown plants in 154 Cedar Creek plots, 2001–2021

We tested whether plants that were not sown offset temporal variation in biomass of sown plants. We used 154 fixed plots in the Cedar Creek Biodiversity II experiment and 20 sorted harvest years from 2001 to 2021. We excluded 2009. We measured both plant groups in the same samples. The geometric mean ratio of total biomass CV to sown biomass CV was 0.806 (95% plot interval 0.773 to 0.839). CV is standard deviation divided by mean. An exact split of the log ratio gave a mean-biomass term of -0.361, an added-variance term of +0.186, and a covariance term of -0.040. The mean covariance share was -0.046 (plot interval -0.079 to -0.013). After removal of linear trends, it was +0.008 (-0.022 to +0.037). The raw estimate also included zero with common-year block sampling. Thus, the higher mean accounts for most of the lower CV. Evidence for a sustained offset of annual variation is limited. The results apply to one managed experiment and recorded harvests. The package contains unchanged source data, provenance, code, full results, figures, and independent validation.

biodiversity · grassland · Cedar Creek · biomass · temporal stability · covariance · coefficient of variation · reproducible research