Including other plants reduced the biomass stability benefit of sown diversity at Cedar Creek in 2012–2021

Does biomass from plants outside a sown list change the measured diversity benefit for stability? We tested 154 fixed plots in the Cedar Creek e120 grassland experiment. We used ten harvests in 2012–2021 and matched sorted strips for sown and all live biomass. Source weed groups were treated as other plants. Temporal variation was the sample standard deviation divided by the mean (CV). Per doubling of sown richness, the log-CV slope was −0.1635 for sown biomass and −0.0638 for all live biomass. Their paired difference was +0.0997, with a nominal 95% plot-bootstrap interval of +0.0630 to +0.1371. Including other plants reduced the slope magnitude by 61% (46–76%), conditional on the observed years. In monocultures, the geometric mean CV ratio was 0.671, but the standard deviation ratio was 1.278. An exact partition showed that added mean biomass explained most of the slope change across richness levels. Negative covariance also helped monocultures, but did not lower total standard deviation below sown standard deviation. Common time-block resampling, separate record-format years, and one-year omissions kept the direction. Removing aggregate weed biomass weakened the contrast and changed the response. Both biomass measures retained a diversity benefit. Limits include one managed site, small harvest samples, ten years, changing sampling methods, and the weed-group classification. This is a biomass accounting result. It does not identify an ecological cause or future climate resistance. The package contains source files, provenance, code, all estimates and resamples, figures, and independent reconstruction checks.

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