In linear models, wetter previous springs were linked to a larger biomass advantage of planted diversity at Cedar Creek, 1998–2020

We tested whether rain in the previous spring was linked to the next year's biomass advantage of planted diversity. We used 154 fixed Cedar Creek grassland plots with 1, 2, 4, 8, or 16 assigned species. The lag model had 23 harvest years, from 1998 to 2020. After adjustment for current April–June rain and linear time, an extra 100 mm of previous spring rain was linked to a 7.7% larger fitted biomass ratio for 16 species versus one species. The nominal 95% three-year residual-block interval was 2.9% to 12.7%. The model used all five richness levels. Planned checks and later added checks kept the lag estimate positive. These included quadratic and piecewise time trends, annual differences, heat, source duplicates, and inter-harvest rain. The delayed link to mean biomass across all plots was uncertain. Rain was not assigned to years, so the result does not identify a drought mechanism or a recovery rate. A simulation with correlated annual errors gave 91.7% coverage for the nominal 95% interval procedure. The short time series, changing station coverage, and one managed site limit inference. The package preserves raw data, source rights and hashes, code, plans, all results, figures, and independent checks.

biodiversity · grassland · rainfall legacy · Cedar Creek · plant biomass · reproducibility