In 261 BioTIME 2.0 studies, start-year dominant species decline equally in time-reversed series, so the pattern is regression to the mean and not a directional loss of dominance

Question: some analyses of assemblage time series report that dominant species decline and rare species increase. If species are put into abundance classes at the start of a series, regression to the mean can make this pattern without real change. We tested how much of the pattern is directional. Data: BioTIME 2.0 (release of 15 April 2025), 2,595 assemblages (study x 1-degree cell) from 261 studies, 1910-2023, at least 10 years each. Each year was rarefied to equal samples. Method: for each species we calculated the trend in log abundance per decade. We grouped species by abundance relative to the top species. We compared start-year classes in forward time with end-year classes in reversed time. If dynamics are time-symmetric, the two results are equal, and only directional change makes them different. We also used classes from whole-series mean abundance, identity-free dominance trends (Berger-Parker, Pielou), a year-permutation null, and simulations with known truth run through the same code. Results (study-weighted means, 95% study bootstrap CIs): with start-year classes, dominant species declined by -0.53 per decade and rare species by -0.26 (gradient -0.27 [-0.43, -0.11]). The reversed series gave a similar or stronger gradient (-0.39 [-0.52, -0.26]). The directional part was +0.06 [-0.05, +0.17]. This CI excludes the -0.13 given by simulated erosion that removes 10% of species log-abundance deviations over 20 years. With whole-series classes, the dominant-minus-rare gradient was +0.08 [-0.08, +0.24]. The Berger-Parker share changed by -0.008 [-0.019, +0.003] per decade. No robustness variant (second rarefaction seed, series of 20 years or more, count data, untreated studies, 3-year baselines, rank classes, each realm) gave a directional gradient in the erosion direction with a CI that excluded zero. Limits: the reversal test assumes time-symmetric dynamics with no directional change. The simulation calibration uses one simple community model. We did not reproduce the exact design of the restricted code of the 'erosion of dominant species' study. Code, derived tables and a reproduction script are included.

BioTIME · regression to the mean · dominance · rarity · population trends · assemblage time series · time reversal · Berger-Parker index · biodiversity change · null model