After removal of counting noise, unsmoothed end levels and post-1800 samples, the late-Holocene rise in pollen rate of change remains in Europe but not in North America (1,166 Neotoma records)

Question: does a late-Holocene rise in pollen rate of change (RoC) over the mid-Holocene remain after removal of counting noise and processing artefacts? This study corrects and supersedes sci:sha256:1021e1b8b3aba4f89438f6588be4ea51185024ef0e546c5b8226aa0ab9399a18, which said that the noise-corrected rise was robust (ratio 1.35). Approach: we used a Python version of R-Ratepol (150 grains, chi-square distance, 500-year windows) on 1,166 Neotoma records of the Mottl et al. (2021) set, with an exact multinomial noise floor. We tested 10 variants: Shepard smoothing with all levels; smoothing without the 2+2 end levels that R-Ratepol leaves unsmoothed; no smoothing; removal of samples younger than AD 1800; and thinning to 200 or 400 years. The statistic is the within-record median ratio of noise-corrected RoC, 0-2 ka over 4-6 ka, with record bootstrap CIs. Results: the unsmoothed end levels inflate the ratio: global 1.35 (1.28-1.39) with all levels, 1.24 (1.19-1.32) without end levels, and 1.19 (1.14-1.22) without smoothing. Post-1800 samples cause most of the remaining North American rise: with end levels dropped or no smoothing, and no post-1800 samples, North America gives 1.00 (0.93-1.05) and 1.03 (0.98-1.07), and 0.90-1.04 with thinning. Europe keeps a rise in all these variants (1.31-1.47; 1.26 with a 4-8 ka reference). The global ratio falls to 1.12-1.17. Asia, Latin America and Africa are not resolved. Thinning has little effect. Counting noise still causes 46% (41-53%) of the global increase in squared RoC, which agrees with sci:sha256:8a56058addd44b2c1c651f09ce043d8794fe0b8ae31e9984fe44982635609e9a. Thus these records support a late-Holocene acceleration over the mid-Holocene in Europe only, not a worldwide one. Limitations: our age models and taxonomy differ from those of Mottl et al.; we use no age-model uncertainty; AD 1800 is an approximate settlement limit; the corrected RoC still includes real short-term variability.

pollen · rate of change · Holocene · Neotoma · palaeoecology · R-Ratepol · sampling noise · smoothing artefact · land use · Europe · North America