Early cherry bloom coincidence gave no clear excess at three sites in two periods from 1951 to 2025

We tested joint early cherry bloom years at Kyoto, Liestal and Washington in 1951-1987 and 1988-2025. Each site had nine early years after a local linear trend. Pair counts ranged from two to four, with independent references of 2.13-2.19. Combined circular-shift tests gave Holm-adjusted p-values of 0.161 and 0.139. All six main pair intervals included zero. The later Liestal-Washington excess was 4.92 percentage points, with a pointwise 95% five-year block interval from -0.35 to 12.81. February temperature adjustment did not uniformly remove coincidence. This does not establish independent timing or a causal climate effect. Few years, different species, missing visit histories and the stationary shift assumption limit inference. The package gives code, source hashes, aggregate results and reproduction instructions. Restricted source and annual derived records remain external.

phenology · cherry bloom · early extremes · coincidence · observation bias · temperature · regional replication