Liestal and Washington cherry bloom dates had positive annual agreement in 1988-2025 after local trends

We tested annual agreement in cherry bloom timing at Kyoto, Liestal, and Washington in 1951-1987 and 1988-2025. Each site had 75 annual records. After separate local linear trends, Liestal and Washington had a correlation of 0.486 in 1988-2025, with a 95% block bootstrap interval of 0.216 to 0.696. A post hoc spectral null test gave p = 0.0228 after correction for six tests. Its earlier correlation was 0.257, with an interval of -0.013 to 0.523. The interval for the period change included zero. Quadratic trends, first differences, and winter ENSO adjustment kept positive later point estimates for this pair. Neither simple shared trends nor the measured ENSO covariate removed its agreement. Other pairs gave less clear results. The first planned circular-shift rule could not pass the multiple-test threshold; we preserve that design failure. The spectral result needs a stationary random-phase null and independent validation. Different species, bloom rules, site histories, and unmeasured local weather limit inference. We did not measure pollution, damage, or biodiversity loss. The package gives code, exact source hashes, annual residuals, all tests, a figure, and reproduction instructions.

phenology · cherry bloom · spatial synchrony · ENSO · observation bias · temporal replication · spectral null