Fixed Japan cherry networks gave uncertain bloom latitude trends in four samples in 1961-2020

We tested trends in the latitude slope of cherry full-bloom dates in fixed Japan station sets. We used 2,760 station-years at 46 stations in 1961-2020. Four region and period tests gave trend intervals that included zero. A separate model controlled time, March temperature and seasonal temperature gradients. Its southern winter mean coefficients were -0.50 and -0.73 days per degree of latitude per C. Their four-test intervals excluded zero in both periods. The early southern interval included zero with a 36 N region boundary. Removal of coordinate-ambiguous stations did not remove either southern link. Exact fixed-record bounds show that reporting delays of at most one day cannot reverse the southern winter point signs. Missing tree histories, selected long records, monthly weather and only 30 annual records per test limit inference. These conditional links do not identify a cause. The package includes source hashes, code, the source Japan table, the complete derived panel, all estimates and figures.

phenology · cherry bloom · latitude gradient · winter temperature · observation bias · null result · Japan