Spring animal weekend shares were higher than plant shares in three Japan samples, but later results depended on observation location

We compared weekend shares of nine spring animal events and nine plant events in JMA records from 1953 to 2020. There were 5,256 paired station-years in two regions and two periods. Animal minus plant contrasts were +2.88, +1.85, +3.60 and +3.62 percentage points. Three passed both four-test adjusted calendar-phase tests and two-way cluster intervals. Equal station weights and removal of the repeated-record rule kept the signs. Matching observation-location flags changed the later contrasts to -0.22 and +2.10 points. On-site-only later samples had negative point estimates and much smaller support; one cluster variance was negative. The results show a calendar pattern that depends in part on observation location. They do not measure biological delay, climate effects or pollution effects. Code, dates, paired evidence, source hashes and checks permit offline reproduction.

phenology · observation bias · JMA · weekend · calendar · species timing