Seven-day sighting delays permit zero butterfly-minus-swallow warmth slopes in Japan under one-day annual change limits, 1953-2020

We tested how detection delays affect a published link between spring temperature and relative butterfly and swallow first sightings. We used the same 4,077 paired station-years at 69 Japan stations in 1953-2020 and rebuilt the panel from raw public data. Four region and period models had recorded slopes of -1.70 to -2.22 days per C. If each species delay is zero to seven days and can change by at most one day per elapsed year, sharp conditional slope bounds include both signs in all four samples. The minimum annual limits for zero are 0.71-0.89 day per species. If only one species delay can vary under the same limits, all four fitted slopes remain negative. We give exact bounds, delay sequences, independent solver checks and an offline reproduction. The delays are assumptions, not measured errors. These bounds do not include sampling uncertainty or correct missing observations. Selected records, unknown effort and abundance, coarse weather and weather windows that extend past sightings limit biological interpretation. The result does not identify actual onset, a causal climate effect, pollution effects or a harmful timing mismatch.

phenology · observation bias · first detection · butterfly · barn swallow · partial identification · temperature · Japan · sensitivity analysis