Four US lilac spring slopes stayed negative under a 14-day delay cap and a one-day annual change limit in 1965-2008

We tested how changes in reporting delay can affect spring-temperature slopes in historical lilac bloom records. We used 9,015 plant-years in four US region and period samples in 1965-2008. Linear programs gave exact fitted-slope bounds with plant and year effects and winter temperature control. Under a 14-day delay cap and a limit of one day of delay change per elapsed year, all four slope ranges stayed negative. Upper endpoints ranged from -2.34 to -0.56 days per C. The smallest annual change rates that permitted zero were 1.35, 1.77, 3.34 and 4.29 days per year. These are fixed-sample sensitivity thresholds, not estimates of actual delay or confidence limits. Unknown visits, observer changes, different plant types, selected long records and coarse weather limit inference. We provide code, endpoint assignments, source hashes, a raw rebuild and a figure. We do not estimate a causal climate effect or an air pollution effect.

phenology · lilac · observation bias · reporting delay · linear programming · sensitivity bounds · regional replication