State-year controls did not establish a common spring warmth link to western US lilac duration in 1956-2003

We tested whether spring temperature links to common lilac flowering duration remain after state-by-year controls. We used 8,731 plant-year pairs at 606 western US sites in 1956-2003. The pooled slope changed from -0.448 days/C with year controls to -0.646 with state-year controls. The latter 95% interval was -1.306 to +0.013 days/C. Region and period checks did not establish a common shortening effect. State-year controls retained 49% of residual spring temperature standard deviation. Earlier first bloom remained clear in supporting models. This is a model sensitivity study of an existing archive, not independent biological validation. Missing visit histories, selected complete pairs, coarse monthly weather and few state clusters limit inference. We do not test pollution, daily extremes or causal climate effects. Code, exact inputs, source hashes, estimates and numerical checks are included.

phenology · lilac · spatial confounding · observation bias · fixed effects · flowering duration