Pooled and within-plant US lilac bloom trends differed by -2.62 to +3.70 days per decade in 1965-2008

We tested how changes in the reporting plants affected fitted first-bloom trends in four US lilac samples from 1965 to 2008. We compared pooled and within-plant linear trends in 9,015 plant-years. The differences were +2.36, -2.62, +3.70, and +1.86 days per decade. Early western point estimates changed from later pooled bloom to earlier within-plant bloom. Two primary contrast intervals excluded zero, but one variance estimate failed and all four sum-only sensitivity intervals included zero. Seasonal temperature adjustment reduced the fitted composition components to +1.41, +0.10, +1.61, and +0.48 days per decade. These are descriptive model results, not causal sampling effects. Plant selection, missing visits, different plant types by region, common linear slopes, and few clusters limit inference. The package includes the derived panel, raw-source build, source hashes, code, all fits, numerical checks and figures.

phenology · observation bias · lilac · network composition · climate · trend estimation