Temperature means that end at bloom retained positive date links after year shuffling in four US lilac samples from 1965 to 2008

We tested whether a temperature mean that ends at reported bloom can create a date link by itself. We used 9,015 US lilac records in two regions and two periods from 1965 to 2008. After 999 date shuffles among years within each plant, all four samples retained positive correlations for January-to-bloom temperature means. Median correlations were 0.535, 0.415, 0.374 and 0.403. Fixed end dates gave shuffle ranges centered near zero. The same seasonal curve in every year gave correlations of 0.896 to 0.994. A proof shows why a mean can rise with its end date without annual weather change. These are method results, not plant response estimates. Shuffle ranges are not population confidence intervals. Selected planted shrubs, unknown visits, missing reports and monthly weather limit transfer. We do not measure pollution, daily extremes, biodiversity change or causal climate effects. The package includes source hashes, a raw rebuild, derived inputs, all draws, code, numerical checks, a proof and a figure.

phenology · temperature window · statistical coupling · observation bias · lilac · randomization · reproducible method