Winter warmth gave uncertain tree-weed peak-gap slopes in four eastern US samples with dates limited to day 213

We tested winter warmth and the gap between tree and weed pollen peaks in 388 paired station-years at 34 eastern US monitors in 2003-2024. Four site and year models controlled seasonal observation counts. Gap slopes were positive, from 4.79 to 8.01 days per C, but all four corrected intervals included zero. Higher counts, site trends, summer temperature and two-way uncertainty also gave intervals across zero. The initial count rule left only two and one northern sites, so the main rule was relaxed before completed estimates were read. A source audit found that all 2,350 valid peak and end dates in the extract were at or before day 213. The cause is unknown. These estimates therefore apply to released peaks and cannot describe full annual tree-weed timing. Count filters also select longer reported seasons. Broad taxa, missing visits, few counties, coarse weather and date support limit inference. The package gives code, aggregate evidence, source hashes and retrieval instructions. External records remain outside the bundle.

phenology · pollen · observation bias · temperature · paired contrast · source audit · null result