Visit intervals permit both signs of March warmth slopes for beech-maple stage gaps at two New England forests in 1990-2015

We tested whether March warmth was linked to a larger beech minus sugar maple spring stage gap. We used fixed monitored groups at Harvard Forest and Hubbard Brook in 1990-2002 and 2003-2015. The four samples had 13, 13, 10 and 9 annual gaps. Models controlled January-February temperature and a linear year trend. Slopes from linear date estimates were +0.09, +0.51, +0.11 and +0.62 days per C. All four HC3 95 percent intervals included zero. When threshold dates could occur anywhere between recorded visits, the conditional slope bounds included both signs in every sample. The bounds ranged from -3.97 to +4.47 days per C across samples. They are not confidence intervals. A lag-1 Newey-West interval was positive for later Hubbard Brook, so uncertainty depended on the estimator. Two source date-field conflicts had little effect in a separate check. The evidence does not establish a common warming response or its absence. Few years, missing pre-event visits, different stage definitions and March weather correlation of 0.991 between sites limit inference. The package includes source hashes, derived event brackets, code, all 48 model fits and figures. It tests no air pollution or frost effect.

tree phenology · observation bias · visit intervals · American beech · sugar maple · Harvard Forest · Hubbard Brook · March temperature · partial identification