Predicting forest responses to warming climates relies on assumptions about niche and temperature sensitivity that remain largely untested. Observational studies have related current and historical temperatures to phenological shifts, but experimental evidence is sparse, particularly for autumn responses. A 4 year field experiment exposed four deciduous forest species from contrasting climates (Liquidambar styraciflua, Quercus rubra, Populus grandidentata, and Betula alleghaniensis) to air temperatures 2 and 4 °C above ambient controls, using temperature-controlled open top chambers. Impacts of year-round warming on bud burst (BB), senescence, and abscission were evaluated in relation to thermal provenance. Leaves emerged earlier in all species by an average of 4–9 days at +2 °C and 6–14 days at +4 °C. Magnitude of advance varied with species and year, but was larger for the first 2 °C increment than for the second. Effect of warming increased with early BB, favoring Liquidambar, but even BB of northern species advanced, despite temperatures exceeding those of the realized niche. Treatment differences in BB were inadequately explained by temperature sums alone. In autumn, chlorophyll was retained an average of 4 and 7 days longer in +2 and +4 °C treatments, respectively, and abscission delayed by 8 and 13 days. Growing seasons in the warmer atmospheres averaged 5–18 days (E2) and 6–28 days (E4) longer, according to species, with the least impact in Quercus. Results are compared with a 16 years record of canopy onset and offset in a nearby upland deciduous forest, where BB showed similar responsiveness to spring temperatures (2–4 days °C−1). Offset dates in the stand tracked August–September temperatures, except when late summer drought caused premature senescence. The common garden-like experiment provides evidence that warming alone extends the growing season, at both ends, even if stand-level impacts may be complicated by variation in other environmental factors.