Earlier spring green-up gave only 0.17 day earlier autumn green-down per day at 133 PhenoCam deciduous sites in 2002–2025, one third of a ground-based carry-over estimate
Some autumn phenology models assume that an earlier spring causes an earlier autumn. Keenan and Richardson (2015) reported 0.52 day per day from ground data and 1.1 day per day from MODIS. We tested this carry-over slope with PhenoCam canopy greenness (gcc_90 transition dates) at 133 deciduous broadleaf camera sites (1,037 site-years, 2002–2025; 119 sites in North America). We compared years within each site, controlled for autumn temperature (NASA POWER; Daymet as a check), and used a bootstrap of 98 location clusters. A 1-day earlier spring green-up (50% of amplitude) came with a 0.17-day earlier autumn green-down (95% interval 0.04 to 0.32). Thus about 83% of a spring advance stayed in the green season length. The upper 95% bound was below 0.52 in 20 of 26 variants and below 1.1 in 25 of 26. A placebo with the next year's spring gave 0.06 (−0.07 to 0.19). The slope depended on the date definition: 0.02 (−0.08 to 0.10) with 10% dates, 0.08 (−0.02 to 0.17) with 25% dates in eastern North America, and 0.36 with our own site-fixed thresholds (0.12 to 0.54 at 25%). A per-year amplitude error does not explain the positive slope. Autumn warmth delayed the green-down by 2.8 days per °C (2.0 to 3.5). In 30-day windows, the temperature effect changed from negative (July, intervals including zero) to positive for windows centred after early August, not at the summer solstice. A joint monthly model gave −1.04 day per °C for July (−1.68 to −0.25), but this July effect was not significant with year fixed effects. Limits: GCC is canopy greenness, not leaf colour or photosynthesis; most sites have 4–10 years; the European subset is small; a cross-camera instrument test in 17 multi-camera clusters was inconclusive. Code, derived tables and retained raw data are included or referenced.
phenology · PhenoCam · autumn senescence · carry-over effect · deciduous forest · growing season length · green chromatic coordinate · summer solstice · placebo test · cluster bootstrap
References
10- challengesdoi:10.1111/gcb.12890
We measured the spring-to-autumn carry-over slope with PhenoCam canopy cameras and found it about one third of their ground-based value and one sixth of their MODIS value.
- usesdoi:10.1038/sdata.2018.28
PhenoCam data processing, GCC metrics and transition dates.
- usesdoi:10.1038/s41597-019-0229-9
PhenoCam Dataset v2.0 methods for the transition-date algorithm.
- usesdoi:10.3334/ORNLDAAC/2129
Daymet V4 R1 daily temperature and precipitation for North American sites.
- doi:10.1126/science.abd8911
Productivity-driven earlier senescence hypothesis; we do not test the productivity mechanism directly.
- doi:10.1126/science.adf5098
Solstice reversal hypothesis. In PhenoCam canopy greenness the temperature effect changed sign in early August, not at the solstice, but single-window intervals before August include zero and the July signal is not robust to year effects.
- supportsdoi:10.1111/gcb.16104
Their weak support for productivity-driven earlier end of season agrees with the small carry-over slope we found.
- doi:10.1073/pnas.1321727111
Experimental evidence for spring-to-autumn carry-over.
- supportsEarlier budburst did not cause earlier autumn yellowing or leaf fall in the same ginkgo and mulberry trees at Japanese weather stations in 1953–2020 when autumn temperature is controlled sci:sha256:11f49f6ad83249631ee87728c5812f778dc56e0d8946e8e047891f31291f2783
Ground observations in Japan also show no strong carry-over when autumn temperature is controlled; our camera data agree.
- Pre-solstice warming advanced and post-solstice warming delayed leaf colouring of four German tree species (1951–2025), but these effects did not predict later colouring in 2016–2025 sci:sha256:d7c52bfcf1d62961fcc2fe0b7d0be067d0bea09726c9ebbdd4f36510d36fc99c
German ground data supported a solstice reversal; our canopy greenness data show a later reversal.
Cited by
1 publicationPublic catalog entries on Sciscape.
- Less summer rain had no clear link to autumn greenness dates at 43 PhenoCam sites in 2008–2024
Related analysis of spring-to-autumn timing in the same camera network, with rain controls in some models.