Is spring starting earlier?

Time series of leafing phenology, air temperature and precipitation along with regional changes in carbon dioxide concentration, are analysed in addressing three questions: `what magnitude of changes in flushing day are likely with climate change?', `which taxa have a chilling requirement?' and `is spring starting earlier?'. Phenological observations of leafing, starting in 1739, are used to calibrate a new model that describes the time of flushing. Statistical methods are used to provide full standard errors on the model coefficients. All 13 tree types studied are found to show an association between flushing date and air temperature. Nine out of the 13 tree types show non-linear temperature behaviour. Model output indicates that climate change has a particularly strong effect on flushing date in oceanic climates, such as that of Britain. The modelled responses can be used to estimate the impact that recent changes in temperature have had on the timing of the start of spring. We find that in any future climate-warming, flushing dates of tree types sensitive to `chilling', for example beech and horse chestnut, are likely to move 30 days, or more, out of synchroneity with those sensitive only to springtime warmth. The new phenological model is also applied to fluctuations in the carbon dioxide mixing-ratio. Changes in the phase of the annual CO 2 cycle are isolated and used as evidence for an earlier spring. We make use of a complex demodulation analysis of carbon dioxide, as monitored at 27 stations around the world, to reveal that the timing of the annual CO 2 cycle has become steadily earlier in recent decades. Finally we show that the estimates of the effect of rising temperatures on the earliness of spring, made using our new phenological model, are in good agreement with global observations of changes in the timing of the annual carbon dioxide cycle if about 50% of the biomass of the world's terrestrial biosphere is fully temperature sensitive and about 50% sensitive to chilling (or to an equivalent inhibitory effect).