The most common models describing the spring phenology of boreal trees start with the autumn dormancy, and proceed to the ontogenetic development is spring. According to Sarvas (1972, 1974), the spring development immediately follows the release of dormancy, and this formulation is widely used in current models. The recent findings in modelling the spring phenology of boreal trees have indicated that there may be something missing between the dormancy and spring ontogenetic development. We tested three modifications of the sequential model (based on Sarvas papers) against the original model to find out what the missing mechanism could be like. We also compared the results agains a simple model with the ontogenetic development starting from a fixed date in spring. This model, albeit being simple and having few parameters, gives a better fit of the model to observed phenological data. Thus it gives a good reference for model performance estimation.

The first model was the Sarvas model as presented in his papers. The autumn dormancy, or dormancy I, similar to the dormancy phase in sequential model, is followed by winter dormancy, or dormancy II. During the latter phase the state of the bud develops following an empirical function rather similar to that for the ontogenetic development, but saturating at a lower temperature than the latter. Temperature-driven ontogenetic development, as described in sequential model, follows the release of winter dormancy. One would expect the model to perform in a similar manner to the sequential model, and this indeed proved to be the case.

The second model assumes the state of ontogenetic development to be able to "reset" if air temperatures drop below a threshold value. The reasoning behind this model could be that in the first phases of the development accumulating temperature increases the concentration of some regulatory compound, which will be broken again if exposed to low temperatures. This model fitted the observed data somewhat better than the sequential one, but not as good as the fixed starting date model.

The third model describes winter dormancy with reversing development. In high temperatures the dormancy develops, in cold it declines, until a threshold of no-return is reached. The reasoning behind this model assumes accumulation of some controlling substance, declining in cold weather, which then triggers irreversible spring development. This model fitted the phenological observations for leaf unfolding and flowering of Betula and Populus as well as the simple, fixed starting day model, but was worse for the early flowering events of two species of Alnus.

To conclude, the presented model simulations seem to suggest that the current phenological models lack some phase between the autumn dormancy and ontogenetic development in spring. For leaf unfolding and flowering of Populus and Betula our results cannot distinguish between reversing, temperature-driven or light-driven models. For the flowering of Alnus, the latter model seems to perform better.