Relationship between flowering phenology of perennial herbs and meteorological data in deciduous forests of Sweden

The relationship between flowering phenology and meteorological measurements, in particular air temperature, was studied for 29 herbaceous species in four areas of deciduous forest near Uppsala, Sweden, Altogether 16 models were tested for their accuracy of predicting flowering. These were cumulative sum models based on the heat unit concept of an accumulation of (modified) temperatures above a threshold base temperature from a chosen starting date. Average temperature was tested as an alternative method. All models were first applied to a data set from the years 1990 to 1992 and then to an independent test data set from 1993. The accumulation of daily mean air temperatures (degree-days) above 5 degrees C from January 1 was chosen as the basic reference model. Despite its simplicity, it was a very accurate model in predicting flowering for these species. Only two models were superior to the reference model in both data sets: the summation of degree hours instead of degree-days from optimized starting dates for each species, and the addition of photoperiod (daylength) to daily mean temperature. In general, the models performed better for the late-flowering species than for the early-flowering species. The accuracy of the models was partly dependent on the actual course of temperature accumulation in a year.