CR 32:253-267 (2006)  -  doi:10.3354/cr032253

ABSTRACT: The annual timing of leaf colouring of deciduous trees in temperate regions is not predictable using phenological models. In this analysis we show that commonly applied hypotheses of leaf colouring triggers have neither satisfactory explanatory power nor significant statistical proof. We tested meteorological parameters, such as monthly mean temperatures, threshold temperatures, monthly sums of precipitation and number of dry days per month of the year of the phenological event and of the previous year and the length of the vegetation period. Their influence on leaf colouring dates for 4 deciduous tree species (horse chestnut, beech, birch, oak) in Germany (1951–2003) was tested by Pearson’s correlations. We created 3 different datasets: (A) phenological observations for single stations within 25 km of meteorological stations, (B) means of phenological observations around (<25 km) meteorological stations, and (C) phenological means for Germany. Only the mean temperature of September had a slight influence on the onset of leaf colouring (mean: r_chestnut = 0.45; r_beech = 0.56; r_oak = 0.51; r_birch = 0.45). Taking all correlation coefficients r > |0.3| into account, we can deduce that a warm September (all species) and August (oak, birch) delayed leaf colouring while a warm June (horse chestnut, oak) and May (horse chestnut) advanced leaf colouring.

KEY WORDS: Response · Leaf colouring · Autumn phenology · Growing season · Vegetation period · Germany · Climate change