Bayesian correlation between temperature and blossom onset data

The recent quantification of changes in time series of phenology data with Bayesian methods has provided compelling evidence for changes during the last 20 years. In this paper we correlate the phenological observations with spring temperature time series. We provide quantitative answers to the question whether changes in temperature and phenological time series should be regarded as coherent or independent. For the three considered species snowdrops, cherry and lime tree we find factors of 1.05, 2.19 and 3.26, respectively, in favor of coherence. The functional behavior and the trend in the temperature time series are presented. They amount to 0.15 degrees C yr(-1) for the January-March average, 0.09 degrees C yr(-1) for February-April and 0.1 degrees C yr(-1) for March-May in 2002. In addition, we compare blossom trends for the coherent and independent hypotheses and find that the transition from trend values slightly positive before 1970 to strongly negative at present becomes sharper as the temperature data are included in the analysis.