Figure 4.12 Bayesian analysis of flowering of lilac (Syringa vulgaris) at Gr¨unenplan, Germany
(1951–2000). On the upper left plot, the original data series is plotted with a notable scatter of the data,
and these data points are superimposed on the remaining graphs. On the lower left plot the change point
probabilities (i.e. the likelihood of a gradient change in the data series), normalised to unit area, are
displayed with a clear maximum in the first half of the 1980s. The rate of change, determined by the
one change point model (lower right plot), reaches −1.19 days/year; the confidence ranges indicate
advancing flowering, clearly different from zero change. On the upper right plot, the average functional
behaviour and its confidence ranges estimated from the one change point model are plotted (Schleip,
2005).