Abstract
The flowering phenology of Coltsfoot ( Tussilago farfara L.) shows a very strong negative response to warmer temperatures in spring (warmer springs = earlier flowering) as reported in numerous studies for other species in Europe and North America. However, despite this, flowering dates in some locations have failed to show a strong advance and, in a small number of cases, have even shown significant delays. Here we use a very extensive data set on first flowering dates (FFD) of Coltsfoot in Germany, firstly on a national scale in the period 1894–2015 and secondly at a regional level in the more thoroughly recorded period from 1951–2015. FFDs were coupled with gridded temperature data using regression techniques to attempt to understand the reasons for the unexpected changes in phenology. National mean FFD did not change significantly over 122 years, although January–March mean temperatures had warmed by 1.2°C. The flowering‐temperature model was significantly improved when the previous September mean temperature was included, which decreased the overall temperature response from 4.5 to 3.7 days/°C. For the more thoroughly recorded period of 1951–2015, January–March mean temperature warmed significantly by 1.7–2.4°C in all 11 German federal state combinations and mean FFD dates advanced by 3–14 days, significant for seven of the States. For 10 of the 11 States, multiple linear regressions explained 79–89% of the observed variation in FFD and evidenced a positive relationship (significant in half of these) with higher temperature in the September preceding flowering (warmer September = later flowering). However, the scale of the September influence is insufficient to explain why Coltsfoot flowering may have stalled or been delayed. The use of controlled experiments to supplement observational data may be necessary to fully understand the drivers of Coltsfoot flowering in spring.