False estimates of the advance of spring

Bias is introduced into almost all recent reports of climate-related trends in the phenology of spring events (for example, the timing of migration, egg laying and ice melt) by giving the calendar date of such occurrences each year, rather than their timing relative to the vernal equinox (refs 1–8, but see ref. 9). Most of these studies overestimate the advance of spring events, as the calendar date of the vernal equinox shows a trend to become earlier throughout any century, although this bias is small in the examples published so far. However, its magnitude cannot be predicted for any data set that is extended into the twenty-first century, because of long-term changes in the date of the vernal equinox. As phenological data are important for studying climate change, trends need to be reported in terms that accurately reflect changes to the Earth system.