I have been at this for some time with Ben VanderWeide, and hopefully we are nearing publication. The focus of most work to date that studies how variation in temperature affects when things flower has focused on species that flower early. I think people focus on these species, in part, because their early flowering is conspicuous. We are all looking for blossoms in spring. These species also clearly respond to temperature around and immediately before they flower. The prevailing view is that later flowering species are less sensitive to warming. This, however, is not borne out by data from 261 species in the northern Flint Hills of Kansas dating back to 1893.

Instead, regardless of when they flower on average, species respond to spring warmth by advancing flowering. Late species, however (and other studies have shown this too) delay flowering in response to warmth during other times of the year, particularly later in summer. The result of this is that later species are actually more sensitive to warming, if one sums sensitivity to warming (slope estimates of regressions of mean first flowering date (FFD) against monthly mean temperatures) across months during the year leading up to flowering. The figure below shows this nicely. Cumulative sensitivity (the sum of absolute values of positive and negative sensitivities) is greater, in general, for species flower later in the year (panel A), and this seems to be because they have greater delays (panel C), while there is not much relationship between mean FFD and phenological advancement (panel B). We used quantile regression, because of heteroscedasticity (more variation for later flowering species), but in panels A and C all positive slopes are significant.