Evidence of phenotypic plasticity and spurious relationships in long-term phenological responses to climate change
Although there is evidence of directional selection for earlier flowering in response to climate change, relatively little is known about the role of phenotypic plasticity in shifted flowering phenology. Plasticity in flowering phenology is likely to play a critical role in avoiding harsh abiotic conditions because it allows individual plants to track interannual variation in the abiotic environment. We quantified plasticity in flowering phenology by regressing the residuals from separate regression models of flowering phenology vs. year and climate vs. year against each other. These residuals represent year detrended climate and phenology variables, so that interannual variation in climate and phenology is de-coupled from long-term directional change. A significant relationship between these residuals is evidence for phenotypic plasticity in flowering time, whereas no relationship between these residuals may represent a statistical byproduct of regressing two variables that both change directionally through time. Using two long-term records of flowering phenology for perennial plant species from subalpine meadow habitat in the Colorado Rocky Mountains (42 species; 1974–2013) and arctic tundra in Greenland (5 species; 1996–2013), we examined evidence of plasticity in the date of peak flowering in response to increasing temperatures and earlier snowmelt dates, separately.
In the subalpine plant community, relationships between peak flowering date and snowmelt date and peak flowering and temperature were still significant after detrending for all 42 species. Peak flowering becomes earlier as snowmelt date becomes earlier and as temperatures increase. None of the detrended slopes differed significantly in magnitude from the regular slopes, suggesting that most of the variation in peak flowering date explained by these abiotic factors represents plasticity in flowering phenology. In the arctic plant community, results differed depending on whether snowmelt date or temperature was used to predict flowering phenology. All phenology-snowmelt relationships persisted after detrending, but 4 of 5 phenology-temperature relationships disappeared after detrending. This suggests that phenology-temperature relationships in the arctic arose as a consequence of regressing two variables that significantly change through time against one another. This study suggests that spurious climate-phenology relationships are most likely when the climatic variable shows a relatively strong directional change through time and when there is a relatively weak relationship between phenology and climate. We encourage future studies of long-term phenological responses to climate change to test routinely for spurious climate-phenology relationships, which will ultimately lead to an improved mechanistic understanding of how phenology responds to abiotic variation.