The limits of phenology

A new statistical method to track climate change

Like Comment

The paper in Nature Ecology & Evolution is here:

Species must either adapt in-place to survive climate change or migrate elsewhere to track their prefered environmental conditions. Increasingly, the phenologies of species - the timing of their life history events - are changing in spring, with flowers opening earlier, or birds migrating sooner. Measuring the degree of this phenological change is challenging because it's difficult to get good data before climate changed (when did plants start flowering in the past?), and it's hard to be certain when an event actually took place (On a Saturday in spring you noticed the first violet flower; did they first flower that day, or during the past week when you were busier?). In our paper, we apply a new method to estimate the onset of events, allowing us to more precisely combine historic herbarium and museum data with contemporary observations to detect evidence of climate change.

This work is a great example of the power of collaboration. One of our collaborators is Henry David Thoreau, who died over 150 years ago. Using a study he initiated, we were able to verify that our approach can accurately estimate the date of first flowering using herbarium samples. We also worked with the National Phenology Network, which has amassed millions of records collected by citizen scientists across the USA. This was possible because our approach estimates first flowering time from the entire distribution of flowering times. This provides a more accurate estimate of the true onset of flowering than using only the first observed flowering date.

Henry David Thoreau monitored flowering times in Concord from 1852 to 1858; his data is a key component in our study. Photo by Richard Primack and Abe Miller-Rushing.

Perhaps the biggest surprise of our study was the increase in variation of date of first flower over time in both the National Phenology Network data and in a multi-decade study of flowering times in the Colorado Rocky Mountains. It's well-known that flowering dates are getting earlier in response to a warming climate; but we also found an increase in variability through time, and it's not clear why it should be happening. One possibility is that we're approaching the limit of plants' capacities to flower earlier, and species might be about to undergo some type of phenological shift. Of course, it could also just be due to more variation in the weather or an artefact of the data; as with all science, we’re still working on this.

A variety of flowers (and some of their pollinators!) from the Colorado Rocky Mountain dataset, arranged in flowering order from the early spring beauty at top to the late gentian.

We hope that others will use this new method to study other biological systems in order to get more accurate estimates of the start and end of phenological events. The approach could be used to estimate other statistical distributions, such as the range of temperatures a species can tolerate or when the first ice cream truck appears in the summer. We also hope it facilitates new collaborations among field ecologists, museum based scientists, and modellers.

William David Pearse

Assistant Professor, Utah State University