NSF Postdoctoral Fellowship in Biology FY 2017: Leveraging digital herbaria and crowd-sourced photos to understand climate-driven disruption of community flowering phenology

Breckheimer Ian K, Seattle WA

Investigators

Abstract

This is an NSF Postdoctoral Research Fellowship in Biology, under the program Research Using Biological Collections. The fellow, Ian Breckheimer, is conducting research and receiving training that utilizes biological collections in innovative ways, and is being mentored by Andrew Richardson at Harvard University. Specifically, the fellow will use digital records from museum collections and crowd-sourced photos uploaded by the public to measure where and when plants in different environments are exposed to risky climatic events like drought and growing season freezing during vulnerable periods in their seasonal cycle. Climate plays an important role in determining which organisms occur in which habitats, but exactly how this happens is still unclear, and understanding the mechanisms will allow us to better forecast the economic and ecological impacts of changes to the environment. Recent work suggests that, for non-woody plants like wildflowers, extreme events such as freezing or drought during vulnerable periods such as flowering strongly influence which plants can survive in which environments. The fellow will test this hypothesis for a large group of plants that inhabit mountain meadows in the Western USA. These are economically important ecosystems that attract millions of visitors each summer, and also experience extreme variations in climate. This project will advance our understanding of how climate affects the distribution of organisms, identify environments and species at risk of climatic disruption, and provide an important proof-of-principle for using crowd-sourced images to track spatial and temporal patterns in ecosystems. At approximately a dozen heavily-visited natural areas in the western USA, the fellow will use species occurrence information from digital herbarium collections along with image classifications from volunteer citizen-scientists and cutting-edge computer vision algorithms to identify common subalpine and alpine wildflowers in public geo-located photos hosted on social media platforms. By accounting for variation in the observation process, new statistical tools will allow the fellow to use these unstructured and imperfect observations to reliably measure the seasonal timing of flowering and fruiting. The fellow will then combine these observations with environmental information from satellites and weather stations to understand how that reproductive timing is affected by climate, and how geographic distributions and species responses affect the risk of disruptive climatic events during reproduction at each site. In addition to developing new knowledge about climatic disruption of ecosystems, this fellowship will support training in cutting-edge modeling and machine learning techniques, and advance new methods for combining citizen observations and museum collections. Results from these studies will be published in peer-reviewed journals and presented at scientific meetings.