Using models to project how climate change might affect oak species distribution

Imagine if crystal balls which allowed us to catch a glimpse of the future, actually existed? What would you use them to see? 

Well, in a way – they do exist. In the abstract world of math and computers, “models” are fed datasets of current day information and asked to project future outcomes. Ecologists use them to forecast how current events will shape our future planet. This is what I have spent my summer, attempting to do. 

My project attempts to project the possible affects that increasing climate change may have on oak species distribution. Focusing on 27 oak species, located across the eastern United States – I am using information collected for over 81,000 sites from both current day Global climate models and future emission scenarios to predict what the region may look like in the year 2050. 

In order to do this, I spent the majority of the summer using an open-source (free) programming language called R to write dozens of computer programs, which allowed me to organize this collection of information into three datasets – present day, future emission scenario 1 and future emission scenario 2. These datasets, which basically look like giant excel sheets, are divided into rows representing site locations described by latitude longitude (over 81,000 rows) and columns representing presence absence data for each of the 27 oak species and relevant environment variables for the given location. In the future emission scenarios datasets, the presence absence data was omitted, because this is exactly what we are trying to project. 

I then analyzed the initial present day data using a statistical model, creating 27 new sub-datasets containing presence absence data for a given oak species and the top 5 environmental variables most relevant to that species survival. 

These new datasets, along with both of the future emission scenarios were fed into an ensemble suite of 9 separate statistical computer models. Maps were produced from these models, for the eastern United States, showing both current and predicted presence absence data for a given oak species. 

This was an amazing project to be part of. In the last few months, I have learned a great deal about the concepts of computer models and the extreme importance of validating the accuracy of our data. I have also fallen in love with programming and developed a strong sense of confidence in not only my abilities as a computer programmer, but also in the role that my skills can play in current day science. 

My project was really just a small part of a much larger goal, which will continue on without me. In the future, my work will be used as a reference against a new improved statistical model currently being developed by my mentors. I’m sad to leave Harvard Forest and the work that I have done here, but am comforted in knowing that my work will go on to become a part of “real science”. It is truly one of my greatest academic accomplishments and I am so grateful to have had this opportunity. 

[Elisabete (Baker) Vail is a senior at Simmons College in Boston, Massachusetts and will complete her double major in Biology and Computer Science this December.]