Announcing a newly funded PhD position to develop models of human cultural evolution under the influence of climate change and apply those models to a large data set on farming practices in rural communities in the northeast. Tasks include developing theoretical models of cultural adaptation to the effects of climate change, fitting and calibrating those models based on data assembled as part of this research project, and using the models to make predictions to help rural communities and policy makers understand and anticipate the needs of adapting to climate change.
The position is part of a newly funded $4 million collaborative research project with the on how both rural human communities and species populations will respond to challenges posed by climate change.
A team at UMaine and the University of Vermont has just been awarded a $4 million dollar grant to study climate adaptation in using data science tools. As part of the project, Dr. Waring will lead development of cultural evolution models of rural community adaptation to climate change. The team will explore what social and economic conditions determine how a natural resources-based community adapts to climate-induced change over time, and whether cultural adaptation models coupled with data on species changes can better inform farming practices in the future.
Niles will develop a large-scale spatiotemporal dataset that will focus on and inform farmer adaptation behaviors. The data will include projected range shifts in crops, models of key crop weeds and pathogens, and socioeconomic and demographic information on the rural resource users.
Together, the work of Waring and Niles will be the first to leverage significant ecological and social datasets to study climate adaptation in a spatiotemporal context. PhD positions will emerge soon.
Master’s student Antonio Jurlina has completed a simulation model of food buying clubs in which people share bulk purchases. Building on the work of Ethan Tremblay, Jurlina fit this model to data from 30 separate food clubs. The model simulates membership as a result of success in cooperative purchasing behavior between agents.
To simulate a club, the model takes a set of input variables: average catalog size, mean items per case, starting members, and the number of orders for a that club. It then creates a set of virtual agents who interact to share food purchases. Happy agents will help others. Unhappy agents will not help out. As a result simulated clubs have a cooperative momentum. This cooperative momentum determines whether members feel satisfied and stay in the club, or leave.
The model has two free parameters:
join_rate : the probability of additional members joining a club in a given order prosociality : the average tendency of club members to be willing to buy something they don’t prefer to help another member
We fit the model to the data by finding the values of join_rate and prosociality that produce the closest membership trend line. These results show that our model can mimic real world clubs with good accuracy. Our next step will be to test the predictive capacity of the model by doing out of sample prediction tests.