ILWS PhD Candidate
The transferability of species distribution models into new environments
Associate Professor Rachel Whitsed and Dr Ana Horta
ILWS PhD scholarship recipient Liam Grimmett is a former CSU Bachelor of Environmental Science graduate who went on to work for what was the Nature Conservation Trust (now the Biodiversity Conservation Trust) as a Geographic Information Science (GIS) officer. While working with the trust he undertook first a Masters in GIS and Remote Sensing, and then his Honours (Environmental Sciences) both part-time, and again with CSU.
Liam's PhD, which he began in February 2019, builds on his honours research in species distribution modelling.
For his study he intends to use virtual simulations to better understand current modelling methods, aiming to develop a valuable tool for environmental modellers and managers tasked with predicting into the future.
"Basically if you want to model where a species might possible occur in the future - say if you had a situation where a species like a cane toad could potentially invade a new area - then there are potentially new environmental conditions and processes occurring compared to its native range, that we don’t have data for," explains Liam. "What I am trying to understand is how well do our current modelling methods work in those sorts of scenarios.
“Because we can’t get data ‘from the future’ to validate our models, I’m looking at developing virtual ecosystems to generate virtual data from processes analogous to those in the real world. By using agent based models which represent individual animals moving through this virtual world with real world processes - dispersal limitation, reproductive rates, responses to environmental condition - their distribution will emerge. Then we can sample that simulated data and model it using our current methods to see whether or not we are actually getting the right results. We can actually validate our models properly and start answering some of the questions that we can’t currently answer because we don’t have enough real data."
“By understanding which models, which methods are more applicable in certain situations, and which evaluation methods are better at characterising the performance of those models, then that can feed back into actually developing informative models that can help managers' decision making to mitigate some of impacts of threatening processes like habitat loss and climate change.”
He expects the kind of people to use this new knowledge would be the people who do modelling in research centres or government departments.
“My research would, hopefully, inform their decision making in terms of developing models and then that information should then flow on to more informative models for land and environmental management,” he says.
Grimmett, L., Whitsed, R. & Horta, A. (2020) Presence-only species distribution models are sensitive to sample prevalence: Evaluating models using spatial prediction stability and accuracy metrics. Ecological Modelling, 431 https://doi.org/10.1016/j.ecolmodel.2020.109194
Bachelor of Environmental Science (Honours) (CSU)
Masters GIS & Remote Sensing (CSU)