So I've built ecological niche models and then species distribution models for the past 3 years preferentially using Maxent. I chose Maxent based on Elith's 2006 Ecography review paper of the different statistical software programs available. Recently, I felt I was becoming too reliant on Maxent and decided to reevaluate the ecological theory behind my modeling choices. This is one of many reasons why I started our species distribution modeling reading group at UT along with Dan Warren. Additionally, the SDM literature is overwhelming in size! it's nice to have a forum to meet and discuss relevant papers we stumble across. There is no way to read every SDM paper published...there are just too many these days!
As a graduate student I've completed all the classes I need for my PhD but I'm a sucker for interesting classes where I think I'll learn new and applicable information. I also hate paying tuition money to sign up for research hours when I'm going to do the research anyway! So this spring semester I signed up for a graduate level course on ecological modeling with a focus on species distribution modeling offered by our geography department. The course is being taught by Dr. Jennifer Miller whose PhD advisor was Dr. Janet Franklin. The course is much more theory focused but also has some fun activities. I'm just happy to have another place where I can talk about SDM!
One of the activities of the class is a SDM competition to see who can build the best model. I don't know if the data we will be given is going to be simulated or real...but I do wonder how it will be evaluated. So one of my goals on this blog is to keep y'all updated on the competition and my different ideas for how to best model the species occurrence data I will eventually receive.
Author
Stavana Strutz is a doctoral candidate studying disease ecology in the Parmesan lab at UT Austin.
Comments
Dan Warren
01/28/2012 14:30
I love the contest idea, it's such a great way to get people invested in thinking about their modeling decisions and trying to get good results out of their models.