changes in distribution areas future and present

[nima...@hotmail.com]

Hello, 

I've run maxent for the present climate, and also projected for future climate scenarios. I'd like to calculate the changes in the area between the future and present distributions (using a cut-off for presence, naturally) and report, for example, species1 had a decrease in potential distribution by XX sq.km, in the RCP45 scenario compared to the present-day distributions, while species2 had a slight increase by YY sq. km compared to the present day. 

I have the ascii files, and I've converted the ascii files to a raster using the "FLOAT" option, so in theory, I should be able to do calculations. I also reclassified the raster, and have the counts for each. But I'm unsure as to whether this is the correct method. If it is, how would I proceed to convert the count to sq. km and calculate the difference, and if it's not, what's the best method to obtain that information? 

Thanks very much, 

Nima

[Jamie M. Kass]

The most often used method is to make your continuous predictions binary by thresholding them using some value, below which the cell = 0 and above which the cell = 1. The Maxent output html will give you some suggestions for thresholds, like the minimum training presence and the 10% training presence, etc. There is no "best" threshold, so you need to experiment to see which leads to the most realistic prediction for potential distribution for your species. You might also want to look at omission rates for each. The binary raster will let you calculate area (if you know the cell size, just square it to get the area, assuming your cells are all squares). Hope this answers your question.

-Jamie