Why would an avg map differ in two maxent models with the same settings?

[mjb...@york.ac.uk]

Hello Maxent group,

I'm hoping someone can shed some light on an issue I have just found for a side project I am working on. 

1) I model a species distribution in maxent for current climate to produce a distribution map (i.e. the _avg.asc file from the maxent output), I apply a threshold value to make a binary map. The map makes sense with my understanding of the species involved and all is well.

2) I then need to project into the future. So, I run the model again, same settings (exactly the same) but this time specifying the future climate directory. The output of this model run gives me a _avg.asc file and let's say a rcp45_avg.asc file. Now, I assume that the _avg.asc file from this output should be the same as the _avg.asc file from 1). 

I apply the new threshold value (but using the same threshold as in 1) but there is a big difference between the binary map from 1) and the binary map from 2). 

Am I missing something here with Maxent outputs. Does the _avg.asc in 2) relate to current climate distributions? 

m  

[Jamie M. Kass]

An idea: the only stochastic element in Maxent models is the random selection of background points. If you have a huge background extent, and you let the software randomly sample again with the default of 10,000 points, you may get a different prediction than the first time you ran it. This would explain what you're seeing. A way to avoid this is to make the random background selection before running, put the coordinates in the same csv, and run Maxent with the SWD setting.

Jamie

[subbuku...@gmail.com]

Hi Jamie,

When you say put the background coordinates with SWD setting does it mean we should put this csv in variable's folder ?

Thanks
kumar

[Jamie M. Kass]

It means that you supply the occurrence and background coordinates, plus the environmental covariate values for each point, in the same csv and input this into the Maxent GUI. There is some info on this in the Help document, and more online if you search.

Jamie

[mjb...@york.ac.uk]

Thanks Jamie,

That makes sense; the background extent is large (Madagascar) because of the nature of the question and how we are trying to manipulate the SDM. I think I'll opt to control background points and see how it goes.

[Samuel Veloz]

You are making a mistake in your #2 model. Rather than rerunning the model and specifying the future model directory, you should be running the model with the current environmental conditions and specifying the future climate directory in the "projection layers" dialog. If you didn't do this you are attributing your species presence and background data with the future climate data which makes no sense. If you used the SWD format as suggested and attributed with current climate than you should be fine also.

Sam

--
You received this message because you are subscribed to the Google Groups "Maxent" group.
To unsubscribe from this group and stop receiving emails from it, send an email to maxent+un...@googlegroups.com.
To post to this group, send email to max...@googlegroups.com.
Visit this group at https://groups.google.com/group/maxent.
For more options, visit https://groups.google.com/d/optout.

[mjb...@york.ac.uk]

Thanks Sam for your thoughts, perhaps I should have been clearer - the future climate was specified in the projection layers dialog, hence the two outputs of current climate response and future climate response.

cheers