Maxent taking so long time

[Ra Zoo]

Dear all, first time i am using maxent (3.4.1) by using worldclim 19 climatic variables, dem and land cover (all at 30m) for around 35,000 km2 area, and also predicting for 2050. But maxent takes more than 24 hours for single species distribution. Please suggest me ideas.... thanks 

[Cass Kalinski]

A lot of data at that small resolution so not a surprise on the time required. I had comparable results on an extent one third your size, also at 30m. One thing to check is the Threads parameter on the Experimental tab of Settings. Increase that number, with the maximum value being the number of cores available on your machine. If the only thing running is MaxEnt, use all of them. Then go take a long nap. 

If you use Maxnet (R) instead Maxent (Java), running in the Microsoft R environment, the time will drop exponentially. See the literature about maxnet for the trade-offs of going that route though. Also check the maxnet related threads on this forum for some issues that you might need to be aware of. 

The Microsoft R product multi-threads many raster operations. My initial testing several months ago was not exhaustive but I did see substantial jumps in performance with their R environment. Grates on me to say that as I am generally not a big fan of theirs. 

Reducing the number of climate variables is also recommended. Avoid collinearity in the variables. Use only those that are environmentally pertinent to your species. 

Cass