Chapter 9 Develop Maxent Model
This chapter will describe the process used to configure and run the Maxent model (Phillips et al. 2017).

Figure 9.1: HREP Mussel Modeling Workflow, Run Maxent Step.
9.1 Convert predictors from .bil
to .mxe
format
The purpose of this step is to describe how to convert a raster format that ESRI products can read to a raster format the Maxent can read and write efficiently. To speed model creation (from 10’s of hours to 10’s of minutes), convert the predictors from the .bil
format to the Maxent .mxe
format.
- Open command prompt and change to the working directory. Change to the correct drive if necessary by typing
c:
. Then usedcd C:\Users\b6pdpdem\Documents\SteamboatMusselModel
to change to your local working directory. - In the working directory, create an
input_bil
folder. Copy the.bil
format predictor rasters created in previous chapters into this folder. - In the working directory, create an
input_mxe
folder.
java -Xmx5g -cp maxent.jar density.Convert C:\Users\b6pdpdem\Documents\SteamboatMusselModel\input_bil bil C:\Users\b6pdpdem\Documents\SteamboatMusselModel\input_mxe mxe
9.2 Run Maxent
The purpose of this section is to describe the process of configuring the Maxent model.
<
9.3 Convert model output from .mxe
back to bil
format
The purpose of this step is to describe how to convert a raster in the Maxent .mxe
format back to the .bil
format that can be read by ESRI.
java -Xmx5g -cp maxent.jar density.Convert C:\Users\b6pdpdem\Documents\SteamboatMusselModel\sb_existing_20181203_2 mxe C:\Users\b6pdpdem\Documents\SteamboatMusselModel\sb_existing_20181203_2 bil
- Use the
Define Projection
tool to set the projection of the.bil
rasters.
9.4 Maxent speed Issues
Follow these simple performance guidelines to reduce Maxent run time from a week to half an hour.
- Run the model on a physical computer, not on a virtual machine.
- Select a computer with the fastest processor and most memory available.
- Increase the memory available to the java virtual machine (JVM). Edit the first line of the
maxent.bat
file to something like the following:java -Xmx5g -jar maxent.jar
, where the-Xmx
parameter sets the maximum heap size for the JVM. Set it to something less than the total amount of physical memory on your computer. - Do not run the model across the LAN (inputs and outputs stored on a network file system). Copy all inputs and write all outputs to the local file system.
- Use a solid state drive for storing inputs and outputs.
- Use the “SWD” format for specifying both samples (species occurrences) and environmental layers (background points).
- Convert all predictor grids to the Maxent
.mxe
format. - Write all outputs to the Maxent
.mxe
format. - Set the
threads
parameter to at least the number of cores on your computer (or a large percentage of logical processors) to speed processing of replicates.
References
Phillips, Steven J, Robert P Anderson, Miroslav Dudík, Robert E Schapire, and Mary E Blair. 2017. “Opening the Black Box: An Open-Source Release of Maxent.” Ecography 40 (7). Wiley Online Library: 887–93. doi:10.1111/ecog.03049.