new open-source release of Maxent software

[Mary Blair]

Dear colleagues,

We are pleased to announce a new open-source release of Maxent software for modeling species niches and distributions, now to be hosted by the Center for Biodiversity and Conservation at the American Museum of Natural History (CBC-AMNH).

As documented in Phillips et al. (2017 Accepted in Ecography), an open-source release of Maxent now exists, including some updates. In addition to fully open-source code, now released under an MIT license, updates to the software include changes to the default feature types (threshold features are now turned off by default) and the addition of a cloglog transform as the default output format (which was formerly the logistic transform).

Also, the Maxent download site is moving to a new hosted server here.  Maxent will now be hosted by CBC-AMNH, which is where the idea for use of Maxent for modeling niches and distributions was first conceived. Steven Phillips and the other developers of the Maxent software are still engaged in its development and maintenance, and the google group will remain the main mechanism for user questions. We hope that these developments promote additional progress in the field, and encourage collaborations and independent contributions.

Please let us know if you have any questions, and thank you for your continued critical use of Maxent.

Sincerely,

Steven J. Phillips, Mary E. Blair, Robert P. Anderson, Miroslav Dudík, Robert E. Schapire 

[Jamie M. Kass]

Thanks Mary! Very exciting that Maxent is open now. I am also pretty thrilled that there is an R implementation of Maxent now (maxnet package).

I think we'll add the option of using maxnet in ENMeval soon. It will eventually be interesting to explore any differences between maxent.jar and maxnet and what effects they potentially could have on models.

And thank you AMNH for hosting the software! It's a very appropriate place for offering Maxent and its documentation.

Jamie Kass
PhD Candidate
City College of NYC

[Caitlin Glymph]

All,

I recently downloaded the new version of maxent described here. However when I change regularization parameters from the default, I've been getting very strange results or in some cases no results (even though I get no error messages during processing). For example, I ran a model with all the defaults except with a regularization of 3 and the output showed all my response curves as completely flat and the map was all solid green. The AUC  score graph was also just a diagonal line opposite the black line (forming an X). I don't understand why this is happening as I am using the same datasets as I did with the older version which I had no problems with. It will work for the default regularization of 1, but no other setting. If anyone has any insight as to why I am getting results like this ( again I received no error messages) I'd very much appreciate it! I'd like to use the cloglog setting for my final model but so far, no luck . 

Thanks!

Caitlin Glymph

Graduate Assistant

Stephen F. Austin State University 

[Pete]

Hi everyone,

Jamie was able to replicate this issue and we were able to bring it to Steven Phillip's attention. He released a newer version of Maxent (v. 3.4.1) that should fix these bugs (see commit notes for bug details on Github). See here to download the newest version from the Center for Biodiversity and Conservation at the American Museum of Natural History (CBC-AMNH).  

Cheers,

Pete G

Biodiversity Informatics Specialist

CBC - AMNH