Transforming the Maxent output maps

828 views
Skip to first unread message

Víctor

unread,
Jan 25, 2018, 8:18:12 AM1/25/18
to Maxent
Hello, I'm working with Maxent for modelling the habitat suitability for several species. 

When I obtained the output, I see that the maps are represented in a default scale of multiple colors expressing more or less suitability. I'm not sure if using that raw result in a paper is correct or I should establish specific ranks of suitability based in some parameter. 

If  for example I would like to represent the outputs just as presence-abscence, how should I decide the size of each rank? Would it be from 0 to 0.49 for absences and from 0.5 to 1 for presences or there is another protocol to decide the thresholds?

Thanks
Message has been deleted

Weverton Carlos

unread,
Jan 26, 2018, 3:21:51 PM1/26/18
to Maxent
Hello Víctor,

I think the choice of output format is related to question of your research. If you just want to compare the distribution between species, using the same background, Raw output is fine.
If you are using different backgrounds, or if you want to find how much a given area is suitable (or unsuitable) to a specie, you should choose Logistic or Cloglog Output. But be careful. When the Maxent performs a logistic transformation of the raw output, these transformations take into account the prevalence of the species. Logistic output uses a default prevalence value of 0.5 (you can change it in Default Prevalence), and Cloglog a default prevalence value of ~0.632 (you can't change it). If the species are very rare or very common, this values may not be appropriate.

The Maxent itself calculates threshold values that allow to make binary maps, and these values tend to be very lower than 0.5. There are several threshold values, some more conservatives and others more flexibles. The choice of appropriate threshold also depends of the questions of your research.

You can apply the threshold values and change the colors using a GIS aplication, as QGIS or ArcGis.

Best regards,

Weverton C. Trindade
Reply all
Reply to author
Forward
0 new messages