I would like to map the current distribution of an insect using MaxEnt based on 15 samples, should be published in a small entomological journal. Is it recommendable to split the data for training and testing with such small sample size? I did a little inquiry how to optimize MaxEnt models and found, for example, Shcheglovitova & Anderson (2013). I believe applying this to my modest case is like cracking nuts with a sledgehammer, are there any simple ways (and tutorials explaining such) for people like me?
Any help would be highly appreciated.
Robert
Hi Roger,
Yes i would agree with you, there are simple tutorials for beginners to start using maxent. I would recommend starting with the manual available when you try to use the software version of MaxEnt available at: https://biodiversityinformatics.amnh.org/open_source/maxent/
I would also recommend you read the publication on Merow et al. 2013 and Elith et al. 2006.
In the matter of splitting you data, I found when modelling a species with a small range, its best to selected the 'cross validate' function and replicated the model the same number of times as the number of occurrence points. That way the model using one point to evaluate each model. For example if you have 15 points, replicate the model 15 times using cross validate function. that way the average model is evaluated on 15 points and modelled on the 15 points as well.
Read my publication for mode details on this. Article Applying species distribution modelling to improving conserv...
The mentioned references are all referenced in my publication as well.
Dear Roger,
Try "a statistical explanation of MaxEnt for ecologists - Elith - 2010 - Diversity and Distributions - Wiley Online Library" and let me know if it works with your study case.
Best,
Guy
- Badges
- Science topic
- Similar topics
- Biological Science
- Zoology
- Invertebrate Zoology
- Arthropods
- Insects
- Insect