How to perform species distribution modeling using the software Maxent

[@jeff3523]

Thank you! Your tutorials and associated materials from your website are an incredible resource for the biology community.

[@DrJBanta]

Thank you so much Jeff. It means a lot to get that feedback and to have your support.

[@MJ-ye7dd]

Thanks Prof. Josh for making your very useful tutorials accessible on your lab website. It matters.

[@briannjoroge2344]

this is very helpful. I just started learning how to use Maxent

[@deepitkotian]

Thanks for this , did really help, please dont delete your videos

[@fatimaawadalla9264]

Hi prof thank you for good explaining . this session it's very helpful .

[@surendi87nisha38]

Thank you very much for detail explanation !!!

[@Pritam_Chhetri_31]

It was very helpful, thanks a lot. I am just starting to learn maxEnt.

[@katrinaluzon1487]

Tutorial is very clear

[@Madhushree.VMadhu]

It was really very useful thing, Thanks Prof. Josh. i have tried it and almost got the result but the thing is i have got it for global but i needed it for particular region. i have seen your maxent settings video also but i couln't figure it out can you please suggest which settings or protocol should be followed to obtain for particular study region. thanks in advance

[@danielreis4709]

Great job. I have a doubt about the choice of variables that I will test the correlation. I shouldn't be testing the correlation of the nine variables, right? Which variables do I select to test the correlation and then decide which of these to use in modeling? Example: I want to model a plant. I have the 19 variables available from Wordclin. But as I select within these variables I will test the correlation.

[@centerforanimalecology268]

Excellent tutorial. I recommend you to please provide more detailed insight on how to choose the best model if we have multiple replicates.

[@aliyahbhatbhat8458]

Hi sir. When i open maxent after download, there opens an WinRAR (evaluvation copy) from which i am not able to drag maxent jar into maxent zip file that we have on desktop. Please sir help.

[@fabiosavini5481]

Hi Professor! I saw your Max Ent tutorial because I have to do a habitat selection work by the wolf in the Po Valley (Italy). Now I have collected and processed my environmental variables but I miss the last step before putting them on Max Ent. Now comes my question: my environmental levels must be in ASCII format in order to be processed on maxent, right? Can I keep my Ascii files separate maybe in a single folder or do I have to create a large and single multilevel ascii file that contains all my environmental layers?

[@DrJBanta]

Dear Fabio, I am glad the video has proved useful for you. I would recommend keeping the environmental layers separate in a single folder. You cannot use a multilevel ascii file with Maxent to my knowledge. Best regards, Josh

[@jinsvj2387]

Can u provide link for the previous tutorial mentioned in this video?

[@TheRockaleo]

Hi Sir, thank you for this tutorial, where can we find the output of the MESS analysis?

[@manuarteteco6153]

Thanks for this tutorial. I understand that in the html output the left map is the distribution, but what's the map on the right then?

[@DrJBanta]

It shows the standard deviation of the suitability scores, in other words the variability of the replicates.

[@manuarteteco6153]

@@DrJBanta I get it, thanks for the comment and for your videos, very helpful!

[@irham9321]

Dear Dr Josh Banta, I tried to run the maxent. But the error 'no background points with data in all layers' popped up. The result from the enmeval is okay. All the layer have same extent and projection. And can generate bias file. I am not sure what is the problem now. Anyway very good tutorial.

[@whitneyholeva2090]

Hi Josh! Thank you for another awesome tutorial!! When I run Maxent, I get a popup box warning "Bias has zero or negative values, ignoring bias grid." I tried searching the internet for this error but didn't find anything. Do you know if this means that it is not using the bias file? I created the bias file using the code from another of your tutorials, so I'm not sure where things went wrong. Thank you again for putting all of these videos out!

[@DrJBanta]

Dear Whitney, this is because some of the values in your bias file were so small that they were estimated to be zero. Change the zeros in the bias file to 0.00001 or something like that and it should work. You can change the zeros using R. If the R object containing the bias file were called "biasfile," then the code would be: biasfile[biasfile == 0]

[@loicpittet1897]

@@DrJBanta Hi ! Thank you for the tutorials, they are amazing :) I'm having the same problem. Unfortunately replacing the zeros was not successful. In fact all the values in the biasfile are really small. Do you have any suggestions ?

[@DrJBanta]

@@loicpittet1897 I would add 0.1 to all of your data in the bias file (except the -9999 entries). The problem is that R is writing the numbers close to zero as zero.

[@sitikhadijahsrioktoviana53]

Thank you very much, Prof. Josh, for giving us an explanation about how to use the MaxEnt model. This tutorial is very helpful. Could you give us some tutorial about how to get enmeval_result?

[@sitikhadijahsrioktoviana53]

and how to get biasdata Prof?

[@DrJBanta]

@@sitikhadijahsrioktoviana53 You can obtain both files by following this tutorial: kzbin.info/www/bejne/fZDbhXqOituMjtEsi=ZK76gdmh1tuRrnaM

[@kukilunbaite6654]

How can I calculate distribution area in qgis... Please help

[@ibrahimhegab2132]

Excellent tutorial. we appreciate your efforts but we would like a little bit explanations for the significance of the results if you do not mind please 😊

[@dishaadak206]

This was really helpful. Can you please tell me that how to choose training data and test data? does the maxent software choose it by itself?

[@psyclistt]

Prof. Josh, I am not able to get install maxent software plz help me through

[@priyanwadaatapattu5900]

In the previous tutorial l could not run raster files. What should I do for that

[@MansiChauhan-wr8ec]

how can i prepare corelation matrix in arcmap

[@luisangelbarrera-guzman7463]

How calculate ‘true skill statistics (TSS)’, to corroborate model performance?

[@samrusaati9173]

Dear Professor, should I choose either SUBSAMPLE or BOOTSTRAP as the "leave-one-out method"?...... I have only 22 geographical data

[@DrJBanta]

Dear Sam, use "CrossValidation" but set the number of replicates to equal the number of occurrence records in your dataset. This will make it the "leave-one-out" method. Best, Josh

[@kalthumomer220]

good morning prof.Josh. thank you for your useful tutorials. whene I run the models alway gave me the warning ; Reseting random test persentage to zero because cross - validation in use. how can I fix thi, I appreciate your answer.

[@DrJBanta]

Dear Kalthum, the problem is that the methods for splitting the data into test and training sets are not compatible with the cross-validation method. If you want to use cross-validation, be sure to set the number of folds to 10. I recommend cross-validation over manually assigning test and training data percentages.

[@kalthumomer220]

@@DrJBanta thank you .

[@mkklindhardt]

Hi Josh, Thank you for this tutorial. I was going through your educational materials on your website. For some reason the link is broken for the link under "How to adjust your environmental layers so that they have exactly the same spatial extent and resolution (which is necessary for species distribution modeling)"

[@DrJBanta]

Dear Kamau, I have fixed this error. Thank you for alerting me. Best regards, Josh

[@mkklindhardt]

@@DrJBanta that's great! Thank you a lot for putting this available to the world. Amazing help and inspiration! Continue spreading the knowledge.

[@jincymathew1516]

Thanks Prof. Josh for the tutorial. I have followed each and every step, but after clicking "run" I am getting an error which says my biasfile and my bio1.asc (which I previously used to generate the biasfile) have different geographic dimensions. First of all I have checked for projection errors and projected the bias file into same projection as of bio1.asc using arcgis. Then I have followed your tutorial on "How to adjust your environmental layers so that they have exactly the same spatial extent and resolution". But the error still persists. Kindly help

[@DrJBanta]

Hello, I think you're using old versions of your files by mistake. Be sure you're using the right files. Open them and manually look at them to make sure they have exactly the same spatial extents and xy corner, and number of columns and rows. Then put all those files onto a separate hard drive or thumb drive. Then run Maxent using those files on that separate hard drive. Good luck, Josh

[@lalalaura8241]

Hi Prof. Banta! Thank you so much for the great tutorials on your website! Even though I currently have a problem running maxent with my own data and thought maybe you could help me. So first I have found out my settings using ENMevaluate like you showed in the previous tutorial. I also used a biasfile which I want to use for running maxent like you show in this video. Now everything works until I click on „run“ in maxent. I get an error („error reading file“) that says that my biasfile and my bio1.asc (which I previously used to generate the biasfile) have different geographic dimensions. So maybe you have an idea what I did wrong? greetings from germany :D

[@DrJBanta]

Dear Laura, Hello and nice to meet you. The easiest way around this is to perform an extra step to ensure your rasters are the same. Include the bias file along with your layers when you follow the steps in this tutorial: kzbin.info/www/bejne/l2e6aIqOhdlmaqM

[@lalalaura8241]

@@DrJBanta Thank you so much! I will try this :)

[@DrJBanta]

@@lalalaura8241 You're very welcome. Good luck. -Josh

[@garryoca3294]

Will maxent still works despite there is a slight difference in the extent of some data?

[@krishnarajmv9816]

Dear Sir, if the number of occurrence is so limited (ie. only two) is it possible to use Maxent?

[@DrJBanta]

I'm afraid not. Have you tried finding more occurrences for your species on iNaturalist?

[@vanuhihambardzumyan1579]

Dear Josh Banta, Thank you one more time for an amazing tutorials! I came to one issue, which I would like to address. I tried building model for 2 species separately, however, for 1 of them MaxEnt says bias file has 0 or negative values and does not accept it. I tried dens.ras2[dens.ras2 == -Inf]

[@DrJBanta]

Dear Vanuhi, your code contains a typo. It should be

[@bryanayres7664]

Hi, I have followed the directions from this tutorial and the tutorial to create a bias file and determine the settings to use in Maxent on my own data. However, when I run Maxent I get an error message reading "Error reading file (biasfile location): java.io.IOException: Sample file needs three columns: species (column 1), longitude (column 1), latitude (column 2)". Do you have any suggestions to remedy this problem? Thanks!

[@bryanayres7664]

Disregard the previous comment, I have figured it out.

[@ronnconcepcion6558]

Hi! May I know how to create the enmeval_results file for my own data? Please advise. Thanks.

[@DrJBanta]

Dear Ronn, Please see: kzbin.info/www/bejne/fZDbhXqOituMjtE Best, Josh

[@spastinator]

If I had

[@DrJBanta]

Crossvalidation. Be sure to set the number of model folds to the number of occurrence points in your dataset.

[@spastinator]

@@DrJBanta Thank you!. Why do you use fewer model folds (i.e. the 10 you use here) when you have more occurrence points?

[@DrJBanta]

@@spastinator When you have a lot of occurrence points, you can divide them up into a few groups, with one group being left out each time the model is iterated. When you don't have a lot of occurrence points, however, you cannot divide the model up into a bunch of groups. So instead, you leave only one single point out every time, and run the model with all the others, and repeat this, until each point has taken a turn being left out. In the Maxent settings, this is how you set it to do that.

[@spastinator]

@@DrJBanta Thanks a lot, that makes sense, and thanks for these tutorials, they are very helpful!

[@suryahafizh5498]

thank you very much

[@VitorOliveira-fe7yq]

Hi professor! Thank you for the tutorial it's awesome! i'm not sure if i missed it in some part of the video but, when did you checked for correlations among the environmental layers?

[@DrJBanta]

Dear Victor, You can learn how to do that at my website: sites.google.com/site/thebantalab/tutorials. Scroll down to "How to check for correlations among environmental layers and remove ones that are highly correlated with each other." It's near the bottom. -Josh

[@VitorOliveira-fe7yq]

@@DrJBanta Thank you! There's anyway to get a better resolution of the image generated by the maxent results?

[@DrJBanta]

@@VitorOliveira-fe7yq Dear Victor, the resolution of the results is a direct output of the resolution of the input layers. The only way to improve the resolution of the results is to improve the resolution of the environmental layers. See: sites.google.com/site/thebantalab/tutorials. Scroll to "Tutorials for using Geographic Information Systems (GIS) in the software GRASS GIS." and then find "Rasters and raster manipulation."

[@ethiopianationnationalitie5043]

Hi Josh Maxent error: Error projecting, two layers have different geographic dimensions. but I have checked on GIS that they are same location please can help me ?

[@DrJBanta]

Hi, what GIS program were you using? I suspect your layers are in different projections. See the tutorials on my website, joshbanta.com, for help with this.

[@dr.farazakrim2239]

superb very helpful

[@amnabatool6597]

Dear Josh Banta, This tutorial is very helpful for me. I have built the current model successfully but when I tried building the model for the future a pop-up box appears such as Error reading file ..java.lang.NumberFormatException: For input string: "1 QNAN". Can you plz help me in solving this issue?

[@DrJBanta]

Dear Anna, thank you very much for the positive feedback. Do you have your future layers named exactly the same as your current layers, and do you have them all in the same folder that is different from your current layers? Best, Josh

[@amnabatool6597]

@@DrJBanta Yes I did the same for the future as I did for the current layers but still this error exist.

[@DrJBanta]

Start by manually re-naming all of the future layers to make sure they exactly match the names of the current layers (make sure there's no hidden spaces or anything). Next, make sure you have all the future layers in the same directory, and be sure you do not put any other layers or files in that directory. Do not put the layers into a subdirectory of that directory. Then see if the problem persists. Good luck! -Josh

[@DrJBanta]

Try opening the future layers in a text editor to see if everything looks in order.

[@yehtetlwin5468]

Could you please support a tutorial about how to prepare environmental variables in ArcGIS for Maxent which are not direct downloads from the website like distance from the village, elevation from GPS data and so on.

[@DrJBanta]

Dear Ye, you may find the other tutorials on my website helpful. I have some designed for ArcGIS, but I have others designed for "open source" (free) programs, such as GRASS GIS. joshbanta.com

[@biplabdas6980]

Hello Dr. Josh, i have gone through the tutorials of yours, one thing I want to ask you, how to go for modeling for the future climatic data. Thank you.

[@jalalkassout4226]

Hi bibpab, you can use the same procedure as for current climatic data. Here a paper as an example : Gilani et al. (2020). Under predicted climate change: Distribution and ecological niche modelling of six native tree species in Gilgit-Baltistan, Pakistan (Ecological Indicators)

[@saharaeducation425]

@@jalalkassout4226 Hi sir can you please mention your email address?

[@عبداللهمحمد-ش8ق1ص]

Url where?!!!

[@SehrishBashir2218]

Hi... I need some help please reply

[@DrJBanta]

Hello, what issues are you encountering?

[@baktimandala]

love it

[@star-ph6nn]

thank you!!!!!!!!!!!!!!!!

[@lynxers9545]

My name is Maxent

[@GlobetrotterGab]

Mate it is a decent video but way too much time spent explaining obvious things and 0 time spent explaining actually useful things, such as omission rate.

[@DrJBanta]

Dear Gab, The fact that you are interested in omission and commission shows that you are at an advanced level. This video is intended for people who need to develop a basic conceptual foundation first. If there are no good videos about those topics, then I encourage you to make them! Best regards, Josh

[@Assasinsweed420]

I got a question for you professor. Do i need a powerfull PC to run the algorithm? I'm a latinamerican student trying to apply this to archaeology and i wondering if my PC will manage to do it because i've got to use a lot of variables

[@DrJBanta]

Hi I think you'll be okay with what you've got!

[@YzabelleLacambra]

Hi sir. When I open maxent after download, there opens a WinRAR (evaluation copy) from which I am not able to drag maxent jar into the maxent zip file that we have on the desktop. Please sir help.

[@DrJBanta]

Hello, download and install the free program 7zip to use to open the Maxent file. www.7-zip.org