Hello Maxent Community,
I have been generating SDM models and projections in Maxent. I have had some success with one set of variables (some Hydroshed variables mixed with World CLIM variables); however, I am having a bit of trouble with a model based solely on World CLIM variables.
I used a Peasron correlation matrix to select eight World CLIM variables, which model well with my species' occurrence data. The issue arises, when I ask Maxent to project on future World CLIM variables - the returned projection is almost entirely blue. There are a few specs of red and yellow (indicating prediction probabilities), but the vast majority of the projection is blue (indicating zero prediction probability). I tried rerunning the model with clamping turned off, but the problem persisted.
I am relatively novice with Maxent, so I was wondering if I am overlooking a setting that may help correct this issue. I am a bit confused, because when I ran the projection with the Hydroshed variables and some of the same World CLIM variables, the future projection looked just fine. Any advice or suggestions would be tremendously appreciated.
Thank you in advance for your help!
Dear Carter,
I am not sure your trouble is fully understand by myside, but might because of the predictor selection as you described.
Can I ask where did you perform MaxEnt model? If you used R package like 'dismo' for modeling, if you like, then you can send your codes to me for review; otherwise, if you perform MaxEnt with software, you can send some screenshots to me for checking the setting.
Hope this helpful.
Best wishes,
Ziyan Liao
Ziyan Liao, thank you for such a quick response.
To answer your question, I am using Arc Pro from Esri and the Maxent application (version 3.4.4) to make my models. I am taking rasters (ASCII files) of each variable, clipping them to my study extent, then feeding them through maxent with my occurrence data to generate the models.
I am attaching pictures ("1" and "2") of my current Maxent settings.
I am also attaching a picture ("3") of my functioning model (top map in the picture) and the problematic blue projection of future variables (bottom map in the picture). As you can see in the future projection, there is some light wisps of color in the far North-East, so it appears the projection is at least being made without any major errors.
As a side note, I've played around with selectively using only certain regions of the occurrence data for modeling and noticed that certain selections will generate more normal looking future projections, while others will produce mostly blue maps, like the one in Picture 3.
Again, thank you for your time and help with this issue.
Dear Carter,
I can not find out what is going wrong with this information.
But I would suggest you:
(1) Re-check all predictors' names for current and future periods and use the same names - remind that, because if not, you can not run it successfully when projecting to the future.
(2) It is important using replicates when running the MaxEnt model. However, the number of replicates of the model depends on whether stable results can finally be obtained and when the model would be converged. Some researchers used at least 10 replications, and some suggest using equal replications to the number of sample points. In such a case, the final average model is both trained and evaluated with all the points.
(3) Select the 'Append summary results to maxentResults.csv files ' checkbox, by which you can get a final summary of all model replications.
(4) Change the Maximum iterations number to at least 5,000.
(5) Also, why you use 'Regularization multiplier' as 4, do you test by R package ‘ENMeval’?
Best wishes,
Ziyan Liao
Dear Carter Cranberg
Current and future have the same names and projection?
if yes
try to make it 10 replicates with 25% test
Dear Carter,
I don't know the exact species but i am sure a crayfish will fancy water... did you mask your rasters so only water bodies remain? i would also (or instead of worldclim) use water variables, such surface temperature, flow rate, etc. or at least freshwater distance and proportion of a cell covered by (fresh)water.
Cheers, R
If the proplem still exist try other platform like openmodeller:
this is a simple start manual
Method Ecological Niche Modeling - openModeller User Manual: a simple start
Dear Carter Cranberg,
I recommend reading this article:
Predicting the impacts of climate change on the distribution of species: Are bioclimate envelope models useful?
Best wishes
Bekhruz
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology