I am new to this informatics field, and this is the first time I have tried Maxent for habitat suitability modelling. Now I want to calculate the area of highly suitable, moderately suitable, suitable and not suitable regions.
Zarreen, you can upload the MaxEnt output into R and calculate the number of pixels above or below a specific threshold as you define highly suitable/ not suitable habitat. Then, multiply the number of pixels by the area of your grid cells.
Hello Zareen. As Megan mentions above, the calculation can be done either in R or in any GIS software. MaxEnt produces and ASCII file in the results folder which can be read by GIS software or R. In a GIS, you could reclassify (e.g. r.reclass command in GRASS) for values above and below your threshold. You can then use GIS commands (e.g. r.stats in GRASS) to calculate the number of cells in each category. Ultimately however, the toughest decision is to decide what threshold to use. There are different opinions on this...Some go with the 0.5 value of logistic regressions. Others use arbitrary values such as 0.7. A quick look at Janet Franklin's book "Mapping Species Distributions" (Chapter 9 - Model Evaluation) shows a whole list of criteria used to decide the suitable vs. unsuitable thresholds of SDMs using various measures of the model's sensitivity/specificity. The article by Liu et al. 2005 ("Selecting thresholds of occurence in prediction of species distribution" Ecography 28:385-393) may help. Also, see Freeman & Moisen, 2008 ("A comparison of the performance of threshold criteria for binary classification in terms of predicted prevalence" Ecological Modelling 217:48-58).
Binary maps (suitable/non-suitable) can be created in Maxent from the probabilities of occurrence by choosing a threshold rule (see Liu et al. 2005).If you using Maxent try the 10% percentile presence threshold rule which allow for georeferencing errors (Morueta-Holme et al. 2010; Kaky and Gilbert 2017). Maxent does not naturally give an average binary map over all the replicates, and so this you can di it manually by allotting ‘presence’ to a pixel that had presence values in more than 50% of the model runs (i.e. >5 replicates). The process you can do it by using the Raster Calculator of ArcGIS