Your can use beta diversity, which is the difference in species between communities. The model is: β = γ/(α–1), where β= beta diversity, γ= gamma diversity, and α= alphadiversity. See: Wilson MV, Shmida A. 1984. Measuring beta diversity with presence-absence data. Journal of Ecology 72: 1055–1064.

I would recommend to use spatially explicit GLM (i.e. using environmental variables as well as spatial coordinates as predictors) fitted in a bayesian framework using the HSMC package (Ovaskainen et al. 2017). This package was designed to fit multispecies models but you can also use it to fit single species models.

More info about the package in:

https://www.helsinki.fi/en/researchgroups/metapopulation-research-centre/hmsc

https://github.com/guiblanchet/HMSC

As an alternative, you can also fit an occupancy model that explicitly model the probability of detection. But to do so, you need to have temporal or spatial replication of the presence/absence data for each site. A simple way to get spatial replicates is to set grid-cells as analysis units at a resolution that allows you to have more than one record of presence/absence. For example, if you surveyed your reptile in plots regularly distributed separated each other by 200m, and you analyse your data at a resolution of 1km2, then you will have several plots per grid-cell. Or you can use points of observation within a transect as spatial replicates (an example here: DOI: 10.1371/journal.pone.0025931). Then you can use a hierarchical Bayesian approach to estimate relationships between environmental predictors and both the occupancy and the probability of detection of the species. If this is what you need, then a good book chapter to read is (DOI: 10.1016/B978-0-12-801378-6.00010-2).

Hope it helps,

cheers,

D.-

I act as series editor for the books published by Cambridge University Press in the 'Ecology, Biodiversity and Conservation' series. Two of the books in the series are appropriate for your question; these are:

Franklin, J. (2009). Mapping Species Distributions: spatial inference and prediction.

Guisan, A., Thuiller, W. & Zimmermann, N.E. (2017). Habitat Suitability and Distribution Models: with applications in R.

I hope that these two books are of assistance to you.

It might be worth looking into MaxEnt, it's free and is extremely powerful, especially for assessing large distributions. You do have to input your own environmental variables, however.

You can use species presence points only in Maxent software. this is simple SDM.

Andrew Paul McKenzie Pegman , David E. Uribe R. , Michael Usher , Monica Matthews , Muhammad Asif , James R, Hatten ...

Thank you all for your valuable comments. I will surely go through what you all have directed.

Thank you ones again,

Cheers,

Debaprasad

A logistic LOGIT would be nice to fit the model if your observtions are in a form of presence- absence and a threshold independnt AUC-ROC can be used for evaluation. However for cryptic species, most of the time you may have "presence only" observation, then a MAXENT may work better, as it will model the probability of occurrence, but using the LOGIT as well most other regression models will show the prediction as relative suitability or likelihood of the habitat use.

Subhasish Arandhara ... thanks for your input