I am modelling a single species too, if you want check my questions, there is a lot of answers about that

The bottom line here is that you have to at least have a big enough sample size to get out of the 'instability zone' that a small sample will give you. It doesn't sound to me like you have a large enough sample size - in any case you can only really use two or maximum three predictors with n=16 - so perhaps your time would be better spent doing more field surveys. (Refs for 'instability zone', see figures in this paper: Cumming, G.S. (2000). Using between-model comparisons to fine-tune linear models of species ranges. Journal of Biogeography 27: 441-455.)

Hey everyone, thanks for your replies.

Answering to you Graeme: unfortunately that is what my supervisor, and more recently I, have been doing since its discovery (2001). It lives in really remote gullies which take 1-2 days of walking to reach, and our success ratio is usually of 1 success every n unsuccessful surveys (n being quite large). Furthermore, we think we covered most of its range now.

Maxent automatically removes duplicates within a cell (unless you change this standard setting). You may wish to consider instead to check neighbouring cells depending on the spatial accuracy of your your presence points, the dates of two neighbouring presence points and distances this Gecko can move between the dates (see attached).

The removal of variables based on expertise introduces a bias.

Graeme's statement on minimum number of presence points is contradicted in the literature (see attached). It seems elegant and avoid over-fitting to reduce the number of environment variable step-wise. Mostly 1-3 variables is sufficient (attached).

Article Where the bears roam in Majella National Park, Italy

Article Transferability of species distribution models: The case of ...

Thanks Hein

I was wondering, is there any good paper discussing how variable choice can influence your model output?

Aren't we always going to have a bias regarding variables used?