I think that performing a species distribution model (SDM) with only a few distribution points can be challenging because it may lead to low accuracy and high uncertainty in the predictions. However, If you are looking more information about these models I suggest:

- Naimi, B., & Araújo, M. B. (2016). sdm: a reproducible and extensible R platform for species distribution modelling. Ecography, 39(4), 368-375.

- https://www.youtube.com/watch?v=83dMS3bcjJM (Demonstration of the R code)

- Araújo, M. B., Anderson, R. P., Márcia Barbosa, A., Beale, C. M., Dormann, C. F., Early, R., ... & Rahbek, C. (2019). Standards for distribution models in biodiversity assessments. Science Advances, 5(1), eaat4858.

I hope this will help you!,

In our research referred below, ten (10) samples were taken as the minimum for SDM even for modelling with maxent, known for its relatively good performance with low sample numbers.

Although not reported in the paper, we tested modelling for species with sample numbers between 5 and 9. For some we reached our chosen threshold (AUC > 0.8), but not for all.

It could also be important for a good model output, to use the entire range of fundamental predictor variables with contiguous empirical values (e.g. elevation/DEM), not only say predicted climate.

What I would do in your case, is to carry out the modeling with your five samples and see whether the distribution area is much larger than the current distribution area. If so, survey the predicted area for unknown occurrences.

I assume you deal with endemic plants If your endeavor refers to (mobile) animals, please refer to our bear and panda papers.

Article Fine resolution distribution modelling of endemics in Majell...

I can strongly recommend the book 'Mapping Species Distributions: spatial inference and prediction' by Janet Franklin (2009). Also 'Joint Species Distribution Modelling with Applications in R' by Otso Ovaskainen & Nerea Abrego (2020). Both books are published by Cambridge University Press in the 'Ecology, Biodiversity and Conservation' series.

I would avoid projections, but if they can be verified with real data in the field, why not?

I share a couple of examples of mine

Article Using species distribution models at local scale to guide th...

Article A practical method to speed up the discovery of unknown popu...

We recently published a paper that goes in this direction: Article ENphylo: A new method to model the distribution of extremely...

Hello Bikram. I believe that with a low number of real and well-confirmed presences in combination with good methodological decisions, you can get models with an acceptable or good performance. Below, I give you a link of our tutorial of rare species from our R "flexsdm" package. This is an example of modeling only with 21 presence locations. We have tried species with fewer occurrences and have had good results as well.

link

Hello, certainly it's too late to answer but do you know the technique of Ensemble of Small Models (ESM) for rare species?

Scherrer, D., Christe, P., & Guisan, A. (2019). Modelling bat distributions and diversity in a mountain landscape using focal predictors in ensemble of small models. Diversity and Distributions, 25(5), 770-782.

Breiner, F. T., Nobis, M. P., Bergamini, A., & Guisan, A. (2018). Optimizing ensembles of small models for predicting the distribution of species with few occurrences. Methods in Ecology and Evolution, 9(4), 802-808.

You can used the ESM method with the R package "ecospat":

https://cran.r-project.org/web/packages/ecospat/index.html