You need to apply binary models then. Logit and Probit models are good in this case.

Besides, you can apply some non-parametric techniques that may be good to compare between the groups you have. For example, you can apply kernel densities. This helps you to compare microsites.

You can also categorize your microsites by defining: Small, Medium, large....and here you can apply ordered logit model

You could try generalized linear models, regression analysis that can also incorporate binary, categorical and ordered predictors. For the case of binary functions you will need a link function such as probit or logit.

The strength of this approach however lies in its capacity to integrate predictors from different formats (metric, categorical etc.), I am thus unsure whether this is the best approach but it should work.

Thank you so much! I will try with GLM. But I forgot to say that every monument does not contain the same number of "positions in which can be found a plant" also called microsites. So there are monuments which contain many microsites (eg. plants growing on a vertical walls, or on small holes, or in breaks of the wall, etc.) and others which contain only few of them. So it can be not easy to compare microsites if some of them occur more than others.. Do you confirm? Do you have any recommendation for this? Thanks

You need to apply binary models then. Logit and Probit models are good in this case.

Besides, you can apply some non-parametric techniques that may be good to compare between the groups you have. For example, you can apply kernel densities. This helps you to compare microsites.

You can also categorize your microsites by defining: Small, Medium, large....and here you can apply ordered logit model

Thank you so much! Last question, would you have some guidelines of how to perform logit, probit and kernel densities? Thank you in advance

Dear Giulio Zangari , I can think of two possible solutions to your problem with varying numbers of microsites. However their merits are dependent on your research questions.

1. You might want to include the number of microsites in your analysis and could check in a first step whether the number of microsites is related to some other important variable or use it in your predictive model.

2. You could discard all data of monuments with 1 or less microsites because in these monuments informatiojn about positions should be useles. For monuments with more than one microsite, you could determine the relative position such as "plant in top/lower half" of the monument (I however do not know what exactly you mean by position)

Sven Greving Thank you so much! Sorry for not have been enough clear to explain. I try again, showing you an example of my data.

monuments substrate microsite fum_parv all_atr falc_vulg (others 144 sp)

PP BS I 0 0 1

Ca BS I 0 1 0

Ca BS D 1 0 0

Ca BS D 0 1 0

CYT GGS M3 1 1 0

PR GGS M3 1 0 1

PR GGS M3 0 0 1

(others 75 data down here..)

144 species. 1 means present, 0 means absent.

9 monuments , 2 substrates, 10 microsites

Microsites is not properly the position (sorry) but the habitat where plants live in the monuments: eg. D means vertical fissures, M3 means holes in which there is some soil, M4a means the angles between two walls.

Some monuments can have more types of microsites than others but also the same monument can have a certain number of microsites of the same type

I hope you can understand better now

Thank you so much for your help!