Anthony Bain put it very clear: it means your data do not follow a null model. I would just add that, after this conclusion, you may consider some forces that shape this pattern you found. The null model does not say which process has shaped this pattern, but maybe you have some information from your work.

Thanks Anthony Bain and Ariel Reis. I ran a different test based on a probabilistic calculation (Veech 2013) with my data, and it coincides with the results of the null models and with what you suggested, in the matrices there are species negatively associated and others occurring together at the same time. I am now looking into the possible reasons behind this pattern, but it might seemthat habitat affinity has something to do with it.