deletions | additions       diff --git a/materials_and_methods.tex b/materials_and_methods.tex  index 36e60b7..d67e1cf 100644  --- a/materials_and_methods.tex  +++ b/materials_and_methods.tex 
...
We faced three different issues to evaluate our models. First, the lack of absences made it impossible to evaluate the commissión error. Second, the low amount of presences prevented the use of data splitting to evaluate omission errors. Third, quasibinomial GLMs in R do not provide AIC values, making difficult to rank the candidate models according to both model fit and complexity. To deal with these issues while providing a robust model evaluation framework, we used a leave-one-out approach to compute two different measures:  \begin{itemize}  \item AUC (ROC curve, see Fielding and Bell...) values as an extrinsic measure to evaluate relative omission errors (CITE PHILLIPS). The AUC values were based on 164 168  pseudoabsences not used to calibrate the models. Pseudoabsences were separated 200 km from each other to reduce pseudorreplication, and separated from the presence records the same distance to avoid an artificial reduction of the AUC values (see appendix for further details). AUC values were computed as the proportion of pseudoabsences with an habitat suitability value lower than the habitat suitability of the test presence. \item Adjusted explained deviance as an intrinsic evaluation measure to assess model goodness of fit and complexity. This measure was computed for each model following this expression: 1-((cases - 1)/(cases - predictors))*(1 - ((null deviance - deviance) / null deviance)).  \end{itemize}