FIGURE 17 in Late Messinian mollusks and vertebrates from Moncucco Torinese, north-western Italy. Paleoecological and paleoclimatological implications

FIGURE 17. Variation of coefficients of the predictor variables of the Binary Logistic Regressions that model the probability of finding a specific murid taxon at MCC assuming the different layers as the only predictor variables. Blue dots indicate the values of the coefficients, bars indicate the 95% CI. MCC3 is chosen as reference category. For each of the identified taxa, the coefficients that are lower (or higher) than 0 indicate how much, respect to MCC3, that variable influences the probability (in terms of log odds) of finding that species at MCC. In other words, lower (or higher) coefficients of each variable (i.e., the given fossil assemblage) indicate that for that assemblage there are lower (or higher) probabilities of finding that species respect to MCC3. Significant statistical differences are obtained when the CI does not include the value 0. On the contrary, values approaching zero indicate that in that assemblage there are the same probabilities of finding that species that we have for MCC3. Stephanomys sp. is not included due to its rarity and not signifacant values. Additional information available in Appendix 8.