The optimal sampling design for litoral habitats modelling: a case study in the north-western Mediterranean [dades de recerca]

Share
The data base contains: Projected coordinates, environmental variables and the presence/absence (1/0) of each habitat are presented for each point. Slope code: 1 = 0º-10.8º; 2 = 10.8º-22.8º; 3 = 22.8º-45.1º; 4 = 45.1º-68.2º; 5 = 68.2º-87.8º. Habitats code: Riv = Rissoella verruculosa; Lby = Lithophyllum byssoides; Tro = Lithophyllum byssoides rims ("Trottoir”); Neo = Neogoniolithon brassica-florida; Hph = Hildenbrandia rubra/ Phymatolithon lenormandi. Tables A-F contain results of logistic regression models. Results of logistic regression models for all sampling strategy designs are presented for each habitat and for all sample sizes. For training data, the number (N) and frequency (F) of the habitat occurrence are presented. Results of null models are shown with the mead and standard deviation of the 10 models calculated. The D2 is the Deviance of the model in the training data; AUC is the area under the receiver operating characteristic (ROC) curve, se and spe are the sensitivity and specificity respectively, for the predictive model in the test data ​
This document is licensed under a Creative Commons:Attribution (by) Creative Commons by4.0