Fig 1.
Location and landscape of three habitats of gray almond in east (A), south (B) and west (C) of Iran on topographic map.
Table 1.
Layers of environmental variables used in this study.
Fig 2.
Visualizations of environmental conditions of P. eburnea occurrence locations in Iran by PCA analysis, which summarize variation among the environmental variables.
Fig 3.
The independent (A), (given as the percentage of the total explained variance) and joint contributions (B), its independent contribution (dark color) and its conjoint contribution with all other variables (light color) of the variable variables for P. eburnea, as estimated from hierarchical partitioning.
Table 2.
Results of the randomization tests for the independent contributions of separate variable variables in hierarchical partitioning for explaining variation of P. eburnea (Z.scores are computed as (observed—mean (randomizations))/sd(randomizations), and statistical significance (*) is based on upper 0.95 confidence limit (Z > = 1.65).
Fig 4.
Results of variation partitioning of P. eburnea in terms of the fractions of variation is explained.
In a venn diagram, each circle shows how much of the variations of response variables is explained by each group of variables and overlap areas show the joint contribution of different variables. It must be considered the size of circles and overlaps in figure did not scale to their numerical values (variation of the environmental data is explained by three groups of explanatory variables: X1 = group of edaphic variables, X2 = group of topographic variables and X3 = group of climate variables. Residuals are undetermined variation and a, b and c represent unique effects of edaphic, topography and climate variables, respectively; while d, e, f and g are fractions indicating their joint effects).
Table 3.
Comparison of area under the curve (AUC), true skill statistic (TSS), sensitivity and specificity statistics using 10 fold cross-validation for each GLM and GBM approaches using different fractions of environmental variables in modeling potential habitat suitability of gray almond in Iran (C.P: Climatic variables; C.E.P: Climatic and edaphic variables; C.T.P: Climatic and topographic variables; C.E.T.P: Climatic, edaphic and topographic variables).
Table 4.
Current and predicted future potential distribution and changes in habitat suitability under climate change condition for gray almond using different fractions of environmental variables by GLM and GBM; units is km2 (C.P: Climatic variables; C.E.P: Climatic and edaphic variables; C.T.P: Climatic and topographic variables; C.E.T.P: Climatic, edaphic and topographic variables).
Fig 5.
Projections using a purely dynamic model (based on climate only) versus static + climatic model by GLM and GBM.
(A: Current distribution using climatic variables by GLM; B: Future distribution using climatic variables by GLM; C: Current distribution using climatic and static variables by GLM; D: Future distribution using climatic and static variables by GLM; E: Current distribution using climatic variables by GBM; F: Future distribution using climatic variables by GBM; G: Current distribution using climatic and static variables by GBM; H: Future distribution using climatic and static variables by GBM).