Fig 1.
Southern California study area and conservation network sub-area with puma GPS telemetry and genetic sample locations used in the analysis.
Table 1.
Predictor variables used in the puma resource selection, movement selection, and landscape genetic analyses.
Fig 2.
Functions used to transform the environmental variables to resistance, with a range of 1–100, for use in the landscape genetic analysis.
Table 2.
Characteristic scales of selection for each predictor variable from the Level II and Level III selection functions, the Path selection functions, and landscape genetic analysis.
Plus or minus indicates preference or avoidance of that variable for resource use or movement. The selected resistance transformation for the landscape genetic analysis are indicated by IR = inverse Ricker, NL = negative linear, NMCc = negative monomolecular concave, NMCv = negative monomolecular convex, PL = positive linear, PMCc = positive monomolecular concave, PMCv = positive monomolecular convex. Blank cells indicate model convergence failures.
Table 3.
Standardized beta estimates, robust standard errors, and 95% robust confidence intervals for the multivariate Level II Home Range Selection Function model variables.
Table 4.
Standardized beta estimates, robust standard errors, and 95% robust confidence intervals for the multivariate Level III Point Selection Function model variables.
Table 5.
Standardized beta estimates, robust standard errors, and 95% robust confidence intervals weights for the multivariate Path Selection Function model variables.
Fig 3.
Predicted relative probability of use surfaces from the multi-scale Level II Home Range Selection Function, the multi-scale Level III Point Selection Function and the combined Multi-Level Resource Selection Function.
Fig 4.
Resistance surfaces derived from the landscape genetics analysis, the PathSF, and the combined Multi-Level Resistance Surface.
Fig 5.
Puma resource use patches, landscape corridors, and the current and proposed protected area network.