Table 1. Optimized univariate resistance models for each region.
Region | Landscape variablea | Optimal resistance surface | Median R2 |
---|---|---|---|
DEVA | |||
Slope | Gaussian (rmax = 100, xopt = 50, xsd = 20) | 0.151 | |
Distance to water | Monotonic positive (rmax = 50, rexp = 4) | 0.132 | |
NDVI | Monotonic negative (rmax = 10, rexp = 0.25) | 0.130 | |
Euclidean distance | — | 0.129 | |
GRCA | |||
Water barriers | Ratio (1,000) | 0.173 | |
Slope | Gaussian (rmax = 10, xopt = 50, xsd = 20) | 0.062 | |
Distance to water | Monotonic positive (rmax = 50, rexp = 1) | 0.063 | |
Euclidean distanceb | — | 0.050 | |
MOJA | |||
Major roads | Ratio (100) | 0.227 | |
Slope | Gaussian (rmax = 10, xopt = 40, xsd = 20) | 0.200 | |
NDVI | Monotonic negative (rmax = 100, rexp = 0.25) | 0.186 | |
Euclidean distance | — | 0.173 |
a Variables not included in the table did not outperform the null model of isolation by distance (i.e., had lower median Akaike weight) and were excluded from further analysis.
b Cost distance was log-transformed only for the GRCA Euclidean distance model because untransformed cost distances produced higher model R2 for all other resistance models.