Skip to main content
. 2023 Mar 25;130(6):381–393. doi: 10.1038/s41437-023-00613-w

Table 3.

Distance-based redundancy analysis (dbRDA) results for all 57 localities.

Marginal tests Conditional tests
Variable F p % VAR F %VAR
FSTLIN
GGD 2.827 0.001 6.66 |GGD
ELE 2.811 0.001 3.36 2.787 3.23
ELEDIF 3.522 0.001 8.81 3.183 7.52
RIVN 2.809 0.001 19.93 2.369 15.49
RIVBIN 2.674 0.001 20.94 2.388 17.57
RDSN 3.843 0.001 9.81 3.269 7.79
RWSN 3.262 0.001 11.56 2.619 8.38
RWSBIN 2.831 0.001 15.14 2.486 12.36
Model AIC ΔAIC %VAR
FSTLIN ~ ELE + ELEDIF + RDSN + RIVBIN + GGD 3.06 0 22.78
FSTLIN ~ ELEDIF + RDSN + RIVBIN + GGD 3.18 0.12 21.34
FSTLIN ~ ELEDIF + RDSN + RIVBIN 3.19 0.13 20.01
FSTLIN ~ 1 (null model) 11.29 8.23

Response variable is linearized FST and predictors are geographic distance between sampling sites (GGD), elevation (ELE), elevation difference (ELEDIF), number of rivers (RIVN), binary rivers matrix (RIVBIN), number of roads (RDSN), number of railways (RWSN) and binary railways matrix (RWSBIN). Size effect (F), p-value (p), and percentage of explained variance (%VAR) by each predictor (Adjusted R2*100). Model selection based on Akaike’s Information Criteria (AIC) is shown. below.