Table 2. Model selection results.
| Introgression metric | Model | AIC |
|---|---|---|
| PRTA > 1% | T + RTrange + DS + E + MAF + DT13 + W95 + S | 575.87 |
| T + RTrange + DS + E + MAF + DT13 + W95 + YCTI + S | 576.23 | |
| T + RTrange + DS + MAF + E + DT13 + YCTI + S | 576.72 | |
| T + RTrange + DS + E + MAF + W95 + S | 576.88 | |
| PRTA > 10% | T + RTrange + DS + DT13 + MAF + E | 433.13 |
| T + RTrange + DS + DT13 + MAF + E + YCTI | 433.51 | |
| T + RTrange + DS + DT13 + MAF + E + W95 | 434.27 | |
| T + RTrange + DS + DT13 + MAF + E + S | 434.32 | |
| PRTA > 20% | T + RTrange + DT13 + MAF + E + DS | 387.75 |
| T + RTrange + DT13 + MAF + E | 388.40 | |
| T + RTrange + DT13 + MAF + E + DS + S | 389.23 | |
| T + RTrange + DT13 + MAF + E + DS + W95 | 389.35 | |
| PFRT > 1% | T + DS + DF3 + MAF + E + YCTI + RTrange + DT13 | 552.82 |
| T + DS + DF3 + MAF + E + YCTI + DT13 | 553.25 | |
| T + DS + DF3 + MAF + E + YCTI + RTrange + DT13 + S | 554.45 | |
| T + DS + DF3 + MAF + E + YCTI + RTrange | 554.72 | |
| PFRT > 10% | T + RTrange + DS + MAF + E + YCTI + S | 551.44 |
| T + RTrange + DS + MAF + E + YCTI | 552.05 | |
| T + RTrange + DS + MAF + E + YCTI + S + DF3 | 553.14 | |
| T + RTrange + DS + MAF + E + YCTI + S + W95 | 553.33 | |
| PFRT > 20% | T + RTrange + DS + MAF + E + YCTI + S | 497.57 |
| T + RTrange + DS + MAF + E + YCTI | 498.20 | |
| T + RTrange + DS + MAF + E + YCTI + S + DF3 | 499.15 | |
| T + RTrange + DS + MAF + E + YCTI + S + DT13 | 499.19 |
Model selection results for logistic regression equations relating environmental covariates to whether sites exceeded 1%, 10%, or 20% rainbow trout alleles (PRTA) or fish with rainbow trout alleles (PFRT). The four top models are ranked from most to least plausible for each metric and threshold. Underlined variables had coefficients that were not significantly different from zero. The model shown in bold font (the consensus model) was used to predict the probability that hybridization would exceed specified thresholds of PRTA in streams across the study area.