Table 3.
Confusion matrix from the random forest analysis using δ13C and δ18O isotope measurements a) from the adult baseline and b) from the yearling baseline samples29.
| (a) Isotopes—adult baseline | ||||
|---|---|---|---|---|
| True origin | Estimated origin | |||
| GoM | Med | UA | %correct | |
| GoM | 85 (77) | 9 (7) | 0 (10) | 90.4 (81.9) |
| Med | 15 (3) | 275 (247) | 0 (40) | 94.8 (85.1) |
| Total | 100 (80) | 284 (254) | 0 (50) | 93.8 (84.3) |
| (b) Isotopes—yearling baseline | ||||
|---|---|---|---|---|
| True origin | Estimated origin | |||
| West | East | UA | %correct | |
| West | 94 (74) | 21 (5) | (36) | 81.7 (64.3) |
| East | 25 (6) | 125 (99) | (45) | 83.3 (66) |
| Total | 119 (80) | 146 (104) | (81) | 82.6 (65.3) |
Figures relate to the numbers of individuals from the baseline samples assigned to the Gulf of Mexico (GoM) and Mediterranean (Med) spawning populations or unassigned (UA) using bootstrap sub-sampling and probability thresholds of 50% (outside parenthesis) and 80% (inside parenthesis).