Table 3.
Cross‐validated linear discriminant analysis (LDA) classification statistics of 31 fossil Microtus m1s from Project 23 at Rancho La Brea
| Landmark dataset | Microtus californicus | Microtus longicaudus | Microtus montanus | Microtus oregoni | Microtus townsendii | Error (%) |
|---|---|---|---|---|---|---|
| Dinolite_NoTilt_EO_T1 | 20 | 1 | 0 | 2 | 1 | 13.8 |
| Dinolite_NoTilt_EO_T2 | 14 | 1 | 0 | 2 | 0 | 17.4 |
| Nikon_NoTilt_EO_T1 | 6 | 3 | 0 | 1 | 0 | 20.2 |
| Nikon_NoTilt_EO_T2 | 3 | 8 | 0 | 1 | 0 | 20.2 |
| Nikon_NoTilt_NO_T1 | 10 | 1 | 0 | 1 | 0 | 27.5 |
| Nikon_NoTilt_NO_T2 | 6 | 3 | 0 | 0 | 0 | 30.8 |
| Dinolite_NoTilt_NO_T1 | 12 | 1 | 0 | 0 | 0 | 34.4 |
| Dinolite_Tilted_EO_T1 | 3 | 1 | 0 | 0 | 0 | 34.4 |
| Dinolite_NoTilt_NO_T2 | 13 | 0 | 0 | 0 | 0 | 38.1 |
| Mean | 9.7 | 2.1 | 0 | 0.8 | 0.1 | 26.3 |
| Range | 3–20 | 0–8 | 0 | 0–2 | 0–1 | 13.8–38.1 |
Column values indicate the number of fossil specimens assigned to each extant species per landmark dataset. Specimens with predicted group membership probabilities <.95 are not included. “Error (%)” indicates the percentage of recent Microtus specimens (n = 247) misclassified within the LDA training set. Italicised values mark a species' presence as “likely” according to the accuracy of its respective LDA training set. See main text and Table 1 for explanations of species likelihood calculations and dataset naming, respectively.