Table 2.
Comparisons in number of outlier SNPs detected between seven outlier detection methods
| Method (Abbreviation) | Model type | Correction for spatial or population structure |
C. punctatus # outliers (# unique) |
P. angulosus # outliers (# unique) |
S. granularis # outliers (# unique) |
|---|---|---|---|---|---|
| BayPass core model (BPC) | Bayesian | Yes, population | 9 (2) | 5 (0) | 19 (5) |
| BayPass auxiliary model (BPA) | Bayesian | Yes, population | 0 | 0 | 4 (0) |
| Latent factor mixed model (LFMM) | Mixed model | Yes, population | 134 (121) | 72 (60) | 125 (101) |
| Moran spectral outlier detection (MSOD) | Multivariate model | Yes, spatial | 15 (14) | 9 (7) | 20 (18) |
| Moran spectral randomization outlier detection (MSR) | Multivariate model | Yes, spatial | 3 (NA) | 3 (NA) | 8 (NA) |
| Redundancy analysis (RDA) | Multivariate model | No | 9 (3) | 9 (1) | 16 (2) |
| Distance-based redundancy analysis (dbRDA) | Multivariate model | Yes, spatial | 0 | 0 | 1 (1) |
Descriptions of outlier detection methods, and the number of total and unique outliers (restricted to that method) detected by each method for each species. Note that MSR could not have unique outliers as it uses those identified by MSOD